Reliability Engineering & System Safety最新文献_第3页

Reinforcement learning based maintenance scheduling of flexible multi-machine manufacturing systems with varying interactive degradation

IF 9.4 1区工程技术

Reliability Engineering & System Safety Pub Date : 2025-03-22 DOI: 10.1016/j.ress.2025.111018

Jiangxi Chen, Xiaojun Zhou

{"title":"Reinforcement learning based maintenance scheduling of flexible multi-machine manufacturing systems with varying interactive degradation","authors":"Jiangxi Chen, Xiaojun Zhou","doi":"10.1016/j.ress.2025.111018","DOIUrl":"10.1016/j.ress.2025.111018","url":null,"abstract":"<div><div>In flexible multi-machine manufacturing systems, variations in product types dynamically influence machine loads, subsequently affecting the degradation processes of the machines. Moreover, the interactive degradation between the upstream and downstream machines, caused by the product quality deviations, changes with the different production routes for the variable product types. These factors, combined with the uncertain production schedules, present significant challenges for effective maintenance scheduling. To address these challenges, the maintenance scheduling problem is modeled as a Hidden-Mode Markov Decision Process (HM-MDP), where product types are treated as hidden modes that influence machine degradation and the subsequent maintenance decisions. The Interactive Degradation-Aware Proximal Policy Optimization (IDAPPO) reinforcement learning framework is introduced, enhancing the PPO algorithm with Graph Neural Networks (GNNs) to capture interactive degradation among machines and Long Short-Term Memory (LSTM) networks to handle temporal variations in production schedules. An entropy-based exploration strategy further manages the uncertainty of production schedules, enabling IDAPPO to adaptively optimize maintenance actions. Extensive experiments on both small-scale (5-machine) and large-scale (24-machine) systems demonstrate significantly reduced system losses and accelerated convergence of IDAPPO compared to the baseline approaches. These results indicate that IDAPPO provides a scalable and adaptive solution for improving the efficiency and reliability of complex manufacturing environments.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"260 ","pages":"Article 111018"},"PeriodicalIF":9.4,"publicationDate":"2025-03-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143683432","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A deep autoencoder with structured latent space for process monitoring and anomaly detection in coal-fired power units

IF 9.4 1区工程技术

Reliability Engineering & System Safety Pub Date : 2025-03-22 DOI: 10.1016/j.ress.2025.111060

Di Hu , Chen Zhang , Tao Yang , Qingyan Fang

{"title":"A deep autoencoder with structured latent space for process monitoring and anomaly detection in coal-fired power units","authors":"Di Hu , Chen Zhang , Tao Yang , Qingyan Fang","doi":"10.1016/j.ress.2025.111060","DOIUrl":"10.1016/j.ress.2025.111060","url":null,"abstract":"<div><div>In the big data era, deep autoencoder (DAE)-based methods for anomaly detection are widely used in monitoring coal-fired power units (CFPUs). However, these methods often overlook essential latent space information crucial for detecting anomalies within the DAE model. This study presents a structured latent space deep autoencoder (SLSDAE) that not only intuitively provides both latent space and reconstruction residual information for anomaly detection but also obviates the need for additional hyperparameters in the model's loss function. Furthermore, by leveraging the support vector data description (SVDD) model, this research extracts anomaly discrimination criteria from the SLSDAE model and introduces an end-to-end, real-time online monitoring framework for CFPUs. Comparative analysis on four public datasets demonstrates that the SLSDAE model enhances the G-mean in anomaly detection by 16.05 % over the DAE model and surpasses the performance of both the βVAE and DAGMM models. When applied to an actual induced draft fan, this framework effectively provides clear status trend tracking and early anomaly detection, up to 20 days in advance.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"261 ","pages":"Article 111060"},"PeriodicalIF":9.4,"publicationDate":"2025-03-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143739302","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

DG-Softmax: A new domain generalization intelligent fault diagnosis method for planetary gearboxes

IF 9.4 1区工程技术

Reliability Engineering & System Safety Pub Date : 2025-03-22 DOI: 10.1016/j.ress.2025.111057

Quan Qian , Qijun Wen , Rui Tang , Yi Qin

{"title":"DG-Softmax: A new domain generalization intelligent fault diagnosis method for planetary gearboxes","authors":"Quan Qian , Qijun Wen , Rui Tang , Yi Qin","doi":"10.1016/j.ress.2025.111057","DOIUrl":"10.1016/j.ress.2025.111057","url":null,"abstract":"<div><div>Many unsupervised domain adaptation models have been explored to tackle the fault transfer diagnosis issues. Nevertheless, their achievements completely rely on the availability of target domain samples during training. Unfortunately, these testing samples are usually unavailable in advance due to routine maintenance and long designed life. Towards the real-time diagnosis demands in actual engineering, this study proposes a decision margin-based domain generalization framework that can indirectly achieve the distribution alignment between source and unseen target domains. Based on the framework, a novel DG-Softmax loss considering the class-level decision margin is proposed to enhance the feature separability. A novel adaptive and anti-interference selection mechanism of class-level decision margin named ACADM mechanism is established to select the decision margin in DG-Softmax loss adaptively. Furthermore, the DG-Softmax model, which only includes a task-related loss without any other auxiliary loss terms, is established to improve the computational efficiency and the diagnosis precision. A two-stage training scheme is utilized, including pre-training and training phases. The proposed DG-Softmax is evaluated on two cross-bearing transfer tasks from laboratory bearing to actual wind-turbine bearing and six cross-speed transfer tasks of the system-level planetary gearbox, and the experimental results validate that it outperforms other typical methods. The related code can be downloaded from <span><span>https://qinyi-team.github.io/2025/03/DG-Softmax</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"260 ","pages":"Article 111057"},"PeriodicalIF":9.4,"publicationDate":"2025-03-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143703934","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

An unsupervised framework for dynamic health indicator construction and its application in rolling bearing prognostics

IF 9.4 1区工程技术

Reliability Engineering & System Safety Pub Date : 2025-03-22 DOI: 10.1016/j.ress.2025.111039

Tongda Sun , Chen Yin , Huailiang Zheng , Yining Dong

{"title":"An unsupervised framework for dynamic health indicator construction and its application in rolling bearing prognostics","authors":"Tongda Sun , Chen Yin , Huailiang Zheng , Yining Dong","doi":"10.1016/j.ress.2025.111039","DOIUrl":"10.1016/j.ress.2025.111039","url":null,"abstract":"<div><div>Health indicator (HI) plays a key role in degradation assessment and prognostics of rolling bearings. Although various HI construction methods have been investigated, most of them rely on expert knowledge for feature extraction and overlook capturing dynamic information hidden in sequential degradation processes, which limits the ability of the constructed HI for degradation trend representation and prognostics. To address these concerns, a novel dynamic HI that considers HI-level temporal dependence is constructed through an unsupervised framework. Specifically, a degradation feature learning module composed of a skip-connection-based autoencoder first maps raw signals to a representative degradation feature space (DFS) to automatically extract essential degradation features without the need for expert knowledge. Subsequently, in this DFS, a new HI-generating module embedded with an inner HI-prediction block is proposed for dynamic HI construction, where the temporal dependence between past and current HI states is guaranteed and modeled explicitly. On this basis, the dynamic HI captures the inherent dynamic contents of the degradation process, ensuring its effectiveness for degradation tendency modeling and future degradation prognostics. The experiment results on two bearing lifecycle datasets demonstrate that the proposed HI construction method outperforms comparison methods, and the constructed dynamic HI is superior for prognostic tasks.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"260 ","pages":"Article 111039"},"PeriodicalIF":9.4,"publicationDate":"2025-03-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143734648","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A full domain decision model for robust risk control based on minimum linkage space and copula Bayesian networks

IF 9.4 1区工程技术

Reliability Engineering & System Safety Pub Date : 2025-03-20 DOI: 10.1016/j.ress.2025.111046

Pei Zhang, Zhen-Ji Zhang, Da-Qing Gong

{"title":"A full domain decision model for robust risk control based on minimum linkage space and copula Bayesian networks","authors":"Pei Zhang, Zhen-Ji Zhang, Da-Qing Gong","doi":"10.1016/j.ress.2025.111046","DOIUrl":"10.1016/j.ress.2025.111046","url":null,"abstract":"<div><div>To effectively manage the complexity and risks inherent in rail transit operations, we propose a robust three-stage decision model. This model integrates a full-domain decision system, minimum linkage space, three-way clustering, and a Copula-Bayesian approach to create a comprehensive framework for data analysis and risk management. In the first stage, we establish a full-domain decision system that maps operational processes to specific risk characteristics, facilitating a unified approach to data interlinkages. The second stage combines minimum linkage space with a three-way clustering algorithm to identify the major risk factors from 25 potential risks, focusing on those crucial to system integrity. The final stage combines Copula theory and Bayesian networks to model and analyze in detail the dependencies and interrelationships among the 13 major risk factors identified. By utilizing advanced analytical tools, such as scatter plots, percentile spider charts, and correlation coefficients, we identify critical risk factors that significantly affect rail transit safety. This enables precise, predictive, and diagnostic interventions to enhance real-time risk assessments, ultimately reducing system risks and preventing accidents. The model provides actionable insights for managing complex risks in rail transit, offering a valuable tool for decision-makers to ensure safer operations.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"260 ","pages":"Article 111046"},"PeriodicalIF":9.4,"publicationDate":"2025-03-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143734651","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Kernel Reinforcement Learning for sampling-efficient risk management of large-scale engineering systems

IF 9.4 1区工程技术

Reliability Engineering & System Safety Pub Date : 2025-03-19 DOI: 10.1016/j.ress.2025.111022

Dingyang Zhang , Yiming Zhang , Pei Li , Shuyou Zhang

{"title":"Kernel Reinforcement Learning for sampling-efficient risk management of large-scale engineering systems","authors":"Dingyang Zhang , Yiming Zhang , Pei Li , Shuyou Zhang","doi":"10.1016/j.ress.2025.111022","DOIUrl":"10.1016/j.ress.2025.111022","url":null,"abstract":"<div><div>Mainstream methods for maintenance scheduling of multi-state systems (e.g. aircraft engines) often encounter challenges such as uncertainty accumulation, the need for extensive training data, and instability in the training process, particularly in life-cycle cost management. This paper introduces an innovative Kernel Reinforcement Learning (KRL) approach designed to enhance the reliability and safety of multi-state systems while significantly increasing decision-making efficiency. The policy and value functions are formulated non-parametrically to capture high-value episodes and datasets. KRL integrates probabilistic setups to imbue reinforcement learning with uncertainty, enhancing exploration of state–action spaces. Prior knowledge can be seamlessly integrated with the probabilistic framework to accelerate convergence. To address the memory issues associated with kernel methods when handling large datasets, the kernel matrix is dynamically updated with screened high-value datasets. Numerical evaluations on a k-out-of-n system, a coal mining transportation system, and an aircraft engine simulation demonstrate that the proposed KRL approach achieves faster convergence and reduced life-cycle costs compared to alternative methods. Specifically, KRL reduces the number of training episodes by 2–3 orders of magnitude, with a maximum cost reduction of 92%.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"260 ","pages":"Article 111022"},"PeriodicalIF":9.4,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143703933","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Virtual-reality-generated-data-driven Bayesian networks for risk analysis

IF 9.4 1区工程技术

Reliability Engineering & System Safety Pub Date : 2025-03-19 DOI: 10.1016/j.ress.2025.111053

Huixing Meng , Shijun Zhao , Wenjuan Song , Mengqian Hu

{"title":"Virtual-reality-generated-data-driven Bayesian networks for risk analysis","authors":"Huixing Meng , Shijun Zhao , Wenjuan Song , Mengqian Hu","doi":"10.1016/j.ress.2025.111053","DOIUrl":"10.1016/j.ress.2025.111053","url":null,"abstract":"<div><div>Risk analysis is crucial to the risk control of major accidents. Therefore, the risk analysis of complex systems has attracted increasing attention from academia and industry. Data-driven Bayesian network (BN) models have proved to be useful for risk analysis in complex systems. Nevertheless, insufficient data remains a challenge for risk analysis. In this paper, we propose a method of virtual reality (VR)-generated data aiming to provide a solution to generate data for risk analysis. To demonstrate the feasibility of VR-generated data applied to data-driven risk analysis, we proposed the following methodology on the example of an emergency response system for deepwater blowout (i.e., a spilled oil collection system). Firstly, a VR model of the spilled oil collection system is established. Secondly, required data is generated from the VR system for the risk analysis of emergency operations. Eventually, the data-driven BN for risk analysis is constructed based on VR-generated data. The results show that VR-generated data can support risk analysis in the presence of limited data.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"260 ","pages":"Article 111053"},"PeriodicalIF":9.4,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143714827","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Safety Integrity Level (SIL) evaluation of safety instrumented systems considering competing failure modes and subsystem priorities

IF 9.4 1区工程技术

Reliability Engineering & System Safety Pub Date : 2025-03-19 DOI: 10.1016/j.ress.2025.111025

Morteza Cheraghi, Sharareh Taghipour

{"title":"Safety Integrity Level (SIL) evaluation of safety instrumented systems considering competing failure modes and subsystem priorities","authors":"Morteza Cheraghi, Sharareh Taghipour","doi":"10.1016/j.ress.2025.111025","DOIUrl":"10.1016/j.ress.2025.111025","url":null,"abstract":"<div><div>Safety Integrity Level (SIL) is a crucial measure of the safety performance of Safety Instrumented Systems (SISs), reflecting their ability to reduce risk. However, SIL analysis has often overlooked the impact of competing failure modes and subsystem priorities within SISs. This paper introduces a novel probabilistic model for evaluating the SIL of safety functions that incorporates these critical aspects. The model calculates the time-dependent Probability of (dangerous) Failure on Demand (PFD) and Probability of Failing Safely (PFS) at the component, subsystem, and system levels. The average PFD (PFD<sub>avg</sub>) and SIL are calculated considering both planned and unplanned proof tests. The proposed model is validated through Monte Carlo simulations and applied to a safety system designed to protect a process vessel from high-pressure hazards. A comparative analysis with existing models demonstrates that competing failure modes and subsystem priorities significantly influence PFD, PFS, PFD<sub>avg</sub>, and consequently SIL, especially in systems with longer proof test intervals and higher Safe Failure Fractions (SFFs).</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"260 ","pages":"Article 111025"},"PeriodicalIF":9.4,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143703936","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Confidence-aware quantile Transformer for reliable degradation prediction of battery energy storage systems

IF 9.4 1区工程技术

Reliability Engineering & System Safety Pub Date : 2025-03-19 DOI: 10.1016/j.ress.2025.111019

Rui Wu , Jinpeng Tian , Jiachi Yao , Te Han , Chunsheng Hu

{"title":"Confidence-aware quantile Transformer for reliable degradation prediction of battery energy storage systems","authors":"Rui Wu , Jinpeng Tian , Jiachi Yao , Te Han , Chunsheng Hu","doi":"10.1016/j.ress.2025.111019","DOIUrl":"10.1016/j.ress.2025.111019","url":null,"abstract":"<div><div>Battery energy storage systems (BESS) play a vital role in grid stabilization, integrating renewable energy, and enhancing resilience through efficient energy storage and distribution. Precisely predicting the BESS degradation status is paramount for timely maintenance, ensuring safety, and upholding reliability. The degradation process of batteries in BESS involves complex chemical reactions and physical changes, compounded by various uncertain factors such as diverse battery usage conditions. To address this challenge, the quantile Transformer (Q-Transformer) method is proposed, which can predict the degradation of batteries within intervals, thereby enhancing the reliability of predictions. Firstly, the voltage and current data of lithium-ion batteries in BESS are converted into capacity increment features through incremental capacity analysis. Subsequently, the constructed Q-Transformer model is trained using these capacity increment features. Finally, the trained Q-Transformer model is employed to predict the capacity of lithium-ion batteries in BESS. The effectiveness of the proposed Q-Transformer method is validated and analyzed using two lithium-ion battery datasets, NASA and CALCE. The experimental results indicate that the proposed Q-Transformer method exhibits superior predictive performance than other popular methods. The errors in terms of RMSE, MAPE, and MD-MAPE are mostly within about 5%. The proposed Q-Transformer method shows promising potential for extensive application in BESS.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"260 ","pages":"Article 111019"},"PeriodicalIF":9.4,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143683420","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Semi-supervised ISA: A novel industrial knowledge graph construction method enhanced by the fault log corpus analysis and semi-supervised learning

IF 9.4 1区工程技术

Reliability Engineering & System Safety Pub Date : 2025-03-19 DOI: 10.1016/j.ress.2025.111021

Jiamin Xu, Siwen Mo, Zixuan Xu, Zhiwen Chen, Chao Yang, Zhaohui Jiang

{"title":"Semi-supervised ISA: A novel industrial knowledge graph construction method enhanced by the fault log corpus analysis and semi-supervised learning","authors":"Jiamin Xu, Siwen Mo, Zixuan Xu, Zhiwen Chen, Chao Yang, Zhaohui Jiang","doi":"10.1016/j.ress.2025.111021","DOIUrl":"10.1016/j.ress.2025.111021","url":null,"abstract":"<div><div>In industrial systems, knowledge graph-based intelligent fault diagnosis methods utilize extensive textual information, such as accumulated fault logs, to effectively construct domain-specific knowledge graphs. These graphs facilitate the use of unstructured data, thereby enhancing both diagnostic efficiency and accuracy. However, much of the existing research applies general knowledge graph construction methods to industrial fault diagnosis, without adapting them to the specific characteristics of fault logs. This oversight poses challenges in ensuring adequate and accurate model training. To address these challenges, this paper offers a comprehensive analysis of the essential attributes of fault logs, and proposes a semi-supervised industrial-adaptive knowledge graph construction method. The method employs a BiLSTM-BIO-based named entity recognition model, followed by a testing-enhanced self-attention relation extraction model designed for semi-supervised learning patterns. The extracted entities and relationships are organized into triplets to construct the knowledge graph. Finally, the proposed method is evaluated using fault logs from a specific heavy-duty train model. Extensive comparisons with various existing knowledge graph construction methods demonstrate the superior performance of the proposed method.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"260 ","pages":"Article 111021"},"PeriodicalIF":9.4,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143683433","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0