Fast and accurate 3D shape generation from point clouds is essential for real-world applications such as robotics, AR/VR, and digital content creation. We present ConTiCoM-3D, a continuous-time consistency model that generates 3D shapes directly in point space, without relying on discretized diffusion steps, pre-trained teacher models, or latent-space encodings. Our approach combines a TrigFlow-inspired continuous noise schedule with a Chamfer Distance-based geometric loss, providing stable training in high-dimensional point sets while avoiding costly Jacobian-vector products. This enables efficient one- to two-step inference with high geometric fidelity. Unlike previous methods that require iterative denoising or latent decoders, ConTiCoM-3D operates entirely in continuous time with a time-conditioned neural network, achieving fast generation. Extensive experiments on the ShapeNet benchmark demonstrate that our method matches or surpasses leading diffusion and latent consistency models in both quality and efficiency, establishing ConTiCoM-3D as a practical solution for scalable 3D shape generation.

Correct and useful system descriptions are central to model-based fault diagnosis, as they describe the structure and behaviour of the system. But so far, system descriptions were always interpreted as complete propositional logic models and were assumed to be given. However, with increasing use of data-driven methods that approximate system descriptions, it cannot be guaranteed that the system description is a complete model of the system. It cannot even be guaranteed that the system description contains all observations, components, and connections that the real system exhibits. This requires novel approaches to determine how well a system description models the real system and how well it can thus be used for fault diagnosis. We present a novel algorithm which takes a syntactic approach to calculate diagnosability of approximated system descriptions. This is different from previous diagnosability research, which determined the diagnosability of real systems, where the algorithm could rely on the model’s completeness and corresponding reliable observations. With approximated models, observations and models are (partially) disconnected. We also present two novel algorithms to calculate new metrics that determine how close two approximated system descriptions are to each other by heuristically solving the graph alignment problem. We believe that our new approach and corresponding algorithms benefit practitioners who want to evaluate the quality of their approximated models for fault diagnosis. We show the usefulness of our results on established benchmarks of tank-systems, the Tennessee Eastman Process, two spacecraft, and a benchmark of different Boolean circuits.

Achieving robust uncertainty quantification for deep neural networks represents an important requirement in many real-world applications of deep learning such as medical imaging where it is necessary to assess the reliability of a neural network’s prediction. Bayesian neural networks are a promising approach for modeling uncertainties in deep neural networks. Unfortunately, generating samples from the posterior distribution of neural networks is a major challenge. One significant advance in that direction would be the incorporation of adaptive step sizes, similar to modern neural network optimizers, into Monte Carlo Markov chain sampling algorithms without significantly increasing computational demand. Over the past years, several papers have introduced sampling algorithms with corresponding theorems stating that they achieve this property. In this paper, we demonstrate that these methods can have a substantial bias in the distribution they sample, even in the limit of vanishing step sizes and at full batch size. Furthermore, for most of the algorithms, we show that convergence to the correct distribution can be restored with a simple fix at the cost of increasing computational demand.

Anomaly detection and diagnosis algorithms for Cyber-Physical Systems, especially Cyber-Physical Production Systems, are increasingly data-driven and rely on sufficient and representative data to be fitted. However, recording such data comes with cost, especially for rare or unsafe operational states. While simulation offers a scalable solution, by generating synthetic data, it often comes with a gap between simulated and real-world environments. In this paper, we investigate the impact of a pretraining with simulation-generated data on anomaly detection algorithms in a case study of a model Cyber-Physical Production Systems and its simulation. In a first step, we therefore train different anomaly detection algorithms on simulated data, and subsequently continue training with real data. We examine, (i) whether pretraining with additional synthetic data enhances the performance of anomaly detection algorithms, and (ii) how the proportion of real versus synthetic data affects a model’s effectiveness when operating on a fixed data budget. Our findings show that pretraining on simulation generated data can increase the performance of anomaly detection algorithms, however, solely training on simulated data is leading to a decrease in performance.

Reliable sensor data are essential for the effective operation and safety of cyber-physical systems (CPS) in industrial environments. However, sensors frequently experience faults or degradation, leading to compromised system performance. In order to increase the resilience of CPS, this paper proposes a novel approach to creating virtual sensors capable of reconstructing missing or faulty sensor data through gradient-based input reconstruction by leveraging neural networks. Specifically, we employ an LSTM-based autoencoder architecture trained both conventionally and with a masking strategy to handle potential sensor data loss scenarios effectively. Our method involves using automatic differentiation and gradient descent to iteratively optimize missing sensor inputs, guided by the pretrained network. We evaluate this approach comprehensively on both simulated and real-world plant data from cyber-phyiscal process plants, demonstrating robust reconstruction performance across various sensor failure scenarios. Additionally, we explore the efficacy of modular clustering methods versus single comprehensive models, highlighting the advantages and limitations inherent to each approach. Our findings reveal significant potential for improving system resilience and maintaining operational continuity in CPS through advanced virtual sensor implementations.

The rapid evolution and continuously growing complexity of cyber-physical systems (CPS), ranging from Industry 4.0 production plants to ship drivetrains and building monitoring, have led to the abundant generation of heterogeneous, poorly-structured, and not standardized data. This situation is further aggravated by retrofitting legacy systems with new sensors for the purpose of data-driven monitoring. In this paper, we introduce Cyber-Physical System Watch (CPSWatch), a lightweight framework that aims to support the monitoring of CPS including the possibility for diagnosis, encompassing a high-level ontology, two sample datasets of different complexity as well as a use case scenario on how it can be applied. Our proposed ontology provides a unified framework for describing data across different CPS applications and aligns with OPC UA, ensuring its general applicability in modern industrial settings. CPSWatch is evaluated in terms of standard ontology evaluation measures, within the scope of condition monitoring of an automation system in the maritime domain and a benchmark in the field of process engineering.

A long-term objective for many research fields, such as anomaly detection, discretization and root-cause diagnosis in Cyber-Phyiscal Production Systems is the realization of resilient and highly autonomous systems. Instances of those systems range from the control and operation of single production systems to controlling entire plants. While notable progress has been made, a key challenge remains unaddressed: the availability of comprehensive and standardized datasets necessary for advancing machine learning based solutions. Typical evaluation datasets comprise systems of (too) little complexity or are either suitable for only data-driven or only for symbolic methods. Yet, a comprehensive dataset and model for developing, training, and testing machine learning methods from anomaly detection to fault diagnosis in a structured and comparable manner does not exist. To bridge this gap, we present a benchmark specifically designed to support both data-driven methods and symbolic reasoning approaches, by extending the BerFiPl benchmark introduced by Ehrhardt et al. [1]. By providing hybrid data streams, labeled system states, and a modular, interpretable system structure, the dataset offers a unique opportunity to develop, train, test, and compare hybrid AI methods to bridge the gap between data-driven and symbolic paradigms.

This paper explores the improvement of forecasting voltages throughout electric grids, particularly under load stress by random load switches. The shift towards renewable energy sources and consumers makes it clear that advanced, property-aware machine learning methods are required to understand the system behavior of electric grids and make them more robust to maintain power quality during sudden voltage spikes. Since electric grids are spatiotemporal cyber-physical systems, we have access to several dimensions: temporal, structural, and physical. We show that a combination of different priors (stochastic, physical, and structural) is highly effective in training a probabilistic model for robust forecasting of voltage spikes in the electric grid. Additionally, with probabilistic models, we can quantify the uncertainty of the predictions as well.

Neural networks are a common approach to learn data-driven models of cyber-physical systems. The learned models can be used to monitor the systems, detect anomalies and enable diagnosis. Complex technical systems usually comprise a modular design. Modular neural architectures can be adopted to mirror this modular design, so that each module within the cyber-physical system is modeled by an individual neural network. This can improve anomaly detection performance compared to monolithic modeling, while providing more detailed insights into the system through a granular structure and decoupling between modules. One of the principal challenges in developing modular architectures is establishing an appropriate degree of granularity and determining whether a unified modeling approach should be applied across all instances of a given module type, or if individualized models are necessary for each instance.In this research we compare the performance of type-level and instance-level learning approaches for the task of anomaly detection. We use simulations to generate normal and anomalous data for multiple distinct instances of the same module types. Subsequently, we learn autoencoder neural networks for residual based anomaly detection and compare the performances of the different modeling approaches.Our experiments show that the learning approach has an impact on the anomaly detection performance and the robustness, with instance modular learning achieving better scores and type modular learning being more robust. All data and code files are made available on GitHub1.

Cyber-Physical Systems (CPS) in domains such as manufacturing and energy distribution generate complex time series data crucial for Prognostics and Health Management (PHM). While Deep Learning (DL) methods have demonstrated strong forecasting capabilities, their adoption in industrial CPS remains limited due to insufficient robustness. Existing robustness evaluations primarily focus on formal verification or adversarial perturbations, inadequately representing the complexities encountered in real-world CPS scenarios. To address this, we introduce a practical robustness definition grounded in distributional robustness, explicitly tailored to industrial CPS, and propose a systematic framework for robustness evaluation. Our framework simulates realistic disturbances, such as sensor drift, noise and irregular sampling, enabling thorough robustness analyses of forecasting models on real-world CPS datasets. The robustness definition provides a standardized score to quantify and compare model robustness across diverse datasets, assisting in informed model selection and architecture design. Through extensive empirical studies evaluating prominent DL architectures (including recurrent, convolutional, attention-based, modular, and structured state-space models) we demonstrate the applicability and effectiveness of our approach. We publicly release our robustness benchmark to encourage further research and reproducibility.

Abstract Forecasting the behaviour of industrial robots, power grids or pandemics under changing external inputs requires accurate dynamical models that can adapt to varying signals and capture long-term effects such as delays or memory. While recent neural approaches address some of these challenges individually, their reliance on computationally intensive solvers and their black-box nature limit their practical utility. In this work, we propose Laplace-Net, a decoupled, solver-free neural framework for learning forced and delay-aware dynamical systems. It uses the Laplace transform to (i) bypass computationally intensive solvers, (ii) enable the learning of delays and memory effects and (iii) decompose each system into interpretable control-theoretic components. Laplace-Net also enhances transferability, as its modular structure allows for targeted re-training of individual components to new system setups or environments. Experimental results on eight benchmark datasets–including linear, nonlinear and delayed systems–demonstrate the method’s improved accuracy and robustness compared to state-of-the-art approaches, particularly in handling complex and previously unseen inputs.

Empirical research plays a fundamental role in the machine learning domain. At the heart of impactful empirical research lies the development of clear research hypotheses, which then shape the design of experiments. The execution of experiments must be carried out with precision to ensure reliable results, followed by statistical analysis to interpret these outcomes. This process is key to either supporting or refuting initial hypotheses. Despite its importance, there is a high variability in research practices across the machine learning community and no uniform understanding of quality criteria for empirical research. To address this gap, we propose a model for the empirical research process, accompanied by guidelines to uphold the validity of empirical research. By embracing these recommendations, greater consistency, enhanced reliability and increased impact can be achieved.

In this paper, we tackle the problem of planning in complex numerical domains, where actions are indexed by control parameters, and their effects may be described by neural networks. We propose a lazy, hierarchical approach based on two ingredients. First, a Satisfiability Modulo Theory solver looks for an abstract plan where the neural networks in the model are abstracted into uninterpreted functions. Then, we attempt to concretize the abstract plan by querying the neural network to determine the control parameters. If the concretization fails and no valid control parameters could be found, suitable information to refine the abstraction is lifted to the Satisfiability Modulo Theory model. We contrast our work against the state of the art in NN-enriched numerical planning, where the neural network is eagerly and exactly represented as terms in Satisfiability Modulo Theories over nonlinear real arithmetic. Our systematic evaluation on four different planning domains shows that avoiding symbolic reasoning about the neural network not only leads to substantial efficiency improvements, but also enables their integration as black-box models.

Condition assessments are crucial in the field of civil engineering to avoid potential safety hazards and expensive repairs. For this, non-destructive testing methods like the Impact Echo method are used to detect defects like honeycombs inside of concrete without causing additional damage. This study introduces a four-step method and developed the U-Net Autoencoder (U-AE) to identify such defects from Impact Echo measurements. Our process involves preparing and learning from both simulated and real-world data to ensure an accurate anomaly detection. The key findings of this work demonstrate crucial aspects of data preparation and a significant performance enhancement achieved through pretraining with simulated data followed by fine-tuning with real-world datasets. Our approach effectively counteracts the information loss typically caused by dimension reduction, discovering an optimal balance in the latent space. Overall, this research introduces a novel architecture for detecting defects in concrete structures, marking a new advancement the practice of maintaining and inspecting civil infrastructure.

Autonomously detecting anomalous behavior based on system observations is a fundamental task for Cyber-Physical Systems (CPS). Due to the high system complexity and large number of subsystems in modern CPS, rule- or knowledge-based approaches for anomaly detection are more and more replaced by Machine Learning (ML) approaches which leverage historical CPS data. Typically, ML approaches learn a system model based on the CPS data and identify anomalous behavior based on the distance of the real CPS behavior to the predicted model behavior. However, most classical ML approaches for anomaly detection are monolithic, meaning a single ML model is fitted on a global CPS observation, making them frail to spurious correlations and confounders that originate on CPS subsystem level. We hence propose a modular approach toward anomaly detection in CPS, specifically a novel Modular Neural Network (MNN) architecture. Our architecture not only models the behavior of individual CPS sub-systems in individual MNN modules, but additionally models the dependencies of the CPS subsystems into the MNN architecture. Thereby, we omit confounding effects and spurious correlations, enabling us to identify and allocate anomalies within the CPS at subsystem level. We benchmark our MNN architecture against monolithic Neural Networks and MNN architectures that do not explicitly model CPS subsystem dependencies using a real-world dataset of an industrial robot with different anomalies. We show that by modeling real-world dependencies into a MNN architecture, we can improve the performance of autonomous anomaly detection in CPS.

Traditionally, engineering designs are created manually by experts. This process can be time-consuming and requires significant computing resources. Designs are iteratively created and simulated to satisfy physical constraints. Generative neural networks have the ability to learn the relationship between physical constraints and their geometric design. By leveraging this understanding, generative neural networks can generate innovative designs that satisfy these constraints. In the context of Industry 4.0, integrating these networks can significantly speed up the design process, reduce costs, and push the boundaries of traditional engineering practices. To achieve this goal, we propose a conditional variational autoencoder to learn the underlying relationship between geometry and physics. We validate this approach on the ShapeNetCore dataset, focusing on subsets that contain three-dimensional objects such as cars and airplanes, which contain both continuous and discrete data.

With the increasing complexity of highly automated cyber-physical systems (CPS), monitoring their behavior has become crucial. Failures in these systems can be costly, halt production, or even pose risks to human safety. Effective diagnosis depends on understanding the system's components, connections, and the influences among them, knowledge typically provided by experts. However, the shift towards self-diagnosing systems necessitates this knowledge be machine-readable and interpretable. This paper introduces a novel methodology that utilizes an ontology to encode knowledge about cyber-physical systems and systematically generate propositional logical expressions. These expressions can then be evaluated using state-of-the-art diagnostic algorithms to identify failure causes. Our methodology was validated using an established AI benchmark for diagnostics. We constructed an ontology description for the underlying cyber-physical system, deduced influences of system sensors from data, and successfully diagnosed induced failures, demonstrating the efficacy and applicability of our approach.

With the increasing complexity in manual assembly and a demographic decline in skilled workforce, the importance of well-documented processes through assembly instructions has grown. Creating these instructions is a time-consuming and knowledge-intensive task that typically relies on experienced employees. Although various automation solutions have been proposed to assist in generating assembly instructions, they often fall short in providing detailed textual guidance. With the rise of generative artificial intelligence (AI), new potentials arise in this domain. Therefore, this paper explores these potentials by employing various large language models (LLMs), prompting techniques and input data in an experimental setup for generating detailed assembly instructions, including the planning of assembly sequences as well as textual guidance on tools, assembly activities, and quality assurance measures. The findings reveal promising opportunities in leveraging LLMs but also substantial challenges, particularly in assembly sequence planning. To improve the reliability of generating assembly instructions, we propose a multi-agent concept that decomposes the complex task into simpler subtasks, each managed by specialized agents.

Root-cause diagnosis techniques, such as consistency-based and abductive diagnosis, offer essential support in explaining symptoms in a cyber-physical system. Developing and maintaining the required (detailed or abstract) models can be a serious challenge. Related issues include the complexity of the required knowledge and the dynamic changes we see in a system over its life cycle. This raises the question regarding strategies and the feasibility of utilizing unsupervised machine learning to learn diagnostic system models based on available time series data in order to address this challenge. This paper presents the novel methodology Discret2DeepDive for automated learning of diagnostic system models for root-cause diagnosis, focussing on the use of automata for state mapping over time and explores advancements related to the handling of dynamic time series data. These advancements are incorporated into both the discretization process and the generation of residuals. The findings demonstrate a notable enhancement in discretizing time series data into modes and residual generation for anomaly detection in sequential data, thereby providing a substantial value for diagnosing faults.

This paper investigates the application of machine learning techniques for discretizing multivariate time series data in cyber-physical systems, emphasizing an unsupervised learning approach. Due to the lack of system information about the systems under investigation, we focus our work on the conversion of high-dimensional, continuous data into discrete state representations to improve the interpretability of the system. The study evaluates several unsupervised machine learning methods using both simulated and real datasets of cyber-physical systems, particularly examining their utility in anomaly detection tasks. Our analysis highlights the trade-offs between the complexity and the purity of the learned state representations, and how this influences the performance in subsequent anomaly detection applications. The paper provides a comprehensive view of how different machine-learning-based discretization methods perform under various conditions and offers practical guidelines for selecting the appropriate method. Through this work, we aim to contribute to the broader understanding and implementation of effective unsupervised machine learning strategies for data analysis in cyber-physical systems.

This study investigates the integration of formal and informal prior knowledge into Neural Ordinary Differential Equations (NODEs) for enhancing the modeling of nonlinear electrical circuits, using a state-of-the-art LSTM as a baseline for comparison. By applying a systematic methodology to incorporate diverse knowledge types, our experiments with 2nd and 4th-order RLC circuits demonstrate improved model accuracy and efficiency. Results highlight that knowledge-infused NODEs outperform traditional LSTM models in handling complexity. Furthermore, the ability of NODEs to identify physical parameters is demonstrated.

Analyzing Prompt Gamma Activation Analysis (PG AA) spectra poses significant challenges, particularly in accurately identifying and quantifying the elements present. Traditional expert analysis, while effective, is time-consuming. This paper addresses the need for an efficient, automated solution to enhance the analysis process. The research investigates the use of machine learning (ML) and deep learning (DL) algorithms in the automated analysis of PGAA spectra. We aim to establish a new metric for comparing automated analysis with expert analysis, providing a baseline using Linear Regression, Random Forest, 1D Convolutional Neural Network, Feed Forward Neural Network, and Autoencoder algorithms. Established metrics like Mean Square Error (MSE) and Mean Absolute Error (MAE) are utilized to compare the performance of these automated approaches against traditional expert analysis. Using actual spectra from various research projects and semi-real augmented data, the study demonstrates that the Feed Forward Neural Network (FFNN) and Autoencoder algorithms can effectively predict the magnitude of the present elements. These findings suggest especially D L algorithms could signifi-cantly assist researchers and industry personnel by providing a rough overview of the material and saving valuable time. However, the automated approach requires further refinement, particularly in handling noisy data, predicting additional crucial information, and integrating more prior knowledge into the anal-ysis. This research offers valuable insights into the application of ML algorithms in spectral analysis and lays a foundation for further advancements in the field.

The efficiency of manual assembly can be significantly improved by utilizing assistance systems that display as-sembly instructions. However, generating and maintaining these instructions require substantial effort, especially for low-volume or highly customized products. If neglected, this can lead to outdated instructions, frustrated operators, and the abandonment of the assistance platform. Recent advancements in Large Language Models (LLMs) have made it possible to generate high-quality assembly instructions given the right input. However, relying solely on LLMs to plan the assembly process risks producing unfeasible assembly sequences due to potential hallucinations by the models. To address this, Reinforcement Learning (RL) can be used in simulations to plan the assembly process, imposing restrictions on impractical movements. We propose a framework that integrates RL and LLMs to generate practical and accurate assembly instructions. In our framework, RL is used within a simulation to generate feasible assembly sequences. These sequences are then transformed into detailed assembly instructions by an LLM. We evaluate our framework on real-world product assemblies, generating comprehensive assembly instructions from corresponding CAD files. Our results demonstrate the potential of combining RL and LLMs to automate the generation of assembly instructions, thereby overcoming the limitations of current assistance systems and enhancing the efficiency of manual assembly processes.

In Industry 4.0, Cyber-Physical Systems (CPS) generate vast data sets that can be leveraged by Artificial Intelligence (AI) for applications including predictive maintenance and pro-duction planning. However, despite the demonstrated potential of AI, its widespread adoption in sectors like manufacturing remains limited. Our comprehensive review of recent literature, including standards and reports, pinpoints key challenges: system integration, data-related issues, managing workforce-related concerns and ensuring trustworthy AI. A quantitative analysis highlights particular challenges and topics that are important for practitioners but still need to be sufficiently investigated by academics. The paper briefly discusses existing solutions to these challenges and proposes avenues for future research. We hope that this survey serves as a resource for practitioners evaluating the cost-benefit implications of AI in CPS and for researchers aiming to address these urgent challenges.

Cyber-Physical Production Systems (CPPS) are becoming increasingly important in modern manufacturing, which leads to a growing need for automated anomaly detection, maintenance decision-making, and fault diagnosis. Artificial Intelligence (AI) and Machine Learning (ML) algorithms are often used to perform these tasks. However, there is a shortage of real data sets with fault modes, making it difficult to train ML algorithms. To overcome this problem, we propose the use of the Flexible Production Simulation (FliPSi) to generate simulated data for the development, training and evaluation of ML algorithms. FliPSi is a simulation tool developed in Unity, a game engine, which is used to simulate modular CPPS. It allows the construction of different CPPS, data collection, and generating data which could realistically originate from CPPS. The focus of FliPSi is to generate data for the development, training and testing of anomaly detection and diagnosis algorithms. In this paper, we show that data generated with FliPSi can be used to train ML algorithms, specifically recurrent neural networks (RNNs) and variations such as Long Short-Term Memory (LSTM) and Gated Recurrent Units (GRU), which are specifically designed to process sequential data and that the algorithms trained with these data sets can be employed to detect simulated anomalies. With this, we show that FliPSi can be employed to overcome a blocking issue in using ML algorithms for anomaly detection.

In computer science and related technical fields, researchers, educators, and practitioners are continuously automating recurring tasks for high efficiency in a wide variety of fields. In higher education, such tasks that educators face are the recurring review and assessment process of students' programming coursework. Thus, various attempts exist to automate the assessment and feedback generation for course homework and practicals in higher education. Those approaches for automated programming task assessment often comprise running automated tests to check for limited functional correctness and potentially style checking for various violations (LINTing). Educators familiar with large-scale automated task assessment are likely used to seeing hard-coded solutions specifically or accidentally designed to just pass the required tests, ignoring or misinterpreting the actual task requirements. Detecting such issues in arbitrary code is non-trivial and an ongoing research topic in software engineering. Software engineering research has yielded various semantic analysis frameworks, such as GitHub's CodeQL, which can be adapted for programming task assessment. We present a work-in-progress programming task analysis framework which employs CodeQL's analysis technology to identify the actual use of task-description-mandated syntactic and semantic elements such as loop structures or the use of mandated data blocks in branching conditions. This allows extending existing course work analysis frameworks to include a semantic check of an uploaded program which exceeds the relatively simple set of input-output test cases provided by unit tests. We use a running example of entry level programming tasks and several solution attempts to introduce and explain our proposed control flow and data flow -based analysis method. We discuss the benefits of including semantic analysis as an additional method in the automated programming task assessment toolbox. Our main contribution is the adaptation of an semantic analysis code framework to analyse syntactic and semantic components in students' programming coursework.

Especially in highly interdisciplinary fields such as automation engineering, contemporary programming education with tailored assignments and individual feedback is a major challenge for educational institutions due to the increasing number of students per teacher and the ever-increasing demand for computer science professionals. To address this gap, we present ”KIAAA” an AI Assistant for Automation Engineering Teaching, a work-in-progress approach for an integrated, customized, and AI-based learning support system for automation and programming courses based on instructor-defined course objectives. Thereby in the KIAAA system, the individual knowledge level of the students is determined and individually tailored virtual learning scenarios are generated based on the knowledge and learning profile of the students. These are iteratively adapted based on the answers given. To achieve this, KIAAA uses several AI components, a hybrid rule-based scenario generation component, a Help-DKT-based cognitive model, and a solution assessor that uses a combination of traditional code analysis methods and AI-based analyses methods for automated programming task assessment. These components are the main parts of KIAAA to generate customized programming scenarios as well as visualization and simulation based on a modern game and physics engine.

The increasing size and complexity of cyber-physical production systems (CPPS) lead to an increasing number of faults, such as broken components or interrupted connections. Nowadays, faults are handled manually, which is time-consuming because for most operators mapping from symptoms (i.e., warnings) to repair instructions is rather difficult. To enable CPPS to adapt to faults autonomously, reconfiguration, i.e., the identification of a new configuration that allows either reestablishing production or a safe shutdown, is necessary. This article addresses the reconfiguration problem of CPPS and presents a novel algorithm calledAutoConf.AutoConfoperates on a hybrid automaton that models the CPPS and a specification of the controller to construct a QSM. This QSM is based on propositional logic and represents the CPPS in the reconfiguration context. Evaluations on an industrial use case and simulations from process engineering illustrate the effectiveness and examine the scalability ofAutoConf.

To improve the autonomy of Cyber-Physical Production Systems (CPPS), a growing number of approaches in Artificial Intelligence (AI) is developed. However, implementations of such approaches are often validated on individual use-cases, offering little to no comparability. Though CPPS automation includes a variety of problem domains, existing benchmarks usually focus on single or partial problems. Additionally, they often neglect to test for AI-specific performance indicators, like asymptotic complexity scenarios or runtimes. Within this paper we identify minimum common set requirements for AI benchmarks in the domain of CPPS and introduce a comprehensive benchmark, offering applicability on diagnosis, reconfiguration, and planning approaches from AI. The benchmark consists of a grid of datasets derived from 16 simulations of modular CPPS from process engineering, featuring multiple functionalities, complexities, and individual and superposed faults. We evaluate the benchmark on state-of-the-art AI approaches in diagnosis, reconfiguration, and planning. The benchmark is made publicly available on GitHub.

Given the growing number of volatile energy producers and consumers and the limitations of traditional static load prediction models, we have analyzed the ability of neural networks to predict the loads of an electrical transformer and to understand the physical relationships in the electric grid. To do this, we use a variational autoencoder to learn the load behavior of a neighborhood with three houses. Since a variational autoencoder learns a latent representation, we analyzed the possibility of learning physical relationships of the electric grid. By adapting the prediction model to learn a physical variable, namely phase shift, we show that a variational autoencoder can learn physical relations. Our results show a significant improvement in terms of the correlation of the latent and physical variables by integrating prior knowledge in the form of the corresponding power values as the training objective.

Optimization loops are often required for the improvement of function and electromagnetic compatibility (EMC) in a product development process. Such optimization can be realized either by simulations or high effort based measurements. The neural network approach is suitable to overcome typical issues of simulation programs like SPICE such as convergence problems and high computation time. This paper addresses a neural network modeling approach for nonlinear passive filters. Long Short-Term Memory (LSTM) networks are applied to model nonlinear passive filters. One measured and two simulated filter circuits are used as application examples. LSTMs are chosen by literature research as a suitable modeling approach. The neural network and training structure is defined by literature research and systematic experiments. The filter behaviors are basically modeled by the trained neural networks. But further improvements have to be done. It is shown that the corresponding voltage and current time series can be learned and predicted by the LSTM networks in their essential characteristics. These voltage and current time series can generally be used in further applications. A possible speed advantage of LSTM networks is also examined.

AI methods for fault handling in Cyber-Physical Production Systems (CPPS) such as production plants and tank systems are an emerging research topic. In the last years many methods for the detection of anomalies and faults, the diagnosis of the root cause and the automated repair have been developed. However, most of the methods are barely evaluated using a wide range of systems but applicability is shown using single use cases. In this paper, an ensemble of simulated benchmark systems is presented, which allows for a broad evaluation of AI methods for fault handling. The ensemble consists of seven different tank systems from process engineering with varying sizes and complexities and is made publicly available on Github. The suitability of the ensemble is shown using AI methods for fault handling such as anomaly detection, diagnosis and reconfiguration.

Data-driven anomaly detection is one of the central issues for the implementation of predictive maintenance in cyber-physical systems (CPS). The increasing nonstationary dynamics in CPS lead to complex shapes of collected process data, e.g., convex and nonconvex. Algorithms with the ability of handling nonconvex data are desired for anomaly detection tasks in CPS. Archetypal analysis selects extreme points (archetypes) to represent a dataset-mainly to improve run times, e.g., for anomaly detection. The classic archetypal analysis uses convex combinations of archetypes to represent a set of observations (data points). This leads to the performance depression of the classic archetypal analysis methods for anomaly detection tasks on nonconvex datasets. Such nonconvex sets are typical for CPS. In this article, the anomaly detection tasks are considered as one-class classification problem due to the lack of abnormal samples. A novel nonconvex archetypal one-class classification algorithm is proposed to address the challenge of nonconvex data, which combines the random projection and the AdaBoost algorithm. The major advantages of this method are its high efficiency, flexibility, and its ability to handle both convex and nonconvex datasets, i.e., it can be applied to CPS analysis tasks. The usefulness of the presented approach is evaluated for fault diagnosis tasks in CPS.

This paper introduces a novel approach to Model-Based Diagnosis (MBD) for hybrid technical systems. Unlike existing approaches which normally rely on qualitative diagnosis models expressed in logic, our approach applies a learned quantitative model that is used to derive residuals. Based on these residuals a diagnosis model is generated and used for a root cause identification. The new solution has several advantages such as the easy integration of new machine learning algorithms into MBD, a seamless integration of qualitative models, and a significant speed-up of the diagnosis runtime. The paper at hand formally defines the new approach, outlines its advantages and drawbacks, and presents an evaluation with real-world use cases.

Currently, detecting and isolating faults in hybrid systems is often done manually with the help of human operators. In this paper we present a novel model-based diagnosis approach for automatically diagnosing hybrid systems. The approach has two parts: First, modelling dynamic system behaviour is done through well-known state space models using differential equations. Second, from the state space models we calculate Boolean residuals through an observer-pattern. The novelty lies in implementing the observer pattern through the use of a symbolic system description specified in satisfiability theory modulo linear arithmetic. With this, we create a static situation for the diagnosis algorithm and decouple modelling and diagnosis. Evaluating the system description generates one Boolean residual for each component. These residuals constitute the fault symptoms. To find the minimum cardinality diagnosis from these symptoms we employ Reiter’s diagnosis lattice.For the experimental evaluation we use a simulation of the Tennessee Eastman process and a simulation of a four-tank model. We show that the presented approach is able to identify all injected faults.

Cyber-physical production systems (CPPS) integrate physical and computational resources due to increasingly available sensors and processing power. This enables the usage of data, to create additional benefit, such as condition monitoring or optimization. These capabilities can lead to cognition, such that the system is able to adapt independently to changing circumstances by learning from additional sensors information. Developing a reference architecture for the design of CPPS and standardization of machines and software interfaces is crucial to enable compatibility of data usage between different machine models and vendors. This paper analysis existing reference architecture regarding their cognitive abilities, based on requirements that are derived from three different use cases. The results from the evaluation of the reference architectures, which include two instances that stem from the field of cognitive science, reveal a gap in the applicability of the architectures regarding the generalizability and the level of abstraction. While reference architectures from the field of automation are suitable to address use case specific requirements, and do not address the general requirements, especially w.r.t. adaptability, the examples from the field of cognitive science are well usable to reach a high level of adaption and cognition. It is desirable to merge advantages of both classes of architectures to address challenges in the field of CPPS in Industrie 4.0.

The vision of smart factories are self-diagnosing, self-optimizing and self-adapting Cyber-Physical Production Systems (CPPS). Self-adaption, on which this paper focuses on, means that the CPPS can adapt itself to a changing environment, so that the downtime costs can be reduced by using the system modules most efficient. An architecture is introduced and demonstrated on a concrete use case to show how this capability can be achieved by using different Artificial Intelligence (AI) techniques. For each technique, we define challenges that have to be solved to use it in a real world environment. Additionally, we illustrate the symbolic and subsymbolic AI and argue why symbolic AI is an important aspect in the context of CPPS.

Cyber-Physical Systems (CPS) are systems that connect physical components with software components. CPS used for production are called Cyber-Physical Production Systems (CPPS). Since the complexity of these systems can be very high, finding the cause of an error takes a lot of effort. In this paper, a data-driven approach to identify causal dependencies in cyber-physical production systems (CPPS) is presented. The approach is based on two different layers of learning algorithms: one low-level layer that processes the direct machine data and a higher-level learning layer that processes the output of the low-level layer. The low-level layer is based on different learning modules that can process differently typed data (continuous, discrete or both). The high-level learning algorithms are based on rule-based and case-based reasoning. Thus, causal dependencies are detected allowing the plant operator to find the error cause quickly.

This work shows how OWL ontologies can be represented into the automation standards AutomationML and OPC UA. It is often asserted that an integration is possible, but no detailed review could be found. The integration of OWL into the standards is relevant, because it enables the collection and usage of data through the whole life cycle in OWL. We show that it is possible, but we identified some restriction regarding the representation in OPC UA.

Integrating data analysis into workflows is a recent tendency in manufacturing sectors. According to the vision of Industry 4.0, data analysis can be automatically performed at any point of workflows if needed. In distributed and complex manufacturing systems, checking the integrity of data analysis processes is becoming more and more challenging and the dependency between (intermediate) analysis results is no more easy to understand for users involved in workflows. Therefore, a mechanism is desired, which is able to assist users in tracking and verifying distributed data analysis processes. In this paper, we extend the concept “data provenance” in the manufacturing domain to acquire information about the data origin and data changes. Furthermore, an architecture is proposed to manage provenance of process data, in which the data provenance is considered as annotation of process data. Different use cases are also given to show how data provenance can have impact on understanding and verifying data analysis processes in the manufacturing domain.

Current production planning and scheduling systems in automation do not meet the requirements of modern individualized production. Today's, static production processes impede customized manufacturing and small-scale production. A new way of thinking towards a dynamic control is required. This paper focuses on automated integrated process planning and scheduling on control level in discrete manufacturing. Existing algorithms in artificial intelligence planning are applied to solve process planning and scheduling problems. The challenge is to model the manufacturing system and products in a way that automated planners can generate efficiently process plans and schedules. Hence, based on a general classification of operations, different modeling options with regard to a successful automated process planning and scheduling are discussed. As a result, a domain modeling approach for discrete manufacturing is presented.

The complexity of industrial production systems is steadily growing. Hence, the plant stuff has to search in an increasing number of documents within the daily work routine, e.g. in manuals, commissioning instructions, service notes, shift books, process data, repair instructions, data sheets, R/I flow charts, CAD drawings etc. To support the plant stuff, an intelligent search engine for industrial production environments is proposed in this paper. Characteristics of the developed search engine with respect to the domain of industrial production environments, e.g. tailored synonym replacements and document classifications, are outlined. Particularly, two methods for document classifications, a k-nearest-neighbor classifier and a Naive Bayes classifier, are evaluated with documents from industrial production environments.

Machine learning techniques have a huge potential to take some tasks of humans, e.g. anomaly detection or predictive maintenance, and thus support operators of cyber physical systems (CPSs). One challenge is to communicate algorithms results to machines or humans, because they are on a sub-symbolical level and thus hard to interpret. To simplify the communication and thereby the usage of the results, they have to be transferred to a symbolic representation. Today, the transformation is typically static which does not satisfy the needs for fast changing CPSs and prohibit the usage of the full machine learning potential. This work introduces a knowledge based approach of an automatic mapping between the sub-symbolic results of algorithms and their symbolic representation. Clustering is used to detect groups of similar data points which are interpreted as concepts. The information of clusters are extracted and further classified with the help of an ontology which infers the current operational state. Data from wind turbines is used to evaluate the approach. The achieved results are promising, the system can identify its operational state without an explicit mapping.

Recent clustering based anomaly detection technologies classify new observations in different ways, e.g. using probability distributions, cluster centers or whole data points. Some of which suffer from high false classification rate, while others require high computational resources. In this paper, we propose a geometric approach to clustering based anomaly detection, in which the boundaries of clusters are utilized to classify new observations instead. To identify the cluster boundaries, a new algorithm for generating n-dimensional non-convex hulls has been developed. The proposed approach can improve the accuracy of clustering based anomaly detection, meanwhile, doesn't need high computational resources. Furthermore, it is universally applicable for any kind of cluster algorithms. The effectiveness of this approach is evaluated with real world data collected from different industrial automation systems.

Deep neural network based condition monitoring systems are used to detect system failures of cyber-physical production systems. However, a vulnerability of deep neural networks are adversarial examples. They are manipulated inputs, e.g. process data, with the ability to mislead a deep neural network into misclassification. Adversarial example attacks can manipulate the physical production process of a cyber-physical production system without being recognized by the condition monitoring system. Manipulation of the physical process poses a serious threat for production systems and employees. This paper introduces CyberProtect, a novel approach to prevent misclassification caused by adversarial example attacks. CyberProtect generates adversarial examples and uses them to retrain deep neural networks. This results in a hardened deep neural network with a significant reduced misclassification rate. The proposed countermeasure increases the classification rate from 20% to 82%, as proved by empirical results.

The automation and digitization creates a multitude of warnings, messages and alarms for plant operators. These so-called alarm floods have become an increasing issue in the industry regarding safety and plant downtime. A novel approach to reduce the overwhelming information presented to an operator during an alarm flood is based on causal models. The task of learning a feasible causal model especially of alarms is difficult. Therefore, we analyse the challenges of learning a causal model based on an industrial plant. Furthermore, we investigate the influence of different disturbances on this process and provide first solutions to address these challenges in the future.

The main feature of cyber-physical production systems is its adaptability. They adapt quickly to new requirements such as new products or product variants. Nowadays, a bottleneck is the automation system, for which high manual engineering efforts are needed: Today, on-site technicians write and rewrite automation software, configure real-time communication protocols and create system configurations consisting of machine timing, physical dimensions of products, sensitivity, and motor control accelerations and velocities. Cyber-physical production systems often solve this dilemma by relying on reusable software components, which are composed in the overall automation software. However, this solution comes with a price, reusable software components need free parameters to adjust to the individual production configurations. This paper addresses this central research question and presents a novel parameter estimation approach to choose automatically optimal system configurations for cyber-physical production systems. Different scenarios from discrete manufacturing plants are used to evaluate the solution approach.

Modern industrial plants become more complex and consequently monitoring them often exceeds the capabilities of human operators. Model-based diagnosis is a commonly used approach to identify anomalies and root causes within a system through the use of models, which are often times manually created by experts. However, manual modelling takes a lot of effort and is not suitable for today's fast-changing systems. Today, the large amount of sensor data provided by modern plants enables data-driven solutions and models can be learned from data, significantly reducing the manual modelling efforts. These data-driven solutions enable tasks such as condition monitoring: anomalies can be detected automatically, giving operators the chance to restore the plant to a working state before production losses occur. The choice of the model depends on a couple of factors, one of which is the type of the available signals. Hybrid timed automata are one type of model which separate the systems behaviour into different modes, e.g. `valve open' or `motor is running' through discrete events which are for example created from binary signals of the plant or through real-valued signal thresholds, defined by experts. The real-valued signals are then separated into the corresponding modes to improve the anomaly detection process in comparison to unseparated data. The anomaly detection for hybrid timed automata combines the detection of timing errors and sequence errors in the mode changes and the detection of anomalies in the real-valued signals. However, binary signals or expert knowledge to generate the much needed discrete events are not always available from the plant and automata can not be learned. The unsupervised, nonparametric approach presented and evaluated in this paper uses self-organizing maps and watershed transformations to allow the use of hybrid timed automata on data where learning of automata was not possible before.

Industrie 4.0 (I4.0) aims at a manufacturer-independent, vertical- and horizontal-oriented communication and cooperation within smart factories. This is only manageable using international standards. The so-called reference architecture model for Industrie 4.0 (RAMI4.0) and the requirement specification of an I4.0 component are already available. RAMI4.0 could be the basis for the interoperability of the interactions. The working group “Semantic and interaction model for I4.0 components” (GMA 7.20) made progress in this direction. The language used for the interaction of I4.0 components needs model definitions for the interaction consisting of the structure of the components, syntax for the description of the messages and the means to assign the meaning to the language elements. This paper discusses the results for a broader audience.

New products or product varieties require adapted processes and new production configurations of Cyber-Physical Production Systems. In this paper the focus is on production planning. The planning task is placed into the existing theory of automated planning. The system should automatically generate new production processes and react to new situations in dynamic environments. A new concept of planning and a new algorithm are presented, so that automated planning methods are applicable in real industrial environments.

A large part of the programmable logic controls (PLCs) used in industrial automation systems is based on automata, which are employed to model the different stages of the automated processes and to determine the discrete control signals. Complex PLCs are typically composed of several parallel automata, which are related to a subset of the IO signals, respectively. In this paper, a novel model learning approach is proposed, which allows to learn the parallel automata from the discrete IO signals during normal operation of the PLC. Learning the parallel automata is accomplished by means of a synchronous side-by-side decomposition of the overall system model. The side-by-side decomposition is based on the clustering of the correlation matrix computed between the individual IO signals. The learnt automata can be employed for automatic fault detection and visualization of the normal operation of the PLC. Evaluations are conducted for both a baseline method, where a single automaton is learned as model for the complete system, and the proposed learning algorithm for parallel automata. Experimental results show that the computed parallel automata are superior to a single automaton with respect to compactness, accuracy and fault detection capabilities.

Wireless communication systems such as bluetooth, WLAN and RFID gain more and more importance in industrial production systems. However, high requirements with respect to availability and determinism have to be met in industrial environments with several sources of interference such as frequency converters and welding machines. Finding the optimal parameter configuration, which allows for low power consumption and reliable operation, is in many cases timeconsuming and error-prone. In this paper, a self-optimizing RFID system is presented, which adapts both the transmission frequency and the transmission power of the RFID reader to the system environment. The proposed method allows for a robust communication with low energy consumption. Evaluation has been conducted for an application scenario in the SmartFactory OWL. In the investigated settings, transmission power of the RFID reader could be reduced from 27dBm to 15.2dBm on average.

In times of increasing connectivity, complexity and automation safety is also becoming more demanding. As a result of these developments, the number of alarms for the individual operator increases and leads to mental overload. This overload caused by alarm floods is an enormous safety risk. By reducing this risk, it is not only possible to increase the safety for humans and machines, but also to correct the failure at an early stage. This saves money and reduces outage time. In this paper we present an approach using a Bayesian network to identify the root cause of an alarm flood. The root cause is responsible for a sequence of alarms. The causal dependencies between the alarms are represented with a Bayesian network, which serves as a causal model. Based on this causal model the root cause of an alarm flood can be determined using inference. There exist different methods to learn the structure of a Bayesian network. To investigate which method suites the best for the purpose of alarm flood reduction, one algorithm from each method is selected. We evaluated these algorithms with a dataset, which is recorded from a demonstrator of a manufacturing plant in the SmartFactoryOWL.

Industrial plant operators regularly observe a high number of alarms generated in a short period of time, a phenomenon which is referred to as alarm flooding. This causes plant downtime, not only because of the repair time but also by the time needed to identify the root cause of machine failure - which is difficult during an alarm flood. Therefore, diagnosis tools that perform root cause analysis to advise plant operators can help reduce the downtime, which is a crucial issue in industry. We analyse the reproducibility and applicability of an existing approach by Ahmed et al. (2013) which is based on agglomerative hierarchical clustering where raw data in the form of alarm logs is preprocessed, floods are detected, and then clustered. The aim is, that resulting clusters represent floods that originate from the same common root cause. We extend the approach with alternative similarity measures and perform experiments regarding their effectiveness in structuring industrial alarm flood data. In our evaluation we use a real industrial use case which contains more diverse data and a larger amount of data points compared with the original study.

One of the many opportunities Industrie 4.0 (I4.0) offers, is to reduce (re-)configuration effort. Today, a variety of vendor-specific information models prevents an efficient configuration of Condition Monitoring (CM) for adaptable Cyber-Physical Production Systems (CPPS). The manual configuration of signals that should be monitored is a tedious and error-prone task. Standardization efforts for I4.0 focus on a generic interface called Asset Administration Shell (AAS), which could increase configuration efficiency. The AAS should represent information, e.g. about components, machines, and plants, based on properties standardized according to IEC 61360. Although frequently used in planning and procurement, benefits of properties for the integration of CPPS have not yet been demonstrated. Additionally, important details for the implementation of the AAS, e.g. its deployment and technical requirements, have not yet been specified. In this paper, an information model for condition monitoring of a servo motor is defined based on IEC 61360 properties. This information model is included in an AAS that is implemented on the Programmable Logic Controller (PLC) of a work cell from the company Lenze. The PLC communicates with a condition monitoring service via OPC Unified Architecture (OPC UA). Using IEC 61360 properties the type of a servo synchronous motor can be identified and thresholds for this motor can be configured without manual configuration effort. Nevertheless, in order to implement the AAS in an I4.0 compliant way, IT-Security and Message-based I4.0 communication have to be supported by the PLC. The effort for the (re-)configuration of condition monitoring services can be reduced based on IEC 61360 properties. The approach introduced in this work could help to achieve the main goals of I4.0: To improve the flexibility and efficiency of adaptable CPPS. Nevertheless, more properties and information models have to be standardized in the future.

In recent years, model-based testing (MBT) of automation systems has gained in importance. However, increasing size and complexity of manufacturing plants also lead to larger models, which again cost time and manpower for modeling tasks. An approach to solve this problem is to subdivide the overall model into several separated models of usual components of automation systems which are reusable. Therefore, we introduce the Synchronized Depth First Search (SDFS), which uses a sub-divided automaton model to generate test cases. The underlying automaton model needs to incorporate synchronous transitions in order to synchronize processes among each other and thus forming an overall model to test with.

Trends in novel manufacturing systems lead to an increased level of data availability and smart usage of these data. Nowadays, many approaches are available to use the data, but because of an increased flexibility of the systems the interaction between machines and humans has become a challenge. Humans have to browse through a huge amount of data, need knowledge about the machine and underlying algorithms to interpret the results; they cannot use their known terms for communication, we call it the conceptual gap. The user should be enabled to communicate with the machine on a more abstract level and in a more natural way. Therefore, a natural language layer is introduced to provide users with a familiar interaction interface. Underlying layers contain knowledge about the domain, the machines and how data can be accessed and processed. This enables users' questions such as “Are there any anomalies in the system?” to be answered. Answers are provided in natural language and evaluated with a test set of 204 questions.

S.280-287

Manufacturing plants become more complex as the desire for modern individual products increases. Programming such plants is a challenging task that consumes a lot of time. This paper proposes a new control-code synthesis algorithm, which aids the programmer by automatically generating parts of the control-code. With the new algorithm, the programmer only has to specify and parametrize the general production process, e.g. drill a hole then paint the workpiece. The control-code for all intermediate processes, like transporting the workpiece from the drilling machine to the paint-spray station is generated automatically. This saves much engineering time and enables the programmer to focus on more challenging tasks, such as process optimisation.

Along with the rapid growth of the system complexity, the capability of self-diagnosis is desired by monitoring complex industrial systems to reduce the unplanned system downtimes. By applying data driven analysis methods such as clustering algorithms on the process data of industrial systems, the health status of systems can be deduced and the anomalous statuses can be automatically detected. The accuracy of clustering based anomaly detection using cluster centers is highly dependent on the geometry of the given data set. By a data set with unsymmetrical and concave boundary, using cluster centers as reference to measure the similarity between new observations and clusters normally leads to a high false alarm rate. This paper presented an approach to improve clustering based anomaly detection by building concave hulls for each cluster. For this purpose, a new algorithm for generating n-dimensional concave hulls is developed. The effectiveness of this approach is evaluated with real world data collected from wind turbines.

Cyber-physical production systems' main feature is adaptability. They shall adapt quickly to new requirements such as new products or product variants. Nowadays, the bottleneck is the automation system, which requires high manual engineering efforts for every new adaption. For new requirements, an automation software is created by combining pre-defined software components. But this also means that software components need degrees of freedom in form of parameters, such as timing parameters, to be applicable to new requirements. This paper presents a solution to determine parameters automatically for the automation software of cyber-physical production systems. A scenario from discrete manufacturing illustrates the underlying concepts.

In the present work, fault detection in industrial automation processes is investigated. A fault detection method for observable process variables is extended for application cases, where the observations of process variables are noisy. The principle of this method consists in building a probability distribution model and evaluating the likelihood of observations under that model. The probability distribution model is based on a hybrid automaton which takes into account several system modes, i.e. phases with continuous system behaviour. Transitions between the modes are attributed to discrete control events such as on/off signals. The discrete event system composed of system modes and transitions is modeled as finite state machine. Continuous process behaviour in the particular system modes is modeled with stochastic state space models, which incorporate neural networks. Fault detection is accomplished by evaluation of the underlying probability distribution model with a particle filter. In doing so both the hybrid system model and a linear observation model for noisy observations are taken into account. Experimental results show superior fault detection performance compared to the baseline method for observable process variables. The runtime of the proposed fault detection method has been significantly reduced by parallel implementation on a GPU.

Cyber-Physical Production Systems (CPPSs) are in the focus of research, industry and politics: By applying new IT and new computer science solutions, production systems will become more adaptable, more resource ef- ficient and more user friendly. The analysis and diagnosis of such systems is a major part of this trend: Plants should detect automatically wear, faults and suboptimal configurations. This paper reflects the current state-of- the-art in diagnosis against the requirements of CPPSs, identifies three main gaps and gives application scenarios to outline first ideas for potential solutions to close these gaps.

The automated planning in Cyber-Physical Production Systems (CPPS) reduces the need for the human-error-prone repetitive planning task. Modeling CPPS and the required products for planning purposes can be done with domain-independent planning languages, like Planning Domain Definition Language or Prolog. However, such domain-independent languages do not satisfy all requirements in CPPS, like concurrency and/or time. Moreover, production engineers should deal with the planning models of their systems and, therefore, such models must be usable (understandable) for these users. Unfortunately, these requirements are missing in the available domain-independent planning languages. To overcome the previous problems, we develop an object-oriented planning language specific for the domain of CPPS. To evaluate the CPPS requirements' satisfaction, we use the language to represent the planning problem of two simulated robotic arms cooperating on an assembly task.

Probabilistic machine learning approaches has been successfully applied in various applications and is gaining more and more popularity. But the success of such approaches are based on the quality of the data. Getting quality data is the biggest challenge for most of the real-life applications and our application domain, i.e. industrial cleaning process, is no exception. In our application domain, the data collection is mostly performed manually without using any standards and is highly influenced by the expertise and interpretation of individual cleaning personnel. We have developed a Bayesain predictive assistance system (BPAS) that uses a real-life cleaning data to provide decision support to the cleaning personnel. In this paper, we extend our BPAS and propose a hybrid approach to develop an assistance system for resource optimization in industrial cleaning processes. The proposed approach, which combines Bayesian network and rule-based system, aims at increasing the robustness and the stability of the assistance system.

This paper presents an approach for exploiting multicore hardware architectures on coding level for the IEC 61131-3. An interface between the IEC 61131-3 code and software of a different programming language outsources the actual parallel workload. For validation purpose, an embedded multicore hardware is used as a controlling device, which executes software for the use case of model based condition monitoring. The case study results show an explicit benefit of the multicore exploiting software in comparison to its singlecore counterpart, which is reflected with a faster processing of up to a factor of 3. Overall, this approach can be used for developing high performance applications or for accelerating existing applications in industry.

In this paper, fault detection in piecewise stationary industrial processes is investigated. Such processes can be modeled as sequences of distinct system modes in which the respective expectation values and variances of process variables do not change. In particular, piecewise stationary processes with autonomous transitions between system modes are considered in this work, i.e. processes without observable trigger events such as on/off signals. A Hidden Markov Model (HMM) is employed as underlying system model for such processes. System modes are modeled as hidden state variables with given transition probabilities. Continuous process variables are assumed to be Gaussian distributed with constant second order statistics in each system mode. A novel HMM-based fault detection method is proposed which incorporates the Viterbi algorithm into a fault detection method for hybrid industrial processes. Experimental results for the proposed fault detection method are presented for a module of the Lemgo Smart Factory.

This paper addresses the efficient generation of models for cyber-physical systems from large historical data sets. A cyber-physical system (CPS) is a system composed of physical subsystems together with computing and networking. CPS models are required for monitoring and control of the physical processes. Such models are in general hybrid models that take into account both discrete control signals and continuous system behaviour. Model-learning is the key to a new generation of intelligent automation systems: Automatic generation of models from system observations allows to model complex CPS in cases where manual model creation is time-consuming, expensive or not even possible. In general, the quality of the generated models increases with the size of training data. However, model learning from large historical data sets is in many cases time-consuming. For this reason, MapReduce algorithms are proposed in the present work that allow for efficient model learning.

Bayesian networks (BNs) have been used in different contexts of decision support solutions such as directive, strategic, tactical and operational. These contexts differ from each other only in the realization of the decision support in terms of time. The real-time implementation of BN in an embedded system for resource optimization is very challenging because of the low computation capacity in embedded systems and, to the best of our knowledge, has not been reported yet. In this paper, we present a BN based predictive assistance system that uses real-life data to perform the real-time decision support in industrial cleaning processes.

In this paper, stochastic models for fault detection in industrial automation processes are investigated. Thereby, nonlinear, time-variant systems are considered. The basic idea consists in building a probability distribution model and evaluating the likelihood of observations under that model. In contrast to the existing methods, this paper considers the practically important case in which measurement noise is negligible and all process variables are observable. This assumption allows the direct evaluation of a probability distribution for fault detection without approximations such as second order statistics or particles. The main part of this paper deals with adequate models for this probability distribution such as Gaussian and Hidden Markov models. Such models require predictions of the expectation values of the respective probability distributions. Regression models such as (multivariate) linear regression models and neural networks are investigated for this purpose. Evaluations are conducted with respect to prediction accuracies and fault detection capabilities of the employed models. Evaluations show superior results of the novel approach compared to existing fault detection methods, which are based on approximations such as second order statistics.

The design processes of todays more and more complex automation systems require computer-based support to maintain their manageability. As a base for that, the authors introduce a holistic design approach for these systems. Requirements on the system to be designed are represented by an extended feature model which serves as consistent requirements model during the entire design process. A grammar-based synthesis applies formalised expert knowledge to generate solutions to these requirements. The paper's main contribution is to combine formalisms from overlapping areas of artificial intelligence and software engineering to obtain a holistic design process for industrial automation systems.

In this paper, a new modeling and learning approach is presented which is based on two assumptions from the field of psychology: 1. The number of Tweets mainly depends on previous dynamics of the discussion, i.e. a state-space modeling approach is used for the first time. 2. Humans mainly react to emotional stimuli, i.e. Tweets are automatically characterized by their emotional content. Therefore, the emotions of conversations are extracted and used for system identification and parameter estimation of a state space model, which deals with events and its transitions.

The problems of planning production processes in Cyber Physical Production Systems (CPPS) require information about e.g., machines, products, and time. These information can be modeled with languages which have different characteristics. Production engineers must select a language carefully when modeling their problems. Otherwise, they have to put an extra effort (e.g., to work around temporal constraints) later when selecting an unsuitable language. A suitable selection can be achieved by a comparison that provides engineers with decision support. In related work, comparisons among such languages exist. However, these comparisons are not tight to the context of CPPS. To provide a decision support, we suggest criteria to compare among languages when used for planning in CPPS. We apply these criteria on the languages: Prolog, Timed Automata, and PDDL. Other languages will be considered in our future work.

Plug and produce (PnP) aims at reducing system engineering effort. Therefore several PnP aspects have to be solved. This paper gives an overview and classifies process types and PnP aspects. Furthermore it identifies coupled processes as the most challenging ones and provides a new engineering approach for such systems. A new two-step descriptive engineering approach is applied to automatically synthesise control code from a given product specification. The main idea of the approach is to use a descriptive view rather than a prescriptive. This means that the engineer specifies what he wants to produce and no longer how he wants to produce it. Thus, it reduces engineering effort and releases more resources to optimize the process.

Electricity, water or air are some Industrial energy carriers which are struggling under the prices of primary energy carriers. The European Union for example used more 20.000.000 GWh electricity in 2011 based on the IEA Report [1]. Cyber Physical Production Systems (CPPS) are able to reduce this amount, but they also help to increase the efficiency of machines above expectations which results in a more cost efficient production. Especially in the field of improving industrial plants, one of the challenges is the implementation of anomaly detection systems. For example as wear-level detection, which improves maintenance cycles and thus leads to a better energy usage. This paper presents an approach that uses timed hybrid automata of the machines normal behavior for a predictive maintenance of industrial plants. This hybrid model reduces discrete and continuous signals (e.g. energy data) to individual states, which refer to either the present condition of the machines. This allows an effective anomaly detection by implementing a combined data acquisition and anomaly detection approach, and the outlook for other applications, such as a predictive maintenance planning. Finally, this methodology is verified by three different industrial applications.

In this paper, the authors present an engineering approach for generating automation system solutions based on formalised requirements. This enables assistant systems which guide engineers during the design phase of todays more and more complex automation systems. A software prototype is used for the evaluation of this approach in practice. The main contribution is to directly use a formal requirements model as input for the automated synthesis of automation systems and to formalise the expert knowledge for this synthesis. The result are consistent, maintainable automation systems and with that shorter and reproducible development cycles.

The efficient handling of complex production systems and the implementation of more flexible and adaptable production lies at the heart of cyber-physical production systems and its german equivalent Industry 4.0. Such scenarios currently face one main difficulty: the creation, configuration and maintenance of the corresponding automation software is time-consuming and error-prone. Two main solutions exist for this problem: (i) model-based software development and (ii) intelligent automation, i.e. the usage of new knowledge-based solution approaches. This article compares these different solutions by applying them to three phases of the life-cycle: the planning phase, the operation phase and the plant modification phase.

Optimizing the resource consumption by the products (machines) and making them environment friendly is the aim of almost all producers today. May it be due to cost of resources, their limited availability, their affect on the environment or consumer awareness. Ample research is being carried out at national and international level for resource optimization. Adding intelligence and learning capability is being increasingly used as an approach for resource optimization. Different methods and models for machine learning are available in the literature. Bayesian network is one of the widely used learning model for resource optimization in wide range of applications [1], [2]. In this paper, we present the use of Bayesian network for resource optimization and decision support system in an industrial cleaning process. The proposed Bayesian predictive assistance system assists the cleaner in choosing the optimal parameters and would be a self-learning system that stores the successful cleaning results in a global database for future cleaning cycle.

Today's production plants are not conceivable without automation systems. Due to the increasing complexity of production plants and therefore of automation systems, delays and interruptions in automation projects are observed. A design model for more efficient planning of industrial automation systems is introduced. It is based on a new and practical proceeding for the construction of a requirements model. Extended feature models as formal requirements representation enable to verify the consistency of requirements. New insights on the necessary capabilities of a formal reasoning system for planning the whole automation system are deduced from the design model.

Model-learning is the key to the new generation of intelligent automation systems: Without the automatic generation of models from system observations, models of the plant's behavior will not be available for most systems. And without such models, no intelligent capabilities such as self-diagnosis or self-optimization can be implemented. This paper therefore presents a novel classification schema for systems, models and model learning algorithms. Based on this analysis of open research questions, the new learning algorithm HyBUTLA is presented. In this paper, this solution approach is applied to the analysis and diagnosis of the energy consumption of production plants. To the best of the authors' knowledge, this is the first learning and adaptable energy anomaly detection solution for complex hybrid production systems.

This paper contributes a framework that helps to fulfill the requirements of the standards DIN EN 16247 and ISO 50001 by combining (i) a synchronized data acquisition, (ii) data integration, (iii) learning of normal behavior models and (iv) a implementation of an anomaly detection as prototype. Both standards require a reliable data acquisition and energy consumption analysis for implementing a certified energy management system. It shows that this framework meets the specifications of the standards by implementing a combined data acquisition and anomaly detection approach.

In the presented work, the detection of anomalous energy consumption in hybrid industrial production systems is investigated. A model-based approach with a timed hybrid automaton as overall system model is employed for anomaly detection. The approach is based on the assumption of several system modes, i.e. phases with continuous system behavior. Transitions between the modes are attributed to discrete control events such as on/off signals. The underlying discrete event system which comprises both system modes and transitions is modeled as finite state machine. The focus of this paper is set on the modeling of the energy consumption in the particular system modes. Sequences of stochastic state space models are employed for this purpose. Model learning and anomaly detection for this approach are considered. The proposed approach is further evaluated in a small model factory. The experimental results show significant improvements compared to existing approaches to anomaly detection in hybrid industrial systems.

A tailored model of a system is the prerequisite for various analysis tasks, such as anomaly detection, fault identification, or quality assurance. This paper deals with the algorithmic learning of a system’s behavior model given a sample of observations. In particular, we consider real-world production plants where the learned model must capture timing behavior, dependencies between system variables, as well as mode switches—in short: hybrid system’s characteristics. Usually, such model formation tasks are solved by human engineers, entailing the well-known bunch of problems including knowledge acquisition, development cost, or lack of experience. Our contributions to the outlined field are as follows. (1) We present a taxonomy of learning problems related to model formation tasks. As a result, an important open learning problem for the domain of production system is identified: The learning of hybrid timed automata. (2) For this class of models, the learning algorithm HyBUTLA is presented. This algorithm is the first of its kind to solve the underlying model formation problem at scalable precision. (3) We present two case studies that illustrate the usability of this approach in realistic settings. (4) We give a proof for the learning and runtime properties of HyBUTLA.

This paper presents a novel approach for data acquisition in distributed, heterogeneous automation systems as a basis for new condition monitoring, anomaly detection and visualization applications. The described solution enables process data acquisition in real time Ethernet systems, using the Precision Time Protocol (PTP) from IEEE 1588 for synchronization and the OPC Unified Architecture for data presentation. As a proof of concept the architecture is implemented using standard hardware components and a x86/x64 architecture without any special operating system. As verified later this setup already achieves a synchronization accuracy of <;10ms, sufficient for the majority of production processes to be monitored.

This paper presents a novel model-based approach for the prediction of energy consumption in production plants in order to detect anomalies. A special Ethernet-based data acquisition approach is implemented that features real-time sampling of process and energy data. Hybrid timed automaton models of the supervised production plant are generated and executed in parallel to the system by using data samples as model input. According to comparisons of predicted energy consumption with the production plant observations, anomalies can be detected automatically. An evaluation within a small factory shows that anomalies of 10 % differences in energy consumption, wrong control sequences and wrong timings can be detected with a minimum accuracy of 98 %. With this approach, downtimes of production systems can be shortened and atypical energy consumptions can be detected and adjusted to optimal operation.

Due to an increasing degree of automation, industrial communication networks grow in complexity and become more vulnerable to errors. Therefore a reliable network diagnosis is getting more and more important. Although modern industrial communication networks, such as PROFINET, have basic diagnosis mechanisms, they do not cover complex error cases. Furthermore, our work shows that a diagnosis of most error cases requires an automatic detection of a network's topology. In general, PROFINET does support topology discovery by providing the necessary protocols. However, implementing a reliable automatic topology discovery is not a trivial task e.g., because different vendors vary in their interpretation of the standards. This paper presents our approach for discovering the topology of a heterogeneous network. Further, a first approach for using the topology discovery for a high-level PROFINET diagnosis is shown.

The increasing complexity of today's production systems and the variety of model-based approaches to their monitoring, diagnosis and testing emphasize the importance of the modeling step. Modeling is mostly done manually, in a costly and time-consuming way. In this paper, an alternative that comes from the learning theory is given: an automated procedure for identifying behavior models from recorded observations. Assuming the system's structure is known, the algorithm presented here is capable of learning behavior models for its components. The algorithm accounts for probabilistic, timing, discrete and continuous aspects of the given system, using the modeling formalism of hybrid automata. The practical usability of identified models is demonstrated using an anomaly detection application for a real production system.

Production facilities have to cope with fast changing market demands-leading to a need for high adaptability of the manufacturing plant. Especially the manual adaption of the automation systems causes high costs and significant downtimes. Here, an approach for an automatic adaption of the automation system is presented. The major challenges for an automatic adaptability of the automation system are (i) dynamically modifiable networks, (ii) a real-time middleware transporting variables within the network independently of the automation topology, and (iii) a semantic-based mechanism that allows one plant module to identify which signals are needed from another module. This paper describes a real-time middleware for industrial automation (mINA). It focuses especially on problem (iii), the mechanism to discover and identify required process signals. The functionality is based on a newly defined description language (mINA-DL), i.e. the middleware can use further semantic information about the signals. The middleware (ii) communicates by means of a real-time publish/subscribe concept and provides OPC UA compliance which also allows an easy integration with the manufacturing execution system (MES). Furthermore, a first prototypical implementation of the middleware is described and assessed using a test environment.

In automotive software and system design, explicit system and especially software models have only recently found their way into the development process. This paper will try to give an overview for what such models have so-far been used and which advantages they brought to vehicle manufacturers and suppliers.

An in-depth analysis of traffic data provides valuable insights into network traffic structures. This paper presents both methods for such an in depth analysis and their implementation within the system STRUCTUREMINER. Moreover, it is shown in which way the analysis performed by STRUCTUREMINER can be used to tackle several administrative network tasks.

The problem of finding a pleasant layout for a given graph is a key challenge in the field of information visualization. For graphs that are biased towards a particular property such as tree-like, star-like, or bipartite, a layout algorithm can produce excellent layouts—if this property is actually detected.