Intelligent algorithms and learning are the basis for evolving smart cyber-physical production systems (CPPSs) and Industrie 4.0. A smart image detection algorithm shall be added to reduce downtime due to the extended operation of glass bottles in a yogurt producing plant. To support the engineers in doing so, a comprehensive domain-specific language (DSL), DSL4hDNCS, is introduced, enabling rapid analysis, addressing hardware/software architectures and network-related delays and uncertainties. DSL4hDNCS is defined by a metamodel to avoid ambiguity and enriched by aspects, such as safety, calculation power, and network transmission time. DSL4hDNCS is used to compare (re)deployment alternatives using different technologies, such as edge, fog, and cloud to implement the additional smart algorithm. The evaluation of DSL4hDNCS using an acknowledged Industrie 4.0 demonstrator plant as a case study confirmed the benefit for engineers during the redesign.

Machine learning and artificial intelligence provide methods and algorithms to take advantage of sensor and actuator data in automated production systems. Product quality monitoring is one of the promising applications available for data-driven modeling, particularly in cases where the quality parameters cannot be measured with reasonable effort. This is the case for defects such as cracks in workpieces of hydraulic metal powder presses. However, the variety of shapes produced at a powder press requires training of individual models based on a minimal sample size of unlabeled data to adapt to changing settings. Therefore, this article proposes an unsupervised product quality monitoring approach based on dynamic time warping and non-linear regression to detect anomalies in unlabeled sensor and actuator data. A preprocessing step that isolates only the relevant intervals of the process is further introduced, facilitating efficient product quality monitoring. The evaluation on an industrial dataset with 37 samples, generated in test runs, shows a true-positive rate for detected product quality defects of 100% while preserving an acceptable accuracy. Moreover, the approach achieves the output within less than 10 seconds, assuring that the result is available before the next workpiece is processed. In this way, efficient product quality management is possible, reducing time- and cost-intensive quality inspections.

The availability of vast amounts of data in automated production systems reveals the potential for data-driven improvements. Jointly using this data across different sites or even across different companies will further increase the validity of data-driven models. However, the throughput in wide area networks is limited, limiting the large-scale transmission of data. Therefore, this paper proposes a data reduction approach to reduce network load based on regression and time series models directly on the shop floor. The machine-learning models are used to predict the signals of the automated production system to prevent the transmission of extensive raw data. It is shown that the approach reduces the network load significantly while still ensuring the fulfillment of the real-time control tasks of the programmable logic controller at any time. Thereby, the reduction of the data is dependent on the error of the reconstructed data that can be tolerated.

For modeling Networked Control System (NCS) a standardized notation is still missing. Approaches like the Unified Modeling Language (UML) or the Systems Modeling Language (SysML) are widespread and accepted, yet remain very general and do not offer support for domain specific problems. Additionally, UML/SysML do not offer graphical support to model domain specific solutions e.g. for network architectures in NCS. UML and SysML-based approaches like MARTE, TURTLE, UML-QoS or SysML-Sec provide more guidance to model network architectures, embedded systems and network systems. However, they still lack a well-defined graphical notation, also referred as concrete syntax, to support the modeling process. To strengthen the modelling, an existing graphical notation is extended, which enables traceability to requirements and the modelling of safety-relevant properties. As case study, a train network architecture is provided to show the feasibility of this notation. Together with industrial experts, the communication architecture of a train-part is modeled and linked to its requirements. The graphical representation includes the necessary information to define the communication in the NCS and is understandable for engineers of the domain without UML/SysML knowledge. The approach was evaluated with experts from industry.

The fourth industrial revolution, Industrie 4.0, motivates research to adapt concepts from Internet of Things and Cyber-Physical Systems to automation. Focus is placed on integrating all data sources, from sensors and actuators to management software, to achieve interoperability of heterogeneous systems. To assure data integration, a data collection architecture that guarantees data access from all systems, including legacy devices, is needed. This allows companies to gradually evolve to Industrie 4.0 deployments while still keeping their mission-critical systems online. A middleware-based architecture is a valid answer to the presented challenges. Numerous communication technologies can act as a middleware, all with different advantages and disadvantages. A technology-independent implementation ensures that a lock-in to a specific technology is prevented and makes the architecture more robust and adequate for various use-cases. This work contributes with a concept of a middleware-based data collection architecture, which is independent of specific communication protocols. Therefore, a protocol-agnostic communication interface is developed. A technology comparison summarizes relevant protocols in the field of industrial automation. The concept is subsequently implemented for a case-study and compared to a peer-to-peer legacy approach. Results show that the concept is feasible, as well as that it reduces complexity and effort for implementation and migration of such communication systems. Future work should focus on further developing the concepts, in order to support additional communication protocols and verify the suitability of the architecture in scenarios with a multitude of technologies.

The availability of huge amounts of process data enables data-driven methods to optimize production processes. Predictive Maintenance is one of the common applications to transfer data to useful information for improving the Overall Equipment Effectiveness. In this paper, a data-driven method for condition monitoring of control valves in industrial process plants is developed based on data collected during test runs. In contrast to the state of the art condition monitoring in control valves, the test runs make it possible to define a threshold that differentiates normal from abnormal valve behaviour. Furthermore, the characteristics of the model results allow the identification of different defects. The application of the proposed method to a historic industrial data set validate the applicability in noisy industrial use cases.

The engineering of automated Production Systems (aPS) involves various disciplines (i.e. mechanical, electrical and software engineers). In order to address discipline-specific objectives in the engineering process, a multitude of models are developed. The information within these models is represented by different variants of engineering documents. For instance, the information contained in an ECAD model can be stored in XML-based documents or can be represented by printed, paper-based electric circuit diagrams. Models and corresponding documents change cyclically, as the engineering process of an aPS is generally not straightforward. Consequently, changes made to those documents need to be propagated back to their original model, so as to maintain consistency. Depending on the meaning and the importance of the change, further action may be required (e.g. re-design of components). This paper proposes and evaluates the retrieval and model-based reconstruction of information from paper-based electric circuit diagrams in aPS, containing handwritten annotations.

Internet technologies are driving developments in networking technologies for mechatronic devices. In the plant and manufacturing automation domain, internet technologies support for example cloud-based data exchange and a seamless connection of the Enterprise Resource Planning (ERP) and Manufacturing Execution System (MES). The communication of sensor-actuator loops on the field level is tackled with fieldbus technologies. Fieldbuses were introduced to tackle requirements such as real-time, safety, reliability and low costs that could hardly be met with Ethernet-based communication. Therefore, a variety of proprietary fieldbuses exists for the communication at the field level. For data exchange between different fieldbuses or for data upstream to higher automation layers, gateways to process data with significant time delays are mandatory. To overcome the shortcomings of Ethernet on the field level different approaches were proposed, labeled as Industrial Ethernet or Real-Time Ethernet (RTE). With Ethernet-based communication technologies, a vertical and horizontal integration of data is supported. While in simple or homogeneous networks, the estimation of Quality of Service (QoS) via simulation or calculation is possible; In complex architectures with heterogeneous protocols, different network layers, switches and gateways measurements are required. In this approach, a testing environment is presented to analyze different Ethernet-based technologies implemented on the single board computer i.e. Raspberry Pi. Therefore, measures for software configuration and a methodic analysis workflow are presented. This is beneficial to estimate QoS in planned communication architectures including different Ethernet-based protocols and gateways after defining requirements, functions and test scenarios.

Knowledge on the operating conditions of chemical processes is important for various applications, for instance failure diagnosis. In absence of the recorded information, it has to be reconstructed from other data. This is commonly carried out by process experts as the identification of the operating states requires a deep understanding of the underlying process. As manual classification is a time-consuming task, an automatic identification can simplify this task greatly. As data-driven approaches often fail due to the complex nature of the processes, two hybrid approaches are proposed in this paper, supporting the expert during the classification. Big data methods are used for processing with experts having the chance to influence and evaluate the algorithms and results. For identification, a k-Means clustering and a combination of self-organizing maps and hidden Markov models are used. While minimizing the expert effort for classification, the results still have to be reliable. Both approaches performed accurate and decreased the expert effort significantly. Future studies are centered on combining both methods as their strengths complement each other.