基于裂纹闭合解析模型和神经网络理论的疲劳裂纹扩展速率的计算  

PREDICTION OF FATIGUE CRACK GROWTH BASED ON THE ANALYTICAL CRACK CLOSURE MODEL AND NEURAL NETWORKS

作　　者：孙道恒[1] 胡俏[2] 徐灏[2] 

机构地区：[1]西南交通大学应用力学与工程系 [2]东北大学

出　　处：《机械工程学报》1998年第5期60-65,共6页Journal of Mechanical Engineering

基　　金：国家自然科学基金

摘　　要：用裂纹闭合的解析模型作为计算模型，运用神经网络理论对其中的协调方程进行求解，给出了相应的神经网络的结构和参数；对２２１９Ｔ８５１铝合金板材的疲劳裂纹扩展速率的计算进行了数值模拟，与实测结果对比相差很小。计算时间为电路时常数数量级。The Newman analytical crack closure model was used to compute the crack open stress and the crack growth rates. To reduce the computational time and increase the precision, a new method based on the neural networks is proposed to solve the compatible equations, the neural networks' structure and parameters are given. It was used to predict crack growth under constant amplitude loading on 2219 T851 aluminum alloy plate material. The predicted crack growth lives agreed well with experimental data. The computational time is within an elapsed time of the circuit time constant (ns).

关 键 词：疲劳裂纹 扩展速率 裂纹闭合理论 神经网络 

分 类 号：O346.1[理学—固体力学]