基于遗传算法和BP神经网络的结构损伤识别

鉴于BP神经网络需要较长的训练时间、易陷入局部极小值、网络权值和阈值难确定等不足。运用遗传算法全局寻优的特点对BP网络的权值和阈值进行优化。同时运用遗传算法进行网络训练,避免网络收敛于局部极小值。通过对一根单梁实验数据的识别,结果表明两者的结合能对结构进行准确的识别。

Structure Damage Recognition Based on Genetic Algorithm and BP Neural Network

In view of the fact that the BP neural network needs the long training time,easy to fall into the partial minimum,the network weight and the threshold value difficult to grade firmly.This article utilizes the genetic algorithm overall situation optimization the characteristic carring on the optimization to the BP network weight and the threshold value.Meanwhile carries on the network using the genetic algorithm the training,avoids network restraining in the partial minimum.Through to Shan beam empirical datum recognition,the result indicates both the binding energy carries on the accurate recognition to the structure.