结构疲劳寿命及可靠度的神经网络预测方法

A Method Based on Neural Network to Predict Fatigue Life and Reliability of Structures

基于疲劳曲线,考虑应力范围的随机性,获得了结构疲劳寿命所服从的分布,计算出期望寿命的可靠度.采用人工神经网络后向传播BP模型,把应力范围与所期望的疲劳寿命作为输入节点,期望疲劳寿命的可靠度作为输出节点;或者把应力范围与所期望的可靠度作为输入节点,期望可靠度下的疲劳寿命作为输出节点,进行网络学习,建立应力范围、结构疲劳寿命及可靠度之间的关系,用于预测某应力范围下所期望可靠度的疲劳寿命,或者所期望疲劳寿命的可靠度.通过具体算例表明该方法正确,精度满足工程要求.

Based on the fatigue curve, the life distribution of structures and the reliability of the expected life can be obtained considering the randomness of the stress range. The relationship among the stress range, the life of the structures and the reliability can be established by using the model of BP networks. In the BP networks, the stress range and expected fatigue life are used as the input nodes, and the reliability of expected life is used as the output node. Similarly, the stress range and expected reliability can also be used as the input nodes, and the fatigue life of the expected reliability is used as the output node. The fatigue life of the expected reliability or the reliability of the expected life under a certain stress range can be predicted by using the trained network. This approach is demonstrated to be correct and can satisfy the accuracy of engineering by an example.