Large-scale Industrial Systems Research Articles

Hierarchical fault propagation path recognition method based on knowledge-driven graph attention autoencoder with bilayer pooling for large-scale industrial system

Large-scale industrial system exhibits hierarchical structure, which consists of multiple subsystem blocks with various corresponding variables. However, existing fault propagation path recognition methods fail to fully leverage the relevant knowledge about industrial system with hierarchal structure, which makes the causality inference result inconsistent with reality, causing inaccurate fault propagation path recognition. In this work, a hierarchical fault propagation path recognition method based on knowledge-driven graph attention autoencoder with bilayer pooling for large-scale industrial system is proposed. First, a knowledge-driven graph attention autoencoder is developed for causality knowledge exploration on intra-subsystem level, which expressly represents the causalities between operation variables in each subsystem into a weighted directed graph. Then, a graph bilayer pooling model is designed to generate the subsystem causality graph by “stacking” the intra-subsystem causality graphs in a hierarchical fashion. Utilizing the knowledge-driven graph attention autoencoder with bilayer pooling, the critical hierarchical characteristics of system causality can be well described. Moreover, to reduce the smearing effect, a nonlinear probabilistic state estimation model with fault isolation is adopted for distributed fault detection. Eventually, the performance of the proposed method is verified on large-scale hierarchical industrial system air separation unit with three fault cases. The fault detection and fault root cause location results show that compared with other methods, the proposed method achieves high fault detection rate with low fault alarm rate, as well as better fault root cause location performance.

CPDet: Circle-Permutation-Aware Object Detection for Heat Exchanger Cleaning

Shell–tube heat exchangers are commonly used equipment in large-scale industrial systems of wastewater heat exchange to reclaim the thermal energy generated during industrial processes. However, the internal surfaces of the heat exchanger tubes often accumulate fouling, which subsequently reduces their heat transfer efficiency. Therefore, regular cleaning is essential. We aim to detect circle holes on the end surface of the heat exchange tubes to further achieve automated positioning and cleaning tubes. Notably, these holes exhibit a regular distribution. To this end, we propose a circle-permutation-aware object detector for heat exchanger cleaning to sufficiently exploit prior information of the original inputs. Specifically, the interval prior to the extraction module extracts interval information among circle holes based on prior statistics, yielding prior interval context. The following interval prior fusion module slices original images into circle domain and background domain maps according to the prior interval context. For the circle domain map, prior-guided sparse attention using prior a circle–hole diameter as the step divides the circle domain map into patches and performs patch-wise self-attention. The background domain map is multiplied by a hyperparameter weak coefficient matrix. In this way, our method fully leverages prior information to selectively weigh the original inputs to achieve more effective hole detection. In addition, to adapt the hole shape, we adopt the circle representation instead of the rectangle one. Extensive experiments demonstrate that our method achieves state-of-the-art performance and significantly boosts the YOLOv8 baseline by 5.24% mAP50 and 5.25% mAP50:95.

Toward Proactive Maintenance: A Multi-Tiered Architecture for Industrial Equipment Health Monitoring and Remaining Useful Life Prediction

This research paper introduces a comprehensive proactive maintenance architecture designed for large-scale industrial machinery systems. The proposed architectural framework integrates supervised and unsupervised machine learning business processes in order to enhance maintenance capabilities. The primary objective of this architecture is to enhance operational efficiency and reduce the occurrence of problems in industrial equipment. The collection of data on the state of industrial machinery is conducted through the utilization of sensors that are attached to it. The recommended framework offers modules that might potentially implement capabilities such as immediate anomaly detection, pre-failure status prediction, and assessment of remaining usable life. We offer a prototype implementation to verify the appropriateness of the proposed framework for testing purposes. The prototype utilizes a simulation framework, Cooja, to model a sensor network. The concept entails the collection of status data from industrial machinery by each sensor. The prototype utilizes a machine learning library for data streams, the MOA framework, to design and implement a business process for anomaly detection using unsupervised machine learning, as well as a business process for early machine fault prediction using supervised machine learning. In addition, deep learning libraries are employed to construct a business process that predicts the remaining operational lifespan of industrial machinery that is anticipated to experience failure. Furthermore, we examined the efficacy of the prototype’s integrated business protocols in this investigation. The proposed framework aligns effectively with software architectures designed to offer maintenance functionalities for industrial machinery, as indicated by our research findings.

Self-supervised cognitive learning for multifaced interest in large-scale industrial recommender systems

Recommender systems play a vital role in e-commerce. Existing algorithms usually learn ranking scores from a user's historical sequence, but they generally pay little attentions to mining the user's precise and varied interests to jointly optimize multiple tasks. In this study, we argue that it is crucial to formulate different user interests based on self-supervised learning (SSL) enhanced data representation for multi-scene objectives. We propose self-supervised cognitive learning (SSCL), a novel framework that can simultaneously model a user's multifaceted interest representation and linked multitask information learning. SSCL exploits SSL to enhance the value of training data. In addition, multi-interest-aggregation and multi-interest-select units are introduced and a customized gate-control model is used to optimize multiple objectives. We evaluate our framework using three real-world datasets, and our results demonstrate its effectiveness and superior performance over the state-of-the-art techniques. Compared with the baselines in three datasets, the click-through rate increased by +7.58%, and the conversion rate increased by +6.04%.

High-dimensional process monitoring under time-varying operating conditions via covariate-regulated principal component analysis

High-dimensional process monitoring, which aims to raise warnings for faulty events, is a fundamental tool in ensuring the safety of large-scale industrial systems. Over the last few decades, there has been significant interest in Principal Component Analysis (PCA)-based Statistical Process Control (SPC) charts. However, most of them rely on PCA’s optimal properties in homogeneous data, which often fails when dealing with heterogeneous data from industrial processes under time-varying operating conditions. To address this issue, this study proposes a Covariate-regulated PCA model (CrPCA) seamlessly integrated into a process monitoring framework. The main idea is to adjust the principal component directions smoothly according to the time-varying operating conditions, resulting in significantly less information loss in the principal component scores compared to existing methods. This approach further yields a low-dimensional feature vector that follows a distribution with fixed parameters, regardless of covariates, for real-time surveillance. We validate the performance of our methodology through a comprehensive numerical study and two real-world applications: wind turbine condition monitoring and hot rolling process monitoring.

Anomaly detection model for large-scale industrial systems using transfer entropy and graph attention network

Abstract Multivariate time series (MTS) anomaly detection is vital for ensuring the safety and reliability of large-scale industrial systems. However, existing deep learning methods often overlook complex interrelationships between different time series and the study of anomalies has been limited to detection. To address this, we propose an MTS anomaly detection model based on transfer entropy (TE) and graph attention network (GAT). In the graph construction module, by combining modified TE with automatic structure learning, we extract intricate relationships between features. In the prediction module, we modify the GAT to implement the dynamic attention mechanism and non-linear interaction between different features to improve the accuracy of model prediction. Finally, our model combines the modified TE with anomaly detection task, which can be used to provide interpretability for the detected anomalies using the constructed causal graph. Experimental results on both real and public datasets show that our approach outperforms the mainstream methods, in particular, achieving optimal results in terms of F1 scores and recall.

Comparison of mean-line methods for hydrogen turbo-expanders in hydrogen liquefiers

The energy consumption of a large-scale industrial hydrogen liquefaction system is primarily determined by the hydrogen turbo-expanders. To estimate the performance of the expander quickly, a one-dimensional preliminary design is conducted. This paper aims to identify dependable one-dimensional design codes for hydrogen turbo-expanders by employing three mean-line design approaches. The precision of the loss models is validated through hydrogen experiments, revealing a maximum deviation of less than 10 %. This study presents a summary of the optimal design parameter ranges for achieving high-efficiency hydrogen turbo-expanders. One common factor among the three methods is the characteristic ratio, which exhibits an optimal value range of 0.67–0.71. Furthermore, the three methods are applied to design hydrogen turbo-expanders of three different sizes, and the compositions of losses are thoroughly analyzed. The PR method exhibits the highest losses related to geometry sizes, while the VC method shows the greatest losses associated with velocities. The SS-CR method with moderate parameters is recommended for hydrogen turbo-expander design due to its superior maximum efficiency design limit and convenience. The two initial design parameters (specific speed and characteristic ratio) of this method play crucial roles in determining the scalability and efficiency of the hydrogen turbo-expander design.

Exergetic and exergoeconomic analyses of a large-scale industrial beer processing system

This research evaluated the performance of components and sections involved in industrial beer production using exergetic and exergoeconomics methodologies. The system was segmented into five production sections, and three energy input types were considered. The exergetic studies showed an operational exergetic efficiency of 3.33%, with an overall exergetic destruction rate of 5.54 MW and a specific destruction rate of 9.72 kW/hl for beer production. The overall improvement potential and sustainability index were estimated at 4.98 MW and 1.03, respectively. The brewhouse and packaging hall were identified as the sections with the highest production inefficiency, 58.73% and 30.40%, respectively. The exergoeconomic studies revealed a cost rate of 0.1704 USD/s for beer production, with the wort kettle, filling and cocking machine, Kieselguhr candle filter, whirlpool, and brite beer tank identified as the top five significant components in descending order. The efficiency of the system was critically affected by the activities in the packaging hall, particularly those involving energy inputs that cannot be recovered or attributed to the processed stream, beer. Further research is required to determine the cost savings of optimization measures identified from additional steam throttling, downsizing of some main pumps, and exergy loss during heating of wort and beer chilling processes.

Failure mode division and remaining useful life prognostics of multi-indicator systems with multi-fault

Modern large-scale industrial systems are complex in structure, and their health states are usually reflected by multiple indicators. The performance degradation of multiple indicators surpasses respective threshold, causing multiple system faults, and multiple failure modes such as redundancy, fusion, and competition among different fault combinations exist, introducing new challenges when predicting remaining useful life (RUL) of the system. This study aims to establish a unified multi-failure mode division framework and the RUL prediction model under different failure modes for multi-indicator systems. Firstly, combination relationship between different fault types caused by multi-indicators outweighing their threshold is analyzed, and different redundancy, fusion, and competition failure modes are defined. Next, the formal fault type definition and its remaining time before occurrence under different failure modes are provided. The distribution calculation model of remaining time for different fault types is derived. The corresponding system's RUL prediction model under multi-fault competition is established, and degradation modeling, parameter estimation, and RUL distribution calculation are performed using the state-space model. Eventually, the validity of the RUL prediction model according to multiple failure modes is verified by numerical experiments. Taking XJTU-SY bearing and C-MAPSS datasets as two examples, the applicability and feasibility of the given method are proved.

Industrial Process Control Using DPCA and Hierarchical Pareto Optimization

The control of large-scale industrial systems has several criteria, such as ensuring high productivity, low production costs and the lowest possible environmental impact. These criteria must be established for all subsystems of the large-scale system. This study is devoted to the development of a hierarchical control system that meets several of these criteria and allows for the separate optimization of each subsystem. Multicriteria optimization is based on the processing of data characterizing production processes, which makes it possible to organize a multidimensional statistical control process. Using neural networks to model the technological processes of subsystems and the method of dynamic principal component analysis (DPCA) to reduce the dimensionality of control problems allows us to find more efficient solutions. Using the example of a two-level hierarchy, we showed a variant of the connection between two subsystems by parameters.

Automatic decomposition of large-scale industrial processes for distributed MPC on the Shell–Yokogawa Platform for Advanced Control and Estimation (PACE)

The kernel of industrial advanced process control (APC) lies in the formulation and solution of model predictive control (MPC) problems, which specify the controller moves according to the solution of an optimal control problem at each sampling time. A significant challenge is the online computation for large-scale industrial systems. As the state-of-the-art APC technology, the Shell–Yokogawa Platform for Advanced Control and Estimation (PACE) has adopted a systematic framework of handling dynamic optimization of large-scale systems, where an automatic decomposition procedure generates subsystems for distributed MPC. The decomposition is implemented on network representations of the MPC models that capture interactions among process variables, with community detection used to maximize the statistical significance of the subnetworks with preferred internal interconnections. This paper introduces the fundamentals of such a decomposition approach and this functionality in PACE, followed by a case study on a crude distillation process to showcase its industrial application.

Understanding Security in Smart City Domains From the ANT-Centric Perspective

A city is a large human settlement that serves the people who live there, and a smart city is a concept of how cities might better serve their residents through new forms of technology. In this paper, we focus on four major smart city domains according to Maslow’s hierarchy of needs: smart utility, smart transportation, smart homes, and smart healthcare. Numerous IoT applications have been developed to achieve the intelligence that we desire in our smart domains, ranging from personal gadgets such as health trackers and smart watches to large-scale industrial IoT systems such as nuclear and energy management systems. However, many of the existing smart city IoT solutions can be made better by considering the suitability of their security strategies. Inappropriate system security designs generally occur in two scenarios: first, system designers recognize the importance of security but are unsure of where, when, or how to implement it; and second, system designers try to fit traditional security designs to meet the smart city security context. Thus, the objective of this paper is to provide application designers with the missing security link they may need in order to improve their security designs. By evaluating the specific context of each smart city domain and the context-specific security requirements, we aim to provide directions on when, where, and how they should implement security strategies and the possible security challenges they need to consider. In addition, we present a new perspective on security issues in smart cities from a data-centric viewpoint by referring to the reference architecture, the Activity-Network-Things (ANT)-centric architecture. This architecture is built upon the concept of “security in a zero-trust environment", to achieve end-to-end data security. By doing so, we reduce the security risks posed by new system interactions or unanticipated user behaviors while avoiding the hassle of regularly upgrading security models.

Detailed Process for Developing an Efficient Anomaly Detection Algorithm for Real-Time Streaming Data in Large-Scale Industrial Systems

Anomaly detection plays a critical role in large-scale industrial systems, where real-time streaming data is generated at high volumes. This research journal presents an in-depth study on developing an efficient anomaly detection algorithm specifically designed for such industrial systems. The goal is to identify abnormal patterns and deviations from normal behavior, enabling proactive maintenance, improved operational efficiency, and reduced downtime. The proposed algorithm leverages machine learning and stream processing techniques to handle the challenges associated with real- time streaming data analysis. The research covers algorithm design, implementation, evaluation, and performance analysis using large-scale datasets from industrial domains. Keywords: anomaly detection, real-time streaming data, large-scale industrial systems, machine learning, stream processing

An Operation Scheme Generation Method for Nuclear Power Plant Operation under the Condition of No Operating Procedures Guided

Large-scale, complex, and high-risk industrial systems such as nuclear power plants have developed detailed operating procedures. Under the expected conditions, the operators operate the system according to the operating procedures to avoid human error. However, under complex and unfamiliar conditions, once the guidance of operating procedures is lost, serious consequences may be caused. This paper proposes a No-Procedure guided Operation Supervision Technology (NoP-OST). The key idea is to identify the success paths by a forward-searching approach from source to sink functions. On this basis, the success paths are combined to generate various operation schemes to achieve the main objectives of the system. The paper illustrates the application of NoP-OST through the case of loss of the Residual Heat Removal System (RHRS) in a Pressurized Water Reactor (PWR) nuclear power plant. The research results show that the method proposed in this paper can provide a valuable operational reference for complex systems by making use of the available functions of the system.

Fast prediction of the performance of the centrifugal pump based on reduced-order model

In the present paper, the prediction of the performance of a centrifugal pump is investigated based on proper orthogonal decomposition (POD) reduced-order model (ROM). Firstly, the POD method is used to analyze the characteristics of the complex flow field inside the centrifugal pump, especially the eigenvalues of the velocity field and pressure field, and then the ROM of the flow field in the centrifugal pump is established then. After that, the established reduced-order model is used to predict the internal flow field. The results show that the POD mode field can reflect the spatial scale distribution of the flow field. The low-order flow field reflects the large-scale flow field structure, while the high-order flow field reflects the small-scale flow field structure. With the established ROM, the overall pump head (0D, 0-dimensional parameter) can be predicted with the relative error of less than 5%, and the standard root mean square error (normalized root mean square error, NRMSE) of pressure field and velocity field (3D, 3-dimensional parameter) is less than 4% and 8%, respectively. The efficiency and accuracy of both 0D and 3D parameters’ prediction motivate the fast prediction of multi-scale simulation of the large-scale industrial system in the future.

A Hybrid Evolutionary Algorithm Using Two Solution Representations for Hybrid Flow-Shop Scheduling Problem.

As an extension of the classical flow-shop scheduling problem, the hybrid flow-shop scheduling problem (HFSP) widely exists in large-scale industrial production systems and has been considered to be challenging for its complexity and flexibility. Evolutionary algorithms based on encoding and heuristic decoding approaches are shown effective in solving the HFSP. However, frequently used encoding and decoding strategies can only search a limited area of the solution space, thus leading to unsatisfactory performance during the later period. In this article, a hybrid evolutionary algorithm (HEA) using two solution representations is proposed to solve the HFSP for makespan minimization. First, the proposed HEA searches the solution space by a permutation-based encoding representation and two heuristic decoding methods to find some promising areas. Afterward, a Tabu search (TS) procedure based on a disjunctive graph representation is introduced to expand the searching space for further optimization. Two classical neighborhood structures focusing on critical paths are extended to the problem-specific backward schedules to generate candidate solutions for the TS. The proposed HEA is tested on three public HFSP benchmark sets from the existing literature, including 567 instances in total, and is compared with some state-of-the-art algorithms. Extensive experimental results indicate that the proposed HEA performs much better than the other algorithms. Moreover, the proposed method finds new best solutions for 285 hard instances.

Blockchain adoption and security management of large scale industrial renewable-based systems: Knowledge-based approach

Smart networks, e.g., smart grids (SGs) as complex interconnected systems have become more complex and require greater processing and communication capabilities due to the growing use of renewable energy (RE). In addition to being a reliable method for collecting and managing information, wireless networks (WNs) are capable of enhancing the performance of RE management. As a result of wireless data transfer risks and centralized power trades, SGs using WNs are unable to provide the security of the electricity market or ensure a high level of RE utilization. Blockchains (BCs), which are new technologies that enable data sharing, are gaining popularity, and it is thought to be able to address these issues. A BC-based, reliable power-exchange scheme is presented for SGs using WNs in this study. Using the BC for recording power data obtained from the WNs allows the smart contract for making appropriate trades according to the information. Energy trading and RE consumption are more efficient in a dual-chain architecture that consists of local energy trade BCs and RE trade BCs. A BC-based RE incentive framework is designed to have greater stability and scale among RE producers. Overall, the proposed technique not only can provide higher security and reliability to the system but also can mitigate fraud, as well as providing less operation cost for the system. Ultimately, the 56-bus test feeder layout is used to validate the framework.

Encrypted Decentralized Multi-Agent Optimization for Privacy Preservation in Cyber-Physical Systems

Decentralized optimizations have been extensively applied in large-scale industrial cyber-physical systems to achieve control scalability. However, state-of-the-art methods heavily depend on explicit communications between participants, exposing the entire control framework to data confidentiality risks. To overcome this challenge, in this article, a privacy-preserving decentralized multi-agent cooperative optimization paradigm was developed via integrating cryptography into decentralized optimization. The proposed approach can effectively protect participants’ privacy against external eavesdroppers, honest-but-curious agents, and the system operator. Theoretical security and correctness analyses are provided. Simulations of numerical examples and experiments on a real-world platform are given to demonstrate the security, accuracy, and applicability of the proposed method.

Reflection on the energy graph-based visualisation approach to FDI of large-scale industrial systems

Modern industrial systems are complex and large-scale, constituting multiple components and interconnections. These components interact through material and energy interfaces. Energy can be considered as a universal quality that allows system characterisation across different physical domains (thermal-fluid, mechanical, electrical and chemical). Therefore, an energy representation of a system can be used as a basis for fault detection and diagnosis. In order to cope with representing complex interconnections of systems, graphs can assist in presenting the structure and attributes of the system components. This paper reviews FDI applications using energy and graph-theoretical methods and then focuses on a unique approach that combines the strengths of both approaches. It furthermore specifically exploits the attributed graph as mathematical formalism for characterising an industrial process in terms of energy. The graph formalism allows capturing energy characteristics while retaining the structural information, i.e. linking the energy attribute to a physical location in the process. This approach is named the energy graph-based visualisation approach. The contribution of this paper lies in the portrayal of the full research endeavour for the purposes of comparison and future reference. Two applications are presented to illustrate the usefulness of the approach namely a gas-to-liquids process and a practical heated two-tank system.

МНОГОКРИТЕРИАЛЬНАЯ ОПТИМИЗАЦИЯ ТЕХНОЛОГИЧЕСКИХ ПРОЦЕССОВ НА ОСНОВЕ ДИНАМИЧЕСКОГО МГК И НЕЙРОСЕТЕВЫХ МОДЕЛЕЙ

The management of largescale industrial systems has several criteria, such as ensuring high productivity, low production costs and the least environmental impact. These criteria must be established for all subsystems of the largescale system. The study is devoted to the development of a hierarchical control system that meets several criteria and allows for separate optimization of each subsystem. Multicriteria optimization is based on the processing of data characterizing production processes, which made it possible to organize a multidimensional statistical control process. Using neural networks to model technological processes and DPCA to reduce the dimensionality of control problems allows us to find more efficient solutions. Using the example of a twolevel hierarchy, we showed a variant of communication by parameters between two subsystems.