Fatigue damage accumulation will not only cause the degradation of material performance, but also lead to the growth of effective stress and critical stiffness. However, the existing fatigue reliability models usually ignore the effective stress growth and its influence on critical stiffness of composite material. This study considers the combined effects of performance degradation and effective stress growth, and a pair of fatigue reliability models for composite material is presented. Firstly, the fatigue damage in composite material is quantified by its performance degradation, and the fitting accuracy of several typical fatigue damage models is compared. Subsequently, the uncertainties of initial strength and initial stiffness are considered, and a pair of probabilistic models of residual strength and residual stiffness is proposed. The performance degradation data of Gr/PEEK [0/45/90/-45] laminates are utilized to verify the proposed probabilistic models. Finally, the effective stress growth mechanism and its influence on failure threshold are elaborated, and a pair of fatigue reliability models for composite material is developed. Moreover, the differences between strength-based and stiffness-based reliability analysis results of composite material are compared and discussed. Keywords: fatigue reliability; performance degradation; effective stress; fatigue damage; composite material

A new algorithm optimization‐based hybrid neural network model is proposed in the present study for the multiaxial fatigue life prediction of various metallic materials. Firstly, a convolutional neural network (CNN) is applied to extract the in‐depth features from the loading sequence composed of the critical fatigue loading conditions. Meanwhile, the multiaxial historical loading information with time‐series features is retained. Then, a long short‐term memory (LSTM) network is adopted to capture the time‐series features and in‐depth features of the CNN output. Finally, a full connection layer is used to achieve dimensional transformation, which makes the fatigue life predictable. Herein, the hyperparameters of the LSTM network are automatically determined using the slime mold algorithm (SMA). The test results demonstrate that the proposed model has pleasant prediction performance and extrapolation capability, and it is suitable for the life prediction of various metallic materials under uniaxial, proportional multiaxial, nonproportional multiaxial loading conditions.

A new algorithm optimization-based hybrid neural network model is proposed in the present study for the multiaxial fatigue life prediction of various metallic materials. Firstly, a convolutional neural network (CNN) is applied to extract the in-depth features from the loading sequence comprised of the critical fatigue loading conditions. Meanwhile, the multiaxial historical loading information with time-series features is retained. Then, a long short-term memory (LSTM) network is adopted to capture the time-series features and in-depth features of the CNN output. Finally, a full connection layer is used to achieve dimensional transformation, which makes the fatigue life predictable. Herein, the hyperparameters of the LSTM network are automatically determined using the slime mould algorithm (SMA). Five data sets of materials are involved for case studies. The predictive performance of the proposed model is compared with those obtained using support vector machine (SVM), LSTM network, and a critical plane model. The results demonstrate the better predictive performance of the proposed model, and it is suitable for the life prediction of various metallic materials under uniaxial, proportional multiaxial, non-proportional multiaxial loading conditions. Besides, the proposed model outperforms the SVM and LSTM network in terms of extrapolation capability.