Shufeng Zhang

The traditional Monte Carlo simulation (MCS) method for time-varying reliability analysis (TRA) typically demands a large number of samples to achieve accurate reliability assessment results, and the total number of samples increases as the nonlinearity of the performance function increases. This paper proposes an adaptive surrogate model framework for TRA based on the back propagation (BP) neural network to reduce computational costs and enhance prediction performance. First, an improved multi-objective dwarf mongoose optimization algorithm (IMODMOA) is proposed to optimize BP neural network parameters, which can effectively improve model stability. Second, aiming to enhance model generalization and prediction accuracy, an adaptive sample selection strategy is designed to address the feature learning requirements of the BP neural network in TRA. Finally, an adaptive optimal BP surrogate model with high robustness is established by integrating IMODMOA parameters optimization, adaptive sample selection and dynamic training strategies, providing an efficient solution for TRA. Compared with conventional methods, the proposed method is validated through two numerical examples and one engineering case to confirm its superiority. The results demonstrate that the presented approach achieves superior performance in terms of prediction stability and accuracy compared with the existing methods.

Hydraulic systems are pivotal components in safety-critical aerospace and industrial applications, making reliable health monitoring essential. However, traditional data-driven diagnosis methods typically rely on static graph structures that fail to capture evolving sensor correlations during different fault modes. Furthermore, existing grid-based models often struggle to extract multi-resolution features and maintain performance under data-limited conditions. To address these challenges, this paper proposes a novel Dynamic Multi-Scale Graph Neural Network (DMS-GNN) for hydraulic system fault diagnosis. The framework integrates a hierarchical multi-scale feature extraction module to capture diverse fault signatures across different frequency bands. Crucially, a self-attention-based dynamic graph learner is introduced to adaptively infer latent sensor topologies end-to-end, eliminating the reliance on predefined physical connections. Experimental validation on a dedicated electro-hydraulic test bench demonstrates that the proposed DMS-GNN achieves a superior diagnostic accuracy of 98.47%, outperforming state-of-the-art baselines such as GraphSAGE, Static GCN, and GAT. The result confirms the efficacy of combining multi-scale temporal learning with dynamic spatial reasoning for robust multi-sensor fusion diagnosis.

This study employs micro-scale finite element modeling to investigate thickness-dependent in-situ transverse shear strength and damage mechanisms in cross-ply carbon fiber-reinforced polymer (CFRP) laminates. A computationally efficient representative volume element (RVE) model is developed, comprising a 90° ply with randomly dispersed fibers, matrix, and interlaminar resin regions embedded between 0° plies. The constitutive behavior is characterized using the Drucker–Prager model for matrix plasticity and bilinear cohesive models for interface debonding. The results demonstrate that the in-situ transverse shear strength increases by 43.74% when the ply thickness decreases from 125 μm to 60 μm, while further reduction to 40 μm causes strength fluctuations within 6.6%. Across the 125-40 μm thickness range, the strength variation remains below 10% with interlaminar resin thickness varying from 0 to 3.5 μm. These findings provide a quantitative theoretical foundation for optimizing ply thickness design and establishing reliability prediction models in CFRP laminated composites subjected to transverse shear loading.

Modern high-intensity armed conflicts often lead to extensive damage to urban infrastructure. The use of high-resolution remote sensing images can clearly detect damage to individual buildings which is of great significance for monitoring war crimes and damage assessments that destroy civilian infrastructure indiscriminately. In this paper, we propose SOCA-YOLO (Sampling Optimization and Coordinate Attention–YOLO), an automatic detection method for destroyed buildings in high-resolution remote sensing images based on deep learning techniques. First, based on YOLOv8, Haar wavelet transform and convolutional blocks are used to downsample shallow feature maps to make full use of spatial details in high-resolution remote sensing images. Second, the coordinate attention mechanism is integrated with C2f so that the network can use the spatial information to enhance the feature representation earlier. Finally, in the feature fusion stage, a lightweight dynamic upsampling strategy is used to improve the difference in the spatial boundaries of feature maps. In addition, this paper obtained high-resolution remote sensing images of urban battlefields through Google Earth, constructed a dataset for the detection of objects on buildings, and conducted training and verification. The experimental results show that the proposed method can effectively improve the detection accuracy of destroyed buildings, and the method is used to map destroyed buildings in cities such as Mariupol and Volnovaja where violent armed conflicts have occurred. The results show that deep learning-based object detection technology has the advantage of fast and accurate detection of destroyed buildings caused by armed conflict, which can provide preliminary reference information for monitoring war crimes and assessing war losses.

This study investigates the characteristics of the FRP failure envelope under combined compression‐shear loading, basing on experimental, theoretical and simulation analysis. The observed discrepancies in the PUCK criterion are shown to depend heavily on the ratio of transverse compressive strength to longitudinal shear strength. Meanwhile, the representative volume element (RVE) models are employed to analysis microscale factors influencing the FRP failure envelope. The results indicate that the failure envelope is significantly influenced by the interface reinforcement coefficient, which is strongly correlated with the ratio of uniaxial compressive strength to shear strength. The significance of this ratio lies in its macroscopic reflection of the interface reinforcement coefficient, and errors can be decreased by modifying the failure criterion according to this ratio. This main finding of this work is to reveal the dependence of the FRP failure envelope on the ratio of uniaxial strengths. On this basis, the optimal values for the inclined angle parameters in the PUCK failure criterion are proposed in relation to FRPs with varying interface reinforcement coefficients, which offering valuable insights for improving the accuracy of FRP failure predictions.Highlights: The interface reinforcement coefficient can influence biaxial failure envelope. Improving the slope parameter can considerably lower the prediction error. Dependence of failure envelope on the ratio of uniaxial strength was clarified.

In order to calculate stress redistribution after a fiber breaking for unidirectional fiber reinforced plastic (FRP) with practically poor fiber distribution uniformity, a new and efficient FE model is developed with high computation efficiency. Based on the proposed FE model, the stress concentration factor (SCF) of intact fibers around a broken fiber is investigated for unidirectional FRP. It is observed that the SCF depends heavily on the local fiber volume ratio (FVR), and a modified analytical SCF model is proposed which takes into account the variation of local FVR. It is proved that the modified analytical SCF model can provide SCF values with much better accuracy for FRP with poor fiber distribution uniformity. The modified analytical SCF model would be very useful for accurate prediction of progressive fiber breaking in FRP.