Reliability Engineering & System Safety最新文献_第8页

Study on unsupervised gas outburst hazard early warning method based on spatiotemporal graph convolution network 基于时空图卷积网络的无监督瓦斯突出危险性预警方法研究

IF 11 1区工程技术 Q1 ENGINEERING, INDUSTRIAL

Reliability Engineering & System Safety

Pub Date : 2026-01-10 DOI: 10.1016/j.ress.2026.112218

Binglong Liu , Zhonghui Li , Chaolin Zhang , Shan Yin

Coal and gas outbursts are jointly influenced by geological factors, gas occurrence, and the physical properties of coal. Therefore, analyzing the spatiotemporal data features during tunneling and constructing a spatiotemporal data fusion model are essential for addressing such disasters. This study proposes, for the first time, the application of a deep learning, based BN-spatiotemporal graph convolution model for gas outburst early warning. First, the risk features of spatiotemporal data were extracted, and effective features were obtained through correlation analysis. Second, an STGAT module was constructed to learn the coupling relationships among spatiotemporal data and to obtain anomaly scores. Finally, a BN layer was employed to construct a lagged graph structure of spatiotemporal data, thereby realizing early warning of coal and gas outbursts. The model was tested using on-site spatiotemporal data, and the results showed that the distribution of unsupervised anomaly scores followed a normal distribution. The model was tested using on-site spatiotemporal data; the risk thresholds were determined based on the observed anomaly rate and risk percentiles, while a normality check of the unsupervised anomaly scores was conducted as a secondary statistical verification.

The final classification results achieved accuracies of 96.7% and 80% for the hazard and high risk, respectively. This methodology provides a conceptual framework for fusing multi-source heterogeneous spatiotemporal data within the coal mining sector, offers a novel approach for gas outburst warning, and significantly enhances the safety and operational efficiency of mines.

煤与瓦斯突出受地质因素、瓦斯赋存状态和煤的物性共同影响。因此，分析隧道掘进过程中的时空数据特征，构建时空数据融合模型是解决此类灾害的关键。本研究首次提出了基于深度学习的bn -时空图卷积模型在瓦斯突出预警中的应用。首先提取时空数据的风险特征，通过相关分析得到有效特征；其次，构建STGAT模块，学习时空数据间的耦合关系，获取异常评分；最后，利用BN层构建时空数据的滞后图结构，实现煤与瓦斯突出的预警。利用现场时空数据对模型进行检验，结果表明，无监督异常分数的分布服从正态分布。利用现场时空数据对模型进行了检验；根据观察到的异常率和风险百分位数确定风险阈值，并对无监督异常评分进行正态性检查，作为二次统计验证。最终的分类结果对危害和高风险的准确率分别达到96.7%和80%。该方法为煤矿领域多源异构时空数据融合提供了一个概念框架，为瓦斯突出预警提供了一种新的方法，显著提高了矿山的安全性和运行效率。

{"title":"Study on unsupervised gas outburst hazard early warning method based on spatiotemporal graph convolution network","authors":"Binglong Liu , Zhonghui Li , Chaolin Zhang , Shan Yin","doi":"10.1016/j.ress.2026.112218","DOIUrl":"10.1016/j.ress.2026.112218","url":null,"abstract":"<div><div>Coal and gas outbursts are jointly influenced by geological factors, gas occurrence, and the physical properties of coal. Therefore, analyzing the spatiotemporal data features during tunneling and constructing a spatiotemporal data fusion model are essential for addressing such disasters. This study proposes, for the first time, the application of a deep learning, based BN-spatiotemporal graph convolution model for gas outburst early warning. First, the risk features of spatiotemporal data were extracted, and effective features were obtained through correlation analysis. Second, an STGAT module was constructed to learn the coupling relationships among spatiotemporal data and to obtain anomaly scores. Finally, a BN layer was employed to construct a lagged graph structure of spatiotemporal data, thereby realizing early warning of coal and gas outbursts. The model was tested using on-site spatiotemporal data, and the results showed that the distribution of unsupervised anomaly scores followed a normal distribution. The model was tested using on-site spatiotemporal data; the risk thresholds were determined based on the observed anomaly rate and risk percentiles, while a normality check of the unsupervised anomaly scores was conducted as a secondary statistical verification.</div><div>The final classification results achieved accuracies of 96.7% and 80% for the hazard and high risk, respectively. This methodology provides a conceptual framework for fusing multi-source heterogeneous spatiotemporal data within the coal mining sector, offers a novel approach for gas outburst warning, and significantly enhances the safety and operational efficiency of mines.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"270 ","pages":"Article 112218"},"PeriodicalIF":11.0,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145978779","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

Data-driven risk analysis and management framework for rail hazmat transportation in Canada: Machine learning approach 加拿大铁路危险品运输的数据驱动风险分析和管理框架：机器学习方法

IF 11 1区工程技术 Q1 ENGINEERING, INDUSTRIAL

Reliability Engineering & System Safety

Pub Date : 2026-01-10 DOI: 10.1016/j.ress.2026.112225

Marjan Izadpanah, Ali Vaezi, Ali Asgary, Anteneh Ayanso, Martin Kusy

Rail transportation of hazardous materials (hazmat) is essential to Canada’s economy but carries significant safety and environmental risks. This study develops a data-driven predictive risk assessment framework for hazmat release following railway accidents, integrating multi-year incident records with operational, environmental, and geographic variables from multiple public sources. Supervised machine learning models—logistic regression, decision trees, and neural networks—are applied to classify hazmat release outcomes following rail incidents. Key predictors include track type, hazmat class, weather conditions, train configuration, and operational density. The best-performing models demonstrated competitive predictive performance, with metrics such as AUC-ROC, F1-score, and balanced accuracy indicating consistent behavior despite substantial class imbalance, while also offering interpretable insights into key risk factors. We also developed the Rail HAZMAT Release Predictor, a web-based tool that applies our models to assess hazmat release risks in rail incidents. Findings inform targeted mitigation strategies aligned with Public Safety Canada’s emergency management framework, including limiting hazmat car counts, implementing predictive maintenance, and tailoring emergency protocols to regional risk profiles. By combining multi-source data integration with advanced modeling, this research advances proactive, evidence-based decision-making for safer hazmat rail operations.

危险物质（危险品）的铁路运输对加拿大经济至关重要，但也存在重大的安全和环境风险。本研究开发了一个数据驱动的预测风险评估框架，用于铁路事故后的危险物质释放，将多年的事故记录与来自多个公共来源的运营、环境和地理变量相结合。监督机器学习模型——逻辑回归、决策树和神经网络——被应用于对铁路事故后的危险物质释放结果进行分类。关键的预测因素包括轨道类型、危险品类别、天气条件、列车配置和操作密度。表现最好的模型展示了具有竞争力的预测性能，诸如AUC-ROC、F1-score和平衡精度等指标表明，尽管存在严重的类别不平衡，但行为一致，同时还提供了对关键风险因素的可解释见解。我们还开发了铁路危险品释放预测器，这是一个基于网络的工具，可以应用我们的模型来评估铁路事故中的危险品释放风险。调查结果为符合加拿大公共安全部应急管理框架的有针对性的缓解战略提供了信息，包括限制危险品车数量、实施预测性维护和根据区域风险概况定制应急协议。通过将多源数据集成与先进的建模相结合，本研究为更安全的危险品铁路运营提供了主动的、基于证据的决策。

{"title":"Data-driven risk analysis and management framework for rail hazmat transportation in Canada: Machine learning approach","authors":"Marjan Izadpanah, Ali Vaezi, Ali Asgary, Anteneh Ayanso, Martin Kusy","doi":"10.1016/j.ress.2026.112225","DOIUrl":"10.1016/j.ress.2026.112225","url":null,"abstract":"<div><div>Rail transportation of hazardous materials (hazmat) is essential to Canada’s economy but carries significant safety and environmental risks. This study develops a data-driven predictive risk assessment framework for hazmat release following railway accidents, integrating multi-year incident records with operational, environmental, and geographic variables from multiple public sources. Supervised machine learning models—logistic regression, decision trees, and neural networks—are applied to classify hazmat release outcomes following rail incidents. Key predictors include track type, hazmat class, weather conditions, train configuration, and operational density. The best-performing models demonstrated competitive predictive performance, with metrics such as AUC-ROC, F1-score, and balanced accuracy indicating consistent behavior despite substantial class imbalance, while also offering interpretable insights into key risk factors. We also developed the <em>Rail HAZMAT Release Predictor</em>, a web-based tool that applies our models to assess hazmat release risks in rail incidents. Findings inform targeted mitigation strategies aligned with Public Safety Canada’s emergency management framework, including limiting hazmat car counts, implementing predictive maintenance, and tailoring emergency protocols to regional risk profiles. By combining multi-source data integration with advanced modeling, this research advances proactive, evidence-based decision-making for safer hazmat rail operations.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"271 ","pages":"Article 112225"},"PeriodicalIF":11.0,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145981142","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

Ontology - and data-driven defect diagnosis with knowledge graphs and causal reasoning: Application to the risk management of gravity dams 基于知识图和因果推理的本体和数据驱动的缺陷诊断：在重力坝风险管理中的应用

IF 11 1区工程技术 Q1 ENGINEERING, INDUSTRIAL

Reliability Engineering & System Safety

Pub Date : 2026-01-10 DOI: 10.1016/j.ress.2026.112224

Liang Pei , Yuanpeng Li , Xiang Lu , Jingren Zhou , Yanling Li

Gravity dams in complex environments, as safety-critical infrastructure, face multi-source defect evolution that challenges reliability and risk management. Traditional diagnostic approaches rely heavily on expert judgment, which limits efficiency and leads to incomplete defect coverage. Therefore, this study develops a reliability-oriented defect diagnosis framework that integrates ontology construction, knowledge graph representation, and causal reasoning. A domain-specific ontology is established to formalize defect concepts, causal relations, and monitoring indicators. Then, a defective knowledge graph is constructed from unstructured inspection reports and monitoring data using large language models. Furthermore, a path attenuation coefficient-driven propagation mechanism and a bidirectional causal reasoning framework are introduced, enabling both forward inference of defect evolution and backward tracing of potential causes. Application to a gravity dam demonstrates that the method can identify critical defect chains, such as “Partial curtain damage → Abnormal uplift pressure → Local dam cracking or damage → Structural failure”, while also tracing hidden risks, such as “Long-term reservoir erosion → Downstream face calcareous segregation”, enhancing the accuracy and reliability of dam structural defect diagnosis significantly. Beyond dam engineering, the methodology applies to other large-scale safety systems, providing a generalizable decision support tool for reliability engineering and system safety.

复杂环境下的重力坝作为安全关键型基础设施，面临着多源缺陷演变的挑战，对其可靠性和风险管理提出了挑战。传统的诊断方法严重依赖于专家判断，这限制了效率并导致缺陷覆盖不完整。因此，本研究开发了一个面向可靠性的缺陷诊断框架，该框架集本体构建、知识图表示和因果推理于一体。建立一个领域特定的本体来形式化缺陷概念、因果关系和监视指示器。然后，利用大型语言模型，从非结构化检测报告和监测数据中构建缺陷知识图。引入了路径衰减系数驱动的传播机制和双向因果推理框架，实现了缺陷演化的正向推理和潜在原因的反向追踪。通过对某重力坝的实际应用表明，该方法能够识别出“部分帷幕损伤→异常隆升压力→局部坝体开裂或损伤→结构破坏”等关键缺陷链，并对“长期库区侵蚀→下游面钙质偏析”等隐患进行追踪，显著提高了大坝结构缺陷诊断的准确性和可靠性。除大坝工程外，该方法还适用于其他大型安全系统，为可靠性工程和系统安全提供了一种通用的决策支持工具。

{"title":"Ontology - and data-driven defect diagnosis with knowledge graphs and causal reasoning: Application to the risk management of gravity dams","authors":"Liang Pei , Yuanpeng Li , Xiang Lu , Jingren Zhou , Yanling Li","doi":"10.1016/j.ress.2026.112224","DOIUrl":"10.1016/j.ress.2026.112224","url":null,"abstract":"<div><div>Gravity dams in complex environments, as safety-critical infrastructure, face multi-source defect evolution that challenges reliability and risk management. Traditional diagnostic approaches rely heavily on expert judgment, which limits efficiency and leads to incomplete defect coverage. Therefore, this study develops a reliability-oriented defect diagnosis framework that integrates ontology construction, knowledge graph representation, and causal reasoning. A domain-specific ontology is established to formalize defect concepts, causal relations, and monitoring indicators. Then, a defective knowledge graph is constructed from unstructured inspection reports and monitoring data using large language models. Furthermore, a path attenuation coefficient-driven propagation mechanism and a bidirectional causal reasoning framework are introduced, enabling both forward inference of defect evolution and backward tracing of potential causes. Application to a gravity dam demonstrates that the method can identify critical defect chains, such as “Partial curtain damage → Abnormal uplift pressure → Local dam cracking or damage → Structural failure”, while also tracing hidden risks, such as “Long-term reservoir erosion → Downstream face calcareous segregation”, enhancing the accuracy and reliability of dam structural defect diagnosis significantly. Beyond dam engineering, the methodology applies to other large-scale safety systems, providing a generalizable decision support tool for reliability engineering and system safety.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"270 ","pages":"Article 112224"},"PeriodicalIF":11.0,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145978783","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 reinforcement learning approach for robust dynamic Bayesian network-based systemic risk analysis in freight forwarding 基于鲁棒动态贝叶斯网络的货运代理系统风险分析的深度强化学习方法

IF 11 1区工程技术 Q1 ENGINEERING, INDUSTRIAL

Reliability Engineering & System Safety

Pub Date : 2026-01-10 DOI: 10.1016/j.ress.2026.112190

Lu Wang , Yunfeng Wang , Na Li

The increasing volatility of global supply chains demands a shift toward robust, worst-case risk assessment. This requirement is particularly critical for the freight forwarding industry, which acts as a central intermediary in global trade. As asset-light coordinators, freight forwarders are exposed to systemic risks originating from their partners. For instance, a minor documentation error can escalate into substantial demurrage fees, and the reliability of a new carrier on a volatile trade lane often presents significant uncertainty. Conventional static risk models are inadequate for addressing these dynamic and interconnected challenges. This paper addresses a critical research question: How can asset-light freight forwarders conduct robust, worst-case risk assessments within such a dynamic, uncertain, and interconnected environment? To address this question, we propose a tailored Robust Dynamic Bayesian Network (R-DBN) framework. Our approach incorporates the Noisy-MAX model to facilitate parameterization in this data-scarce sector. We then employ a Deep Reinforcement Learning (DRL) algorithm to solve the resulting high-dimensional, non-convex optimization problem. Our computational results demonstrate two primary contributions. First, the trained DRL agent identifies critical worst-case scenarios and learns a reusable policy. This policy enables near-instantaneous risk assessments for new problem instances, offering a significant operational advantage. Second, sensitivity analysis and a case study provide a key managerial insight: a forwarder’s internal operational resilience is the most critical factor in mitigating systemic risk, substantially outweighing the direct impact of external disruptions.

全球供应链的波动性越来越大，需要转向稳健的最坏情况风险评估。这一要求对于作为全球贸易中心中介的货运代理行业尤为重要。作为轻资产协调者，货运代理面临来自合作伙伴的系统性风险。例如，一个小小的文件错误可能会升级为大量的滞期费，而在不稳定的贸易航线上，新承运人的可靠性往往会带来很大的不确定性。传统的静态风险模型不足以应对这些动态的、相互关联的挑战。本文解决了一个关键的研究问题：轻资产货运代理如何在这样一个动态的、不确定的、相互关联的环境中进行稳健的、最坏情况的风险评估？为了解决这个问题，我们提出了一个定制的鲁棒动态贝叶斯网络（R-DBN）框架。我们的方法结合了noise - max模型，以促进数据稀缺领域的参数化。然后，我们采用深度强化学习（DRL）算法来解决由此产生的高维、非凸优化问题。我们的计算结果证明了两个主要贡献。首先，经过训练的DRL代理识别关键的最坏情况并学习可重用策略。该策略支持对新问题实例进行近乎即时的风险评估，提供了显著的操作优势。其次，敏感性分析和案例研究提供了一个关键的管理见解：货代的内部运营弹性是减轻系统性风险的最关键因素，大大超过了外部中断的直接影响。

{"title":"A deep reinforcement learning approach for robust dynamic Bayesian network-based systemic risk analysis in freight forwarding","authors":"Lu Wang , Yunfeng Wang , Na Li","doi":"10.1016/j.ress.2026.112190","DOIUrl":"10.1016/j.ress.2026.112190","url":null,"abstract":"<div><div>The increasing volatility of global supply chains demands a shift toward robust, worst-case risk assessment. This requirement is particularly critical for the freight forwarding industry, which acts as a central intermediary in global trade. As asset-light coordinators, freight forwarders are exposed to systemic risks originating from their partners. For instance, a minor documentation error can escalate into substantial demurrage fees, and the reliability of a new carrier on a volatile trade lane often presents significant uncertainty. Conventional static risk models are inadequate for addressing these dynamic and interconnected challenges. This paper addresses a critical research question: How can asset-light freight forwarders conduct robust, worst-case risk assessments within such a dynamic, uncertain, and interconnected environment? To address this question, we propose a tailored Robust Dynamic Bayesian Network (R-DBN) framework. Our approach incorporates the Noisy-MAX model to facilitate parameterization in this data-scarce sector. We then employ a Deep Reinforcement Learning (DRL) algorithm to solve the resulting high-dimensional, non-convex optimization problem. Our computational results demonstrate two primary contributions. First, the trained DRL agent identifies critical worst-case scenarios and learns a reusable policy. This policy enables near-instantaneous risk assessments for new problem instances, offering a significant operational advantage. Second, sensitivity analysis and a case study provide a key managerial insight: a forwarder’s internal operational resilience is the most critical factor in mitigating systemic risk, substantially outweighing the direct impact of external disruptions.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"271 ","pages":"Article 112190"},"PeriodicalIF":11.0,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145981148","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

Dynamic opportunistic maintenance optimization for multi-state systems with overlapping maintenance time and duration-dependent costs 具有重叠维护时间和持续时间相关成本的多状态系统的动态机会维护优化

IF 11 1区工程技术 Q1 ENGINEERING, INDUSTRIAL

Reliability Engineering & System Safety

Pub Date : 2026-01-10 DOI: 10.1016/j.ress.2026.112198

Yuanming Song , Yan-Fu Li , Chen Zhang , Rui Peng , Cristiano A.V. Cavalcante

Opportunistic maintenance is a cost-effective strategy for reducing maintenance costs and improving system performance, particularly in multi-state systems. However, most existing models focus on simplistic maintenance triggers based on time or condition thresholds, neglecting dynamic system evaluation and economic dependencies beyond shared setup costs, such as overlapping maintenance durations and production losses. This paper proposes a dynamic optimization approach for opportunistic maintenance in multi-state systems, incorporating multiple maintenance phases and overlapping maintenance times. The approach is modeled as an infinite-horizon Markov decision process with a hybrid discrete-continuous state and action space. Maintenance decisions are made by integrating corrective replacement, preventive maintenance, and opportunistic strategies. We solve the optimization problem using a multi-task proximal policy optimization framework. Validation on a realistic computing network hosting large language models (LLMs) demonstrates that our approach effectively balances maintenance costs and system performance, offering innovative solutions for dynamic maintenance optimization in practical engineering applications.

机会维护是降低维护成本和提高系统性能的一种经济有效的策略，特别是在多状态系统中。然而，大多数现有模型关注的是基于时间或条件阈值的简单维护触发器，忽略了共享设置成本之外的动态系统评估和经济依赖性，例如重叠的维护持续时间和生产损失。提出了一种多状态系统的机会维修动态优化方法，该方法考虑了多维修阶段和重叠维修时间。该方法被建模为具有离散-连续混合状态和动作空间的无限视界马尔可夫决策过程。维护决策是通过整合纠正性更换、预防性维护和机会策略来做出的。我们使用一个多任务近端策略优化框架来解决优化问题。在承载大型语言模型（llm）的现实计算网络上的验证表明，我们的方法有效地平衡了维护成本和系统性能，为实际工程应用中的动态维护优化提供了创新的解决方案。

{"title":"Dynamic opportunistic maintenance optimization for multi-state systems with overlapping maintenance time and duration-dependent costs","authors":"Yuanming Song , Yan-Fu Li , Chen Zhang , Rui Peng , Cristiano A.V. Cavalcante","doi":"10.1016/j.ress.2026.112198","DOIUrl":"10.1016/j.ress.2026.112198","url":null,"abstract":"<div><div>Opportunistic maintenance is a cost-effective strategy for reducing maintenance costs and improving system performance, particularly in multi-state systems. However, most existing models focus on simplistic maintenance triggers based on time or condition thresholds, neglecting dynamic system evaluation and economic dependencies beyond shared setup costs, such as overlapping maintenance durations and production losses. This paper proposes a dynamic optimization approach for opportunistic maintenance in multi-state systems, incorporating multiple maintenance phases and overlapping maintenance times. The approach is modeled as an infinite-horizon Markov decision process with a hybrid discrete-continuous state and action space. Maintenance decisions are made by integrating corrective replacement, preventive maintenance, and opportunistic strategies. We solve the optimization problem using a multi-task proximal policy optimization framework. Validation on a realistic computing network hosting large language models (LLMs) demonstrates that our approach effectively balances maintenance costs and system performance, offering innovative solutions for dynamic maintenance optimization in practical engineering applications.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"270 ","pages":"Article 112198"},"PeriodicalIF":11.0,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145978782","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 reliability-oriented conflict robust optimization model for berth allocation under operational disturbances 运行扰动下面向可靠性的冲突鲁棒泊位分配优化模型

IF 11 1区工程技术 Q1 ENGINEERING, INDUSTRIAL

Reliability Engineering & System Safety

Pub Date : 2026-01-10 DOI: 10.1016/j.ress.2026.112217

Xi Xiang , Lina Yu , Xin Liu , Changchun Liu

The berth allocation problem concerns assigning service times and quay spaces to vessels at container terminals. Ensuring reliable and safe port operations under uncertain disturbances has become increasingly critical. This study addresses the berth allocation problem considering unpredictable vessel arrivals. Possible scenarios are represented by uncertainty sets rather than probability distributions, which are often unavailable before decision making. A conflict-based reliability measure is introduced to quantify schedule robustness and recoverability. A two-stage conflict robust optimization model is proposed, where the first stage determines baseline berthing times and positions, and the second stage minimizes conflicts that represent potential operational failures. Two model extensions, namely the expanded variant and the risk-constrained variant, explicitly incorporate reliability trade-offs between conservatism, cost, and safety. An exact iterative algorithm is developed to solve the models efficiently. Computational experiments show that the proposed algorithm outperforms several benchmark methods. The results indicate that a small increase in operational cost can substantially enhance schedule reliability and system resilience, while the expanded and risk-constrained models provide flexible options for balancing cost, reliability, and worst-case performance.

泊位分配问题涉及到在集装箱码头为船舶分配服务时间和码头空间。确保在不确定的干扰下港口运行的可靠和安全已变得越来越重要。本文研究了船舶到港不可预测情况下的泊位分配问题。可能的情景是用不确定性集合而不是概率分布来表示的，而概率分布在决策之前通常是不可用的。引入了一种基于冲突的可靠性度量来量化调度鲁棒性和可恢复性。提出了一种两阶段冲突鲁棒优化模型，其中第一阶段确定基线靠泊时间和靠泊位置，第二阶段最小化代表潜在操作故障的冲突。两个模型扩展，即扩展的变体和风险约束的变体，明确地将可靠性在保守性、成本和安全性之间进行权衡。提出了一种精确的迭代算法来求解这些模型。计算实验表明，该算法优于几种基准方法。结果表明，运行成本的小幅增加可以显著提高调度可靠性和系统弹性，而扩展和风险约束模型为平衡成本、可靠性和最坏情况性能提供了灵活的选择。

{"title":"A reliability-oriented conflict robust optimization model for berth allocation under operational disturbances","authors":"Xi Xiang , Lina Yu , Xin Liu , Changchun Liu","doi":"10.1016/j.ress.2026.112217","DOIUrl":"10.1016/j.ress.2026.112217","url":null,"abstract":"<div><div>The berth allocation problem concerns assigning service times and quay spaces to vessels at container terminals. Ensuring reliable and safe port operations under uncertain disturbances has become increasingly critical. This study addresses the berth allocation problem considering unpredictable vessel arrivals. Possible scenarios are represented by uncertainty sets rather than probability distributions, which are often unavailable before decision making. A conflict-based reliability measure is introduced to quantify schedule robustness and recoverability. A two-stage conflict robust optimization model is proposed, where the first stage determines baseline berthing times and positions, and the second stage minimizes conflicts that represent potential operational failures. Two model extensions, namely the expanded variant and the risk-constrained variant, explicitly incorporate reliability trade-offs between conservatism, cost, and safety. An exact iterative algorithm is developed to solve the models efficiently. Computational experiments show that the proposed algorithm outperforms several benchmark methods. The results indicate that a small increase in operational cost can substantially enhance schedule reliability and system resilience, while the expanded and risk-constrained models provide flexible options for balancing cost, reliability, and worst-case performance.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"271 ","pages":"Article 112217"},"PeriodicalIF":11.0,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145981147","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 hybrid modeling framework for predicting the multilayer ceramic capacitor reliability under thermal-electrical coupling operating conditions 热电耦合工况下多层陶瓷电容器可靠性预测的混合建模框架

IF 11 1区工程技术 Q1 ENGINEERING, INDUSTRIAL

Reliability Engineering & System Safety

Pub Date : 2026-01-10 DOI: 10.1016/j.ress.2026.112211

Donghui Li , Yuguang Zhong , Xue Zhou , Guofu Zhai , Rui Kang

Physics-of-failure (POF) methods based on testing are largely inadequate in meeting the reliability prediction demands of large-scale production. In contrast, pure data-driven methods depend heavily on sample quality and completeness, resulting in weak extrapolation and unclear mechanisms. This paper presents a modeling framework that synergizes POF with data-driven methods, combining their scalability in extrapolation and accuracy to address the shortcomings of reliability prediction for multilayer ceramic capacitors (MLCCs) under thermo-electric coupling (TEC) conditions. A performance-distributed degradation POF model is refined for MLCCs, considering both operating conditions and manufacturing parameters, and a breakdown reliability POF model is developed that incorporates the Schottky barrier and Weibull distribution. Furthermore, the neural networks are enabled to compensate for the prediction errors of the POF model by capturing the operating conditions, manufacturing parameters, and time-series features. Additionally, Bayesian optimization (BO) is employed to optimize the hyperparameters, thereby enhancing the prediction accuracy with limited sample sizes. The root mean square error (RMSE) of the hybrid model decreases by up to 82.78% compared to the POF model. Finally, the modeling framework demonstrated its potential through a 2000-h extrapolation proof-of-concept, which used a new MLCC type with the same structure and studied material.

基于测试的失效物理（POF）方法在很大程度上不能满足大规模生产的可靠性预测需求。相比之下，纯数据驱动的方法严重依赖于样本质量和完整性，导致外推能力弱，机制不明确。本文提出了一个将POF与数据驱动方法相结合的建模框架，将其外推的可扩展性和准确性相结合，以解决热电耦合（TEC）条件下多层陶瓷电容器（mlcc）可靠性预测的缺点。考虑运行条件和制造参数，对mlcc的性能分布退化POF模型进行了改进，并建立了包含Schottky势垒和Weibull分布的击穿可靠性POF模型。此外，神经网络能够通过捕获操作条件、制造参数和时间序列特征来补偿POF模型的预测误差。此外，采用贝叶斯优化（BO）对超参数进行优化，从而提高了有限样本量下的预测精度。与POF模型相比，混合模型的均方根误差（RMSE）降低了82.78%。最后，建模框架通过2000小时的外推概念验证展示了其潜力，该模型使用了具有相同结构和研究材料的新型MLCC类型。

{"title":"A hybrid modeling framework for predicting the multilayer ceramic capacitor reliability under thermal-electrical coupling operating conditions","authors":"Donghui Li , Yuguang Zhong , Xue Zhou , Guofu Zhai , Rui Kang","doi":"10.1016/j.ress.2026.112211","DOIUrl":"10.1016/j.ress.2026.112211","url":null,"abstract":"<div><div>Physics-of-failure (POF) methods based on testing are largely inadequate in meeting the reliability prediction demands of large-scale production. In contrast, pure data-driven methods depend heavily on sample quality and completeness, resulting in weak extrapolation and unclear mechanisms. This paper presents a modeling framework that synergizes POF with data-driven methods, combining their scalability in extrapolation and accuracy to address the shortcomings of reliability prediction for multilayer ceramic capacitors (MLCCs) under thermo-electric coupling (TEC) conditions. A performance-distributed degradation POF model is refined for MLCCs, considering both operating conditions and manufacturing parameters, and a breakdown reliability POF model is developed that incorporates the Schottky barrier and Weibull distribution. Furthermore, the neural networks are enabled to compensate for the prediction errors of the POF model by capturing the operating conditions, manufacturing parameters, and time-series features. Additionally, Bayesian optimization (BO) is employed to optimize the hyperparameters, thereby enhancing the prediction accuracy with limited sample sizes. The root mean square error (RMSE) of the hybrid model decreases by up to 82.78% compared to the POF model. Finally, the modeling framework demonstrated its potential through a 2000-h extrapolation proof-of-concept, which used a new MLCC type with the same structure and studied material.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"270 ","pages":"Article 112211"},"PeriodicalIF":11.0,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145978678","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

Variational Bayesian data assimilation with time-varying multi-physics-informed neural network for solving dimension-reduced probability density evolution equation 变分贝叶斯数据同化与时变多物理信息神经网络求解降维概率密度演化方程

IF 11 1区工程技术 Q1 ENGINEERING, INDUSTRIAL

Reliability Engineering & System Safety

Pub Date : 2026-01-08 DOI: 10.1016/j.ress.2026.112216

Teng-Teng Hao , Wang-Ji Yan , Jian-Bing Chen , Ka-Veng Yuen

The dimension-reduced probability density evolution equation (DR-PDEE) offers a powerful approach for analyzing how probability density functions (PDFs) evolve within intricate high-dimensional nonlinear stochastic dynamic systems. Effectively tackling the DR-PDEE relies on precisely identifying elusive spatiotemporal intrinsic drift functions that drive the evolutionary PDFs. Recent progress introduced a time-varying multi-physics-informed neural network (MPINN) to solve the DR-PDEE as an inverse problem, particularly in situations marked by noisy data, through a comprehensive utilization of available representative samples. However, two challenges including uncertainty quantification (UQ) in predictions and complexities in manually tuning loss weights are still not properly addressed. To tackle these issues, this study proposes a variational Bayesian data assimilation scheme that integrates Bayesian neural networks (BNNs) into the time-varying MPINN architecture to enable UQ for both intrinsic drift functions and response PDFs. An iterative coupled optimization strategy is employed for variational inference to iteratively estimate the posterior distributions of two types of parameters with different scales, encompassing the BNN hyperparameters and prediction error variances. By utilizing a pre-trained MPINN model as prior knowledge, the initial estimates of these variances are analytically derived, thereby obviating the requirement for time-consuming manual loss weight tuning. The proposed methodology is validated through several numerical examples, showcasing its robustness and accuracy. Compared to traditional deterministici MPINN, the new scheme not only provides enhanced UQ but also infers the prediction error variances that correlate with the data noise levels, offering valuable insights into determining optimal sample sizes for DR-PDEE-based stochastic response analysis.

降维概率密度演化方程（DR-PDEE）为分析复杂高维非线性随机动力系统中概率密度函数（PDFs）的演化提供了一种强有力的方法。有效地解决DR-PDEE依赖于精确识别驱动进化pdf的难以捉摸的时空内在漂移函数。最近的进展引入了时变多物理信息神经网络（MPINN）来解决DR-PDEE作为一个逆问题，特别是在有噪声数据的情况下，通过综合利用可用的代表性样本。然而，包括预测中的不确定性量化（UQ）和手动调整损失权重的复杂性在内的两个挑战仍然没有得到适当的解决。为了解决这些问题，本研究提出了一种变分贝叶斯数据同化方案，该方案将贝叶斯神经网络（bnn）集成到时变MPINN架构中，以实现对固有漂移函数和响应pdf的UQ。采用迭代耦合优化策略进行变分推理，迭代估计两种不同尺度参数的后验分布，包括BNN超参数和预测误差方差。通过使用预训练的MPINN模型作为先验知识，可以分析得出这些方差的初始估计，从而避免了耗时的手动减重调优的需要。通过数值算例验证了该方法的鲁棒性和准确性。与传统的确定性MPINN相比，新方案不仅提供了增强的UQ，而且还推断出与数据噪声水平相关的预测误差方差，为基于dr - pdee的随机响应分析确定最佳样本量提供了有价值的见解。

{"title":"Variational Bayesian data assimilation with time-varying multi-physics-informed neural network for solving dimension-reduced probability density evolution equation","authors":"Teng-Teng Hao , Wang-Ji Yan , Jian-Bing Chen , Ka-Veng Yuen","doi":"10.1016/j.ress.2026.112216","DOIUrl":"10.1016/j.ress.2026.112216","url":null,"abstract":"<div><div>The dimension-reduced probability density evolution equation (DR-PDEE) offers a powerful approach for analyzing how probability density functions (PDFs) evolve within intricate high-dimensional nonlinear stochastic dynamic systems. Effectively tackling the DR-PDEE relies on precisely identifying elusive spatiotemporal intrinsic drift functions that drive the evolutionary PDFs. Recent progress introduced a time-varying multi-physics-informed neural network (MPINN) to solve the DR-PDEE as an inverse problem, particularly in situations marked by noisy data, through a comprehensive utilization of available representative samples. However, two challenges including uncertainty quantification (UQ) in predictions and complexities in manually tuning loss weights are still not properly addressed. To tackle these issues, this study proposes a variational Bayesian data assimilation scheme that integrates Bayesian neural networks (BNNs) into the time-varying MPINN architecture to enable UQ for both intrinsic drift functions and response PDFs. An iterative coupled optimization strategy is employed for variational inference to iteratively estimate the posterior distributions of two types of parameters with different scales, encompassing the BNN hyperparameters and prediction error variances. By utilizing a pre-trained MPINN model as prior knowledge, the initial estimates of these variances are analytically derived, thereby obviating the requirement for time-consuming manual loss weight tuning. The proposed methodology is validated through several numerical examples, showcasing its robustness and accuracy. Compared to traditional deterministici MPINN, the new scheme not only provides enhanced UQ but also infers the prediction error variances that correlate with the data noise levels, offering valuable insights into determining optimal sample sizes for DR-PDEE-based stochastic response analysis.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"271 ","pages":"Article 112216"},"PeriodicalIF":11.0,"publicationDate":"2026-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146039124","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

Systemic risk analysis of complex socio-technical systems from the safety-II perspective 基于安全ii视角的复杂社会技术系统的系统性风险分析

IF 11 1区工程技术 Q1 ENGINEERING, INDUSTRIAL

Reliability Engineering & System Safety

Pub Date : 2026-01-07 DOI: 10.1016/j.ress.2026.112200

Massoud Mohsendokht , Huanhuan Li , Christos Kontovas , Chia-Hsun Chang , Zhuohua Qu , Zaili Yang

Modern complex socio-technical systems demand systemic risk analysis approaches that can holistically address the interdependencies between human, technological, and organizational components. Traditional models often fall short in capturing the dynamic and emergent nature of these interactions. This study introduces a novel, integrated risk analysis framework grounded in the Safety-II paradigm, which emphasizes understanding how systems succeed under varying conditions rather than focusing solely on failure. The proposed methodology combines the Functional Resonance Analysis Method (FRAM) with Bayesian Networks to overcome FRAM’s qualitative limitations and enable quantitative assessment of performance variability. The framework is further enriched by integrating complementary techniques, including Monte Carlo Simulation and canonical probabilistic models. This holistic toolkit enables a rigorous and scalable approach for modelling uncertainty and systemic variability across complex operational environments. The methodology is demonstrated through a case study of seaport operations, a representative example of a complex socio-technical system. The results show that the integrated Safety-II-informed framework improves the quantification of systemic risk and enhances the capacity to manage complexity and uncertainty in real-world settings.

现代复杂的社会技术系统需要系统的风险分析方法，可以全面地解决人、技术和组织组件之间的相互依赖关系。传统模型在捕捉这些交互的动态性和涌现性方面常常存在不足。本研究引入了一种基于Safety-II范式的新颖的综合风险分析框架，该框架强调理解系统如何在不同条件下成功，而不是仅仅关注故障。该方法将功能共振分析方法（FRAM）与贝叶斯网络相结合，克服了FRAM的定性限制，实现了性能可变性的定量评估。通过整合互补技术，包括蒙特卡罗模拟和规范概率模型，该框架进一步丰富。这个整体工具包为在复杂的操作环境中建模不确定性和系统可变性提供了严格和可扩展的方法。该方法通过海港业务的案例研究来证明，这是一个复杂社会技术系统的代表性例子。结果表明，综合安全ii知情框架改善了系统性风险的量化，增强了在现实环境中管理复杂性和不确定性的能力。

{"title":"Systemic risk analysis of complex socio-technical systems from the safety-II perspective","authors":"Massoud Mohsendokht , Huanhuan Li , Christos Kontovas , Chia-Hsun Chang , Zhuohua Qu , Zaili Yang","doi":"10.1016/j.ress.2026.112200","DOIUrl":"10.1016/j.ress.2026.112200","url":null,"abstract":"<div><div>Modern complex socio-technical systems demand systemic risk analysis approaches that can holistically address the interdependencies between human, technological, and organizational components. Traditional models often fall short in capturing the dynamic and emergent nature of these interactions. This study introduces a novel, integrated risk analysis framework grounded in the Safety-II paradigm, which emphasizes understanding how systems succeed under varying conditions rather than focusing solely on failure. The proposed methodology combines the Functional Resonance Analysis Method (FRAM) with Bayesian Networks to overcome FRAM’s qualitative limitations and enable quantitative assessment of performance variability. The framework is further enriched by integrating complementary techniques, including Monte Carlo Simulation and canonical probabilistic models. This holistic toolkit enables a rigorous and scalable approach for modelling uncertainty and systemic variability across complex operational environments. The methodology is demonstrated through a case study of seaport operations, a representative example of a complex socio-technical system. The results show that the integrated Safety-II-informed framework improves the quantification of systemic risk and enhances the capacity to manage complexity and uncertainty in real-world settings.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"270 ","pages":"Article 112200"},"PeriodicalIF":11.0,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145978679","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

Probabilistic risk uncertainty assessment for driver over-trust and under-trust in Level 3 human-automated driving systems cooperative driving based on the drift-diffusion model 基于漂移扩散模型的3级人-自动驾驶系统协同驾驶中驾驶员过度信任和信任不足的概率风险不确定性评估

IF 11 1区工程技术 Q1 ENGINEERING, INDUSTRIAL

Reliability Engineering & System Safety

Pub Date : 2026-01-07 DOI: 10.1016/j.ress.2026.112212

Song Ding , Lunhu Hu , Xing Pan , Jiacheng Liu , Fu Guo

Over-trust in automated driving systems (ADS) can trigger severe accidents, whereas under-trust may reduce system acceptance and efficiency. Thus, assessing risk uncertainty is critical for ensuring driving safety and enhancing system performance. This study aims to develop a cognitive model–based framework for risk uncertainty assessment in human-ADS cooperative driving, enabling precise tracking of the evolving risks of over-trust and under-trust. We propose a drift-diffusion model (DDM)–based risk uncertainty assessment approach applicable across diverse driving task scenarios. A driving simulation experiment was conducted with three levels of ADS reliability and five levels of task difficulty, yielding 7200 behavioral observations for model fitting and validation. The hierarchical Bayesian DDM demonstrated strong predictive performance, with simulated distributions closely matching experimental data. Results reveal that higher ADS reliability significantly shortens trust decision time, while the impact of task difficulty is non-monotonic. More importantly, the model successfully quantifies the time-varying risk uncertainty of over-trust and under-trust. These findings highlight the proposed framework as an effective and interpretable tool for evaluating time-varying risk uncertainty in human-ADS cooperation, providing a crucial model foundation for the future development of real-time risk prediction and intervention systems.

对自动驾驶系统（ADS）的过度信任会引发严重的事故，而信任不足则会降低系统的接受度和效率。因此，评估风险不确定性对于确保驾驶安全和提高系统性能至关重要。本研究旨在建立基于认知模型的人- ads合作驾驶风险不确定性评估框架，实现对过度信任和信任不足风险演变的精确跟踪。本文提出了一种基于漂移扩散模型（DDM）的风险不确定性评估方法，适用于不同驾驶任务场景。采用3个ADS信度等级和5个任务难度等级进行驾驶模拟实验，获得7200个行为观察值用于模型拟合和验证。分层贝叶斯DDM具有较强的预测性能，模拟分布与实验数据吻合较好。结果表明，较高的ADS信度显著缩短了信任决策时间，而任务难度对信任决策时间的影响是非单调的。更重要的是，该模型成功地量化了过度信任和信任不足的时变风险不确定性。这些发现突出表明，该框架是评估人类ads合作时变风险不确定性的有效且可解释的工具，为未来实时风险预测和干预系统的发展提供了重要的模型基础。

{"title":"Probabilistic risk uncertainty assessment for driver over-trust and under-trust in Level 3 human-automated driving systems cooperative driving based on the drift-diffusion model","authors":"Song Ding , Lunhu Hu , Xing Pan , Jiacheng Liu , Fu Guo","doi":"10.1016/j.ress.2026.112212","DOIUrl":"10.1016/j.ress.2026.112212","url":null,"abstract":"<div><div>Over-trust in automated driving systems (ADS) can trigger severe accidents, whereas under-trust may reduce system acceptance and efficiency. Thus, assessing risk uncertainty is critical for ensuring driving safety and enhancing system performance. This study aims to develop a cognitive model–based framework for risk uncertainty assessment in human-ADS cooperative driving, enabling precise tracking of the evolving risks of over-trust and under-trust. We propose a drift-diffusion model (DDM)–based risk uncertainty assessment approach applicable across diverse driving task scenarios. A driving simulation experiment was conducted with three levels of ADS reliability and five levels of task difficulty, yielding 7200 behavioral observations for model fitting and validation. The hierarchical Bayesian DDM demonstrated strong predictive performance, with simulated distributions closely matching experimental data. Results reveal that higher ADS reliability significantly shortens trust decision time, while the impact of task difficulty is non-monotonic. More importantly, the model successfully quantifies the time-varying risk uncertainty of over-trust and under-trust. These findings highlight the proposed framework as an effective and interpretable tool for evaluating time-varying risk uncertainty in human-ADS cooperation, providing a crucial model foundation for the future development of real-time risk prediction and intervention systems.</div></div>","PeriodicalId":54500,"journal":{"name":"Reliability Engineering & System Safety","volume":"271 ","pages":"Article 112212"},"PeriodicalIF":11.0,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145981143","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