基于神经网络的结构疲劳寿命仿真的研究

Study on Simulation of Structural Fatigue-life Prediction Based on the Neural Network

疲劳是一个相当复杂的过程 ,受大量难以监控的因素影响。目前的疲劳寿命估算基本是采用传统的经验公式 ,计算程序复杂 ,精度低 ,试验成本高。随着人工神经网络的迅速发展 ,使用人工神经网络对疲劳寿命计算成为可能。该文通过对疲劳寿命的影响因素分析 ,采用改进的BP网络对疲劳寿命计算进行仿真 ,仿真结果表明这种方法对疲劳寿命估算是可行的 ,并且有很好的精度。

Fatigue reveals a important complex process, which is affected by many uncontrolled factors. The present method of fatigue-life calculation is based on traditional formulae, which have intricate process, considerable error. With the fast development of artificial neural network(ANN), it is possible that fatigue-life prediction(FLP) is simulated by ANN. In the paper, a method for the study of FLP is proposed. The factors that influence FLP are analyzed, improved BP network is applied to the simulation of FLP, the results show that the simulation can get better result than normal method.