最新刊期
卷 2 ， 期 1 ， 2026
- 摘要：As a critical link in the drive system of the finishing mill, distribution boxes play a vital role in rolling force transmission and distribution. Therefore, researching fault diagnosis for distribution boxes is essential for ensuring the security and reliability of process operation. However, when fault characteristics are indistinct and the fault type is unknown, traditional diagnostic methods may fail to provide reliable diagnosis. To address these challenges, this paper proposes a cross-domain open-set model that leverages adversarial learning to enhance feature representations under indistinct fault characteristics, and integrates distribution-based methods for identifying unknown faults. Considering the importance of feature representation for adversarial learning and unknown fault identification, we build an information-exchange feature extractor integrates depthwise separable convolution and GCN. This structure extracts depth features and addresses the channel independence limitation of traditional depthwise separable convolution. To identify unknown faults in distribution boxes, the saddle-point approximation (SPA) method is employed. This method is used during training to establish a distribution for each known class, and these distributions are utilized in testing to identify unknown samples. Unlike traditional methods for establishing distributions, the SPA method directly uses data without relying on predefined distribution assumptions or formulas. To verify the superiority of the proposed method, we conducted a comparative experiment based on historical fault data of distribution boxes. The results demonstrate that the proposed method achieves an accuracy of 94.13% and an H-score of 91.35%, which demonstrates the superior performance in classifying known faults and identifying unknown faults in distribution boxes.关键词：fault diagnosis;open-set;adversarial network;finishing mill;distribution boxes1|0|0更新时间：2026-04-14
- 摘要：Industrial process systems often struggle to achieve high availability due to complex interdependencies, limited redundancy, and rapidly escalating costs in high-reliability regimes. This study presents a reliability-centered design review for a wastewater treatment plant's evaporation system, combining reliability block diagram (RBD) modeling with a cost-sensitive reliability allocation framework. Two alternative configurations, series–parallel (equipment-level redundancy) and parallel–series (train-level redundancy) were evaluated. To address the absence of detailed cost data in early design, a dimensionless exponential cost index was developed to represent nonlinear cost escalation as reliability approaches practical limits. This index, derived from fuzzy linguistic ratings and defuzzification, was incorporated into a severity-effort-cost weighting scheme to cascade system-level reliability targets to subsystems and individual equipment. Application to the multi-train evaporation system shows that equipment-level redundancy improves overall system reliability by 11.45% over three months compared with train-level redundancy. The proposed allocation consistently prioritizes failure-prone equipment while avoiding over-investment in already robust units, producing reliability targets that are technically justified and economically rational. The integrated RBD and fuzzy cost-driven allocation framework provides a practical method for early-stage design and retrofit, guiding redundancy planning, focusing improvement actions where they yield the greatest reliability benefits, and enhancing operational resilience in complex engineered systems.关键词：series–parallel configuration;k-out-of-n system;redundancy optimization;reliability allocation;cost-sensitivity1|0|0更新时间：2026-04-14
- 摘要：A theoretical system for digital modeling, decoupling, and prediction of the reliability for aerospace products is proposed in order to solve the problem of the reliability issues of aerospace ‘small sample’ products facing high-density launch missions under multi stress coupling. Key technologies such as digital modeling of the reliability of aerospace electromechanical products based on response surfaces, and digital decoupling analysis of reliability under multi stress coupling, and life reliability analysis of aerospace products under multi stress coupling have been overcome. Engineering applications have been carried out in aerospace machinery, electromechanical, and electronic products, achieving comprehensive level improvement of aerospace product support for aerospace equipment with ‘long life, high reliability, high precision, and high performance’.关键词：aerospace products;performance and reliability;digital simulation;modeling;decoupling;prediction28|54|0更新时间：2026-04-14
-
Combined fatigue reliability analysis of engine turbine blade via direct probability integral method AI导读
摘要：The stochastic fatigue life and reliability evaluation for aero-engine turbine blades under complex operating conditions is a crucial yet challenging task in engine design and maintenance. In this study, a unified framework for hybrid high-, low-cycle and creep fatigue life and reliability analysis of turbine blades based on direct probability integral method (DPIM) is proposed. Firstly, the basic theories of combined high-and-low-cycle fatigue (CCF), creep–fatigue cumulative damage, and fatigue reliability are introduced. Secondly, the formulas of the mean, standard deviation, and probability density function of stochastic fatigue life are derived via the probability density integral equation. For CCF reliability analysis, an efficient hybrid time–frequency domain method is established, in which the stress power spectral density function is obtained via frequency-domain method, and the stochastic stress time histories are generated. Moreover, DPIM is utilized to perform the creep–fatigue reliability estimation of turbine blades. Finally, the high accuracy and efficiency of the proposed method are validated against direct Monte Carlo simulation (MCS) and surrogate model-based MCS. The results demonstrate that DPIM can uniformly and efficiently solve the statistical moments of stochastic combined fatigue life and fatigue reliability for turbine blades. Additionally, the significant effect of component coupling terms in combined fatigue damage on reliability is revealed.关键词：combined fatigue reliability analysis;engine turbine blade;high-and-low-cycle fatigue;creep–fatigue;direct probability integral method0|0|0更新时间：2026-04-14 - 摘要：In the evaluation of system reliability based on dependent competing failure process, it is essential to consider the effect of self-healing, and ignoring self-healing may lead to errors in the evaluation results. Existing models only considers the self-healing in soft failure or hard failure, without considering self-healing of both failure processes simultaneously which lacks reality. In this paper, a general reliability model for multi-component system based on different shock models is established. During the soft failure process, a first-kind self-healing factor is introduced to make the degradation increment change with time in this model. Two second-type self-healing factors are introduced into the hard failure process, which causes the hard failure threshold under extreme shock model and the cumulative damage of cumulative shock model to change with the shock magnitude and numbers of shock. A new reliability model for a parallel-series system is established and the analytical expression is derived through combining dynamic self-healing mechanism with multi-component system. Finally, an illustrative example of self-healing gearbox is given to verify the feasibility of the model.关键词：self-healing;multi-component;dependent competing failure process;random shock;internal degradation0|0|0更新时间：2026-04-14
- 摘要：Traditional weighted k-out-of-n systems often assume fixed component weights, which limits their applicability in complex engineering scenarios where component weights exhibit continuous randomness. This limitation is particularly critical in redundant safety instrumented systems with common cause failures, such as renewable energy systems, where components experience heterogeneous degradation, dependent failure patterns, and multi-stage performance decline. To address this limitation, this study proposes a novel k-out-of-n: G system that incorporates multi-type components with continuous random weights, and evaluates its reliability and mean time to failure across two operating states: normal operating state and degraded state. First, the system's operating mechanism is defined, incorporating both the number of surviving components and their sum of weights. Second, the distribution of the system Integrity and the system reliability are derived, considering the dependencies modeled by different Copula functions. Then, Monte Carlo simulations are conducted to investigate the impacts of different dependence structures (fully dependent, inter-class dependent, fully independent) and lifetime correlation patterns on system reliability. Furthermore, the simulation results indicate that mis-specifications of components' relationships can lead to estimation errors. Additionally, the presence of dependence among components can enhance overall system reliability to some extent. Finally, a numerical case study of an aircraft pneumatic system validates the feasibility and practicality of the proposed model.关键词：continuous random weight;k-out-of-n system with multi-type components;degraded state;Copula-based dependence;mean time to failure (MTTF)34|51|0更新时间：2026-04-14
- 摘要：Data streams generated by complex industrial systems commonly exhibit concept drift together with evolving spatial-temporal dependencies. These factors jointly impair the performance and reliability of conventional prediction models that assume stationary data distributions. To address this challenge, this paper proposes an uncertainty-aware hierarchical graph neural network (UHGNN) tailored for drifting data streams. The method explicitly extracts temporal and spatial features in data streams through a hierarchical graph architecture that integrates physical topology with high-order interactions across sensors. In addition, an epistemic uncertainty-driven drift detection mechanism is incorporated to identify concept drift in real time and to initiate fine-tuning for rapid adaptation. Experiments conducted on a real-world online industrial system, the diesel hydrofining process dataset, show that UHGNN effectively alleviates the adverse influence of concept drift and achieves higher predictive accuracy compared with baselines for drifting data streams. The method further demonstrates strong robustness with respect to key hyperparameters and maintains computational efficiency that supports industrial deployment.关键词：concept drift;data stream;hierarchical graph neural network;epistemic uncertainty28|46|0更新时间：2026-04-14
Papers
- 摘要：Intelligent autonomous systems (IASs), encompassing autonomous vehicles, unmanned aerial vehicles, and robotic platforms, have revolutionized sectors ranging from transportation to defense. By leveraging artificial intelligence (AI) techniques, these systems are characterized by data-driven learning paradigms, integrated architectures, and adaptive decision-making. However, these novel capabilities introduce distinctive failure challenges, including data-induced biases, limited interpretability, and vulnerability to adversarial perturbations. Such features necessitate enhanced reliability to ensure dependable operation in dynamic, safety-critical environments. This review synthesizes the state-of-the-art in IAS reliability engineering, framed through four interconnected dimensions: accuracy, generalization, robustness, and explainability. Drawing from interdisciplinary advancements in machine learning, AI and reliability, we examine methodologies such as multi-sensor fusion, meta-learning, adversarial training, and ante-hoc interpretability constraints, supported by empirical evidence from real-world deployments. Key insights highlight substantial progress in mitigating IASs vulnerabilities, yet persistent challenges including performance-reliability trade-offs, degradation under extreme conditions, and scalability limitations impede widespread adoption. We propose future directions emphasizing hybrid frameworks, causal inference and lightweight models to advance reliable IASs. By bridging theoretical foundations with practical implementations, this work provides a comprehensive roadmap for developing reliable and trustworthy autonomous systems that prioritize safety, efficiency, and societal well-being.关键词：autonomous systems;reliability engineering;Artificial Intelligence0|0|0更新时间：2026-04-14
- 摘要：Valves constitute the largest population of active components in the ‘Defense in Depth’ philosophy of nuclear power plants (NPPs), and their reliability is crucial for nuclear safety. Operational experience and numerous studies indicate that valves remain a critical weak link, yet existing academic reviews are often limited to a single failure mechanism or the operation & maintenance (O&M) phase. To address the lack of a systematic perspective centered on this critical component, this paper constructs a full life-cycle reliability framework, systematically reviewing the current status and progress of NPP valve reliability research from four interrelated pillars: (1) reliability-oriented design and optimization; (2) key failure modes and mechanism analysis; (3) reliability assessment and life prediction; (4) condition monitoring and intelligent O&M. Complementing prior reviews limited to specific phases, this paper utilizes the proposed full life-cycle framework to analyze the profound paradigm shift in NPP valve reliability research—transitioning from static, isolated, physics-dominated analysis toward dynamic, systematic, deep physics-data fusion. We identify four core contradictions hindering this evolution: the conflict between small samples and data-greedy methods, complex physics and simplified models, mechatronic systems and single components, and the black-box nature of models versus the need for explainability. To address these challenges, we provide a forward-looking perspective focused on deepening physics-data fusion, coupling component-level remaining useful life with system-level probabilistic risk assessment, and exploring solutions for model trustworthiness and uncertainty quantification.关键词：nuclear power plant;valves;reliability;full life-cycle;physics-data fusion0|0|0更新时间：2026-04-14
Topical Reviews
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备09064830号-19
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0