Accurately identifying the root causes of alarm floods is essential for safe and efficient operations of industrial processes. While similarity analysis methods have been developed for analyzing and clustering alarm floods, they are insufficient to pinpoint root causes. Thus, recent studies have attempted to train classification models to identify alarm flood root causes usually based on substantial amounts of labeled training data, which is impractical and costly in real-world scenarios. Accordingly, this article proposes a semisupervised method named alarm dual-graph fusion-based manifold regularization for root-cause identification (RCI) of industrial alarm floods, which can achieve improved performance by utilizing both limited labeled alarm flood sequences and substantial unlabeled sequences. The contributions are as follows: first, a local sequence alignment-based graph construction method is proposed to construct an undirected graph for historical alarm flood sequences; second, a modified manifold regularization method based on dual-graph Laplacian fusion is developed for semisupervised alarm flood RCI modeling; third, an online RCI strategy is devised for recognition of root causes of incoming alarm flood sequences. The effectiveness of the proposed method is demonstrated by a case study with alarm data generated by the public vinyl acetate monomer process simulation model.

Modeling the evolution process of industrial cascading faults by capturing causal relationships has achieved widespread attention in the attack detection and security control field. However, the highly coupled nature of numerous sensors and control loops in complex systems makes it difficult for existing root cause identification (RCI) methods to accurately and clearly characterize the fault-induced causal relationships. To address this challenge, this article proposes the hierarchical causal graph neural networks (HCGNNs) to reveal the root cause in an end-to-end manner with the aid of cascading failure analysis. First, the cause–effect weighted adjacency matrix is designed to expand the connections between nodes from correlations to causalities in a graph neural network (GNN). On this basis, the proposed hierarchical network structure can collaboratively learn the intraunit and interunit causalities, which significantly reduces the spurious and redundant causal relations while improving the modeling efficiency. Additionally, a new cascading failure analysis method is formulated to quantify the root cause and clarify propagation paths from a more objective perspective. Finally, the effectiveness and superiority of the proposed method are verified by the Tennessee Eastman platform and the real-world coal mill group.

Microwave filters (MFs) are indispensable in communication systems for selecting specific frequency signals. The tuning of MFs is a demanding and time-consuming task, which can be addressed by the inverse decision-making model (IDMM). However, two main challenges arise in the sampling process for IDMM, namely, low efficiency due to the large number of samples and poor adaptability in the presence of uncertain initial positions. To overcome these challenges, a fuzzy nonuniform sampling (FNUS) method is proposed, leveraging the flexibility of the fuzzy logic system. Specifically, an adaptive sampling framework based on a fuzzy logic system is presented to handle the uncertainty of initial positions. Under this framework, a nonuniform sampling approach is devised to collect fewer samples far from the target and more samples close to the target. Given the similarity and single-sided distribution of samples in the raw dataset oriented to modeling, the tailored enhancement strategies are designed to improve dataset quality. Finally, the efficiency and adaptability of FNUS are demonstrated to be superior to the existing methods through simulations. Furthermore, the practicality of FNUS is validated by experiments on physical MFs.

Autonomous driving systems are fundamentally composed of sequential modular tasks, i.e., perception, prediction, and planning. For connected autonomous vehicles (CAVs), cooperative perception offers a promising solution to surpass their perception limitations, such as occlusion, by sharing sensing data with each other through wireless communication. Existing works typically prioritize sharing data from potential object-containing areas to maximize object detection accuracy under constrained communication resources. However, such detection-oriented approaches ignore a crucial fact that more accurate detection does not equal safer planning. Sharing large amounts of sensing data for detection accuracy can lead to communication resource wastage and performance degradation of subsequent driving tasks. To address this, we introduce Plan2comm, a communication-efficient cooperative perception framework via planning-oriented feature sharing, which shares only sensing data around planned trajectories to enable safer planning rather than mere detection accuracy. Specifically, a planning-oriented communication mechanism is designed to select and transmit the most valuable features from the perspective of the planning task. Moreover, an uncertainty-aware spatial-temporal feature fusion strategy is proposed to enhance high-quality information aggregation. Comprehensive experiments demonstrate that Plan2comm outperforms all other cooperative perception methods on motion prediction performance, and is more communication-efficient.

Alarm systems serve as the first layer of protection for modern process industries to monitor industrial processes and ensure operational safety. However, the presence of alarm floods is common in alarm systems and may distract operators from critical alarms. To provide decision supports to operators during alarm flood situations, alarm prediction becomes a potential effective solution. The difficulty of alarm prediction lies in how to achieve high prediction accuracy over long-term consecutive alarm monitoring periods, including situations with either high or low alarm rates. Consequently, this paper presents an industrial alarm prediction method for both alarm flood periods and non-alarm flood periods. Specifically, a strategy of combining fixed-length and fixed-time sliding windows is designed. Further, an alarm prediction model based on Informer is proposed for alarm prediction under different alarm rates. A case study is presented to prove the validity of the proposed alarm prediction method.

Microwave cavity filters are essential electromechanical coupling devices in communication systems. Structural-parameter tuning by experienced operators improves the filter performance but is demanding and time-consuming. The automatic tuning method has received extensive research attentions using data-driven modeling approaches. However, two main issues affect the accuracy and efficiency of the model construction: 1) features of tuning processes, as model inputs, have limited adaptability and extraction accuracy to different resonant states and 2) models require plentiful training data and the training process is time-consuming. Thus, dynamic hybrid models are developed in this study with self-selected inputs, self-organized samples, and a self-learning structure. First, spatial features are extracted to flexibly depict the tuning characteristic, and double-domain (spatial or circuital) features are selected adaptively to accommodate distinct resonance states. Second, a trustworthiness-curiosity-driven active sampling method is exploited to attain fewer and better-training data. Third, an improved glsms broad learning system acrlong BLS is developed using new modules of incremental node calculation and weight pruning, characterized by more lightweight and flexible structures. The proposed method is effective and flexible demonstrated by simulations and experiments, and the tuning task of microwave cavity filters is fulfilled in a more accurate and efficient manner.

Alarm systems are commonly deployed in modern industrial facilities to monitor process operations. However, due to the presence of nuisance alarms and alarm floods, their efficiencies are much degraded. Especially, alarm floods are among the most difficult issues in industrial alarm management and recognized as the main causes of many industrial accidents. To address alarm floods, this article proposes a root cause identification (RCI) method for industrial alarm floods based on word embedding and few-shot learning. The contributions are threefold: 1) A textual encoding method based on word embedding is proposed to convert alarm messages into numerical word vectors that can be used in the modeling of RCI; 2) an alarm-priority-based adaptive weighting strategy is designed to make the RCI more sensitive to alarms of higher priorities and appearing earlier; 3) a few-shot learning method based on long short-term memory is adapted to identify the root causes of alarm floods based on limited instances of labeled data. The effectiveness and superiority of the proposed method are demonstrated by a case study based on data from the vinyl acetate monomer public model.

Prompt detection of bit bounce can prevent serious incidents and is of great importance for safe and efficient deep geological drilling. In the early stage of bit bounce, signal changes are relatively weak. In addition, there are differences in the topological relationships of samples at different time instances in normal state and bit bounce. These factors present a challenge to timely and accurate bit bounce detection. Therefore, this paper proposes a bit bounce detection method based on multi-feature graph and graph convolution networks. A multi-feature graph construction method using process variables, mean value, Mahalanobis distance, and Euclidean distance is proposed, and a two-layer graph convolutional network is designed to realize deep feature extraction and incident detection. The effectiveness and superiority of the proposed method are demonstrated by a real drilling industrial case.

To balance the inference speed and detection accuracy of a grasp detection algorithm, which are both important for robot grasping tasks, we propose an encoder–decoder structured pixel-level grasp detection neural network named the attention-based efficient robot grasp detection network (AE-GDN). Three spatial attention modules are introduced in the encoder stages to enhance the detailed information, and three channel attention modules are introduced in the decoder stages to extract more semantic information. Several lightweight and efficient DenseBlocks are used to connect the encoder and decoder paths to improve the feature modeling capability of AE-GDN. A high intersection over union (IoU) value between the predicted grasp rectangle and the ground truth does not necessarily mean a high-quality grasp configuration, but might cause a collision. This is because traditional IoU loss calculation methods treat the center part of the predicted rectangle as having the same importance as the area around the grippers. We design a new IoU loss calculation method based on an hourglass box matching mechanism, which will create good correspondence between high IoUs and high-quality grasp configurations. AEGDN achieves the accuracy of 98.9% and 96.6% on the Cornell and Jacquard datasets, respectively. The inference speed reaches 43.5 frames per second with only about 1.2 × 106 parameters. The proposed AE-GDN has also been deployed on a practical robotic arm grasping system and performs grasping well. Codes are available at https://github.com/robvincen/robot_gradet .

The tuning of microwave cavity filters (MCFs) is mostly conducted by experienced operators, and thus, is time and resource intensive. The automatic tuning method has received lots of research attention. However, the experience of tuning experts is not described systematically, and the existing methods are not adequate for MCFs with individual differences. Considering that fuzzy expert systems are superior in flexibility and interpretability, a dynamic-attention-based heuristic fuzzy expert system is proposed in this article to achieve automatic tuning with high applicability, accuracy, and efficiency. First, the dynamic tuning process is divided into several stages, and the evaluation function of filtering performance is designed for each stage. Second, an improved expert system with dynamic attention is constructed to facilitate the comprehensive evaluation of filtering performance and to balance tuning efficiency and accuracy. Third, a heuristic fuzzy logic system is presented to obtain appropriate tuning values, with low design complexity of fuzzy rules and high utilization of tuning process information. Finally, the effectiveness, applicability, and practicality of the proposed method are demonstrated by both simulations and experiments.

Alarm systems are critical for process safety and efficiency of complex industrial facilities. However, the presence of alarm floods severely compromises the performance of alarm systems. To cope with alarm floods, data mining has been applied to discover interesting patterns from historical alarm data, and such patterns can be used for alarm suppression, root cause analysis, and decision supports. However, most existing methods ignored the timestamps in pattern extraction or obtained complete patterns with significant redundancy. In this article, a new method is proposed to extract alarm flood patterns using a modified CloFAST algorithm. The contributions are twofold: first, a closed alarm sequence mining approach is proposed based on the CloFAST algorithm with improvements to incorporate timestamps and tolerate alarm order switchings; second, a pattern distillation strategy is designed to merge similar alarm sequences and export compact alarm sequential patterns. The proposed method is capable of avoiding influences of order ambiguities and also minimizing the redundancy of extracted patterns. The effectiveness of the proposed method is demonstrated by an industrial case study involving alarm data from a large-scale industrial facility.

Accurately and promptly detecting downhole abnormalities is of great importance to ensure safe and efficient operation of geological drilling systems. In view of the lack of abnormal data and the presence of multiple operating states, this article proposes a new abnormality detection method for geological drilling processes based on the Jensen-Shannon divergence and adaptive alarm limits. The major contributions are twofold: 1) a novel framework is proposed to detect downhole abnormalities by identifying normal operating conditions first and then establishing the normal operating zones; 2) an adaptive alarm limit design method is proposed to determine the normal operating zones by modeling the relation between the distribution feature and the variational trend feature based on the boundary points. The effectiveness and practicability of the proposed method are demonstrated by an industrial case study. The results demonstrate that the proposed method has superior performance to other methods in downhole abnormality detection.

The objective of this paper is to facilitate the improvement of the control and operation of industrial facilities, by providing decision support tools. More specifically, this paper has three main contributions. First, this paper presents the definition of operation metrics that provide insight into the behavior of: 1) annunciated alarms, 2) alarm floods, and 3) operator actions in industrial facilities. Second, this paper presents visualization plots named multilayered radar plots that can present information in an elegant, dense, and comprehensive fashion. Three types of plots are proposed, which collectively compare the behavior of metrics, variables, and operation times in industrial facilities. Third, this paper presents a ranking method and a reordering design procedure of displayed alarms during an alarm flood to reorder the alarms based on the proposed alarm-flood criticality index. The purpose is to provide additional assistance to operators to focus on more critical issues. As the operation metrics, visualization plots, and the ranking in alarm floods heavily utilize historized data and given the industrial-oriented application of these decision support tools, this paper also addresses the extraction of information and the integration of the tools into industrial automation platforms. Note to Practitioners-In a control system used for an industrial facility, a large amount of data is collected and historized. The data include sensor measurements, status of actuators, alarms, and operator actions, and it therefore contains valuable information. The information can be extracted and utilized to assist in the improvement of the control and operation of the industrial facility. The objective of this paper is to provide decision support tools by: 1) defining operation metrics that can characterize the information extracted from the historized data; 2) presenting visualization plots that allow for a clear presentation and comparison of the metrics, where three types of plots are proposed for different purposes of comparison; and 3) a ranking method and a reordering design procedure of displayed alarms during an alarm flood. Furthermore, this paper discusses how the information is extracted from the historized data and how to integrate the proposed decision support tools into existing industrial automation platforms.