查看更多>>摘要：The inherent spatial variability of soil is reported to significantly impact landslide debris behaviors. In this study, the effect of spatial variability on the inundation and impact processes of debris flow is investigated using a multi-phase depth-averaged model. The dynamic process of a debris flow, considering spatial variabilities of basal friction and initial solid concentration, is explored via Monte Carlo simulation. The results show that due to the flow channel constrain and spatial averaging, the influences of spatial variability on the global impact of debris flow are not significant. However, remarkable influences on the local impact are found. From the upstream of flow channel to the downstream of river, there is a decreasing trend in uncertainties regarding the material composition and flow dynamics at local spots. In the flow channel, the mean values of flow depths are smaller than those in the deterministic analysis, while those of flow velocities are larger. In the river, both the mean values of flow depths and velocities are close to those in the deterministic analysis while their variations remain significant even downstream of river. The findings provide insights into the spatial variability effects on debris flow impact and facilitate risk assessment.
查看更多>>摘要：The health state of equipment will decline in the long-term operation, resulting in the need to rotate multiple equipment to fulfill the operation task (OT). In the current engineering, three available equipment rotation strategies are summarized. However, the selection of these strategies is arbitrary and the health state of the equipment during operation rotation is neglected, which causes poor benefits and heavy risks. For this purpose, a quantitative decision-making mechanism using belief rule base (BRB) and evidential reasoning (ER) is proposed to determine the preferred strategy. Specifically, BRB serves as the preferred decision model, which reflects the mapping relationship between the OT and the rotation strategy. A parameter optimization model is then designed to improve decision rationality. To obtain the labeled historical OTs required for the parameter optimization model, a hierarchical ER method is developed to evaluate the performance of the rotation strategy to obtain the labeled historical OTs, where the health state of the equipment is quantitatively analyzed. The proposed method comprehensively utilizes knowledge and data and provides a quantitative decision-making framework for equipment operation rotation. A case of the natural gas storage tank (NGST) verifies the effectiveness of the proposed method.
查看更多>>摘要：Intelligent fault diagnosis and remaining useful life (RUL) prediction are essential for the reliable operation of rotating machinery. These technologies enhance safety, availability, and productivity in the manufacturing industry. Graph Convolutional Networks (GCNs), an extension of deep learning (DL) to graph data, provide superior performance due to their unique data representation capabilities. Traditional condition monitoring (CM) typically requires separate models for fault diagnosis and RUL prediction, leading to challenges such as ineffective knowledge sharing and high costs associated with preparing and deploying DL models. To address these issues, this study proposes a multi-task graph isomorphism network with an attention mechanism for simultaneous fault diagnosis and RUL prediction. This method considers the interrelationship between tasks, introducing both a parameter-sharing mechanism and a self-attention mechanism. Compared to traditional single-task methods, the proposed approach offers higher accuracy, greater practicality, and reduced costs of developing the model. The effectiveness of the method is validated using experimental degradation data, demonstrating its capability to address key issues in fault diagnosis and RUL prediction, exhibiting strong potential in practical applications.
查看更多>>摘要：Commercial high-voltage circuit breaker (CB) condition monitoring systems rely on directly observable physical parameters such as gas filling pressure with pre-defined thresholds. While these parameters are crucial, they only cover a small subset of malfunctioning mechanisms and usually can be monitored only if the CB is disconnected from the grid. To facilitate online condition monitoring while CBs remain connected, non-intrusive measurement techniques such as vibration or acoustic signals are necessary. Currently, CB condition monitoring studies using these signals typically utilize supervised methods for fault diagnostics, where ground-truth fault types are known due to artificially introduced faults in laboratory settings. This supervised approach is however not feasible in real-world applications, where fault labels are unavailable. In this work, we propose a novel unsupervised fault detection and segmentation framework for CBs based on vibration and acoustic signals. This framework can detect deviations from the healthy state. The explainable artificial intelligence (XAI) approach is applied to the detected faults for fault diagnostics. The specific contributions are: (1) we propose an integrated unsupervised fault detection and segmentation framework that is capable of detecting faults and clustering different faults with only healthy data required during training (2) we provide an unsupervised explainability-guided fault diagnostics approach using XAI to offer domain experts potential indications of the aged or faulty components, achieving fault diagnostics without the prerequisite of ground-truth fault labels. These contributions are validated using an experimental dataset from a high-voltage CB under healthy and artificially introduced fault conditions, contributing to more reliable CB system operation.
查看更多>>摘要：In recent years, deep learning-based approaches have demonstrated superior performance in few-shot fault diagnosis. Nevertheless, many of these methods lack explicit interpretability, making it difficult to intuitively understand their diagnostic logic. To tackle this issue, an interpretable deep learning model called the adaptive frequency attention-based interpretable Transformer network is proposed for few-shot fault diagnosis of rolling bearings. From a frequency interpretability perspective, the standard Transformer network architecture has been innovatively improved. First, an adaptive frequency attention mechanism is developed that quantifies the importance of various frequency components during the diagnostic process, adaptively identifying and emphasizing key frequency components closely associated with fault modes. This boosts both diagnostic performance and model interpretability. Second, to enhance the diversity of fault features under limited sample conditions, a multiscale convolutional architecture is developed to replace the linear projection layer in input embedding. This architecture employs parallel multiscale convolution kernels to extract both local and global fault features, enabling a comprehensive capture of fault information and further supporting the interpretability of the diagnostic model. Finally, Experiments on interpretable few-shot fault diagnosis are carried out on three rolling bearing datasets, and the diagnostic results further validate the effectiveness and interpretability of the proposed method.
查看更多>>摘要：Efficient time-variant reliability assessment for complex systems is of great interest but challenging as the highly complex multiple output responses under time-variant uncertainties are hard to quantify. The task becomes even more challenging if the interconnected dependencies between multiple failure modes are involved. In this study, an eXtreme physics-embedding multi-response regressor (X-PMR) is presented for time-variant system reliability assessment. Firstly, by transforming time-variant multiple responses to time-invariant extreme values, an eXtreme multi-domain transformation concept is presented, to establish the time-invariant multi-input multioutput (TiMIMO) dataset; moreover, by embedding physics/mathematics knowledge into multi-objective ensemble modeling, a physics-embedding multi-response regressor is proposed, to synchronously construct the surrogate model for highly complex multiple output responses. The validation effectiveness and benefit illustration of the X-PMR method are revealed by introducing three numerical systems (i.e., series system, parallel system and series/parallel hybrid system) and a real application system (i.e., dynamic aeroengine turbine blisk), in comparison with a number of state-of-the-art methods investigated in the literature. The current efforts can provide a novel sight to address the time-variant system reliability assessment problems.
查看更多>>摘要：This contribution presents a novel approach for estimating the Sobol' index, which has been commonly employed in variance-based sensitivity analysis of computational models that may often involve multiple uncertain parameters. Specifically, a single-loop Monte Carlo simulation (MCS) method, which is independent of the dimensionality of inputs, is proposed to reduce the computational cost of complicated models. proposed method is realized by developing a new estimator of the Sobol' index computed via the twodimensional kernel density estimation, which can be easy programming while ensuring high accuracy. Numerical examples are studied to demonstrate the advantages of the proposed method.
查看更多>>摘要：The probability distribution of the performance function plays an important role in many fields. However, it is challenging to obtain this distribution because of the difficulty in capturing the tails on both sides, particularly for high-dimensional problems. To estimate the probability distribution of the performance function efficiently and accurately, this study proposes a piecewise fitting method based on the simulation-based Bayesian postprocessing method. The method first divides the whole distribution into the main body and the left and right tail distributions. Subsequently, the samples for the main body are generated by a randomized Sobol sequence, while the samples for the left and right tails are produced through Markov chain Monte Carlo sampling. Thereafter, the shifted generalized lognormal distribution model is applied to reconstruct the main body distribution, and the truncated shifted generalized lognormal distribution is used to fit the tail distributions. Finally, the overall distribution is obtained, and the shape parameters of the distribution model are determined using Bayesian estimation methods. The efficiency and accuracy of the proposed method are demonstrated through four numerical examples, including a simple toy example and cases involving strongly nonlinear, implicit, highdimensional performance functions.
查看更多>>摘要：It is important in many respects to have information about the number of failed components in the system when or before a system fails. This paper investigates the mean number of failed components at or before the failure time of the linear consecutive k-out-of-n:F system which is a useful structure to model various engineering systems such as transportation and transmission systems. In particular, closed form expressions for the mean number of failed components within the system that have discretely distributed components lifetimes are obtained. The results are used to estimate the unknown parameter of the components' lifetime distribution and to find the optimal replacement cycle that minimizes the expected cost per unit of time under a certain age-based replacement policy.
查看更多>>摘要：This paper introduces a comprehensive framework for managing interdependent delay risks in nuclear power plant (NPP) construction by integrating network theory and topological analysis. Spent fuel disposal, nuclear plant safety and nuclear weapons proliferation are known important concerns for nuclear power development, but costs remain the fundamental problem, as NPP projects are plagued by schedule delays that substantially increase total costs. Such complex megaprojects are exposed to numerous risks of different sources that behave interdependently. Most of the studies understand the risks of delay in NPP construction projects in isolation without taking interdependencies into account. The proposed methodology employs a Design Structure Matrix (DSM) to construct a Risk Interaction Network (RIN), enabling a topological assessment to identify critical risks that may cause cascading delays in project tasks. An algorithmic search for these critical risks is conducted, considering the impact of their removal on the RIN's characteristics. We define a bi-objective optimization problem aimed at generating a project schedule that minimizes both the project's makespan and the reachability density of the RIN. The solution is obtained using an evolutionary algorithm. Applied to a Double-Containment Pressurized Water Reactor (DC-PWR) project, this approach effectively uncovers risks neglected by classical analysis and offers scheduling options for different risk attitudes, enhancing decision-making capabilities.
NETL
NSTL
Elsevier