摘要
应用人工神经网络技术,提取结构的固有频率的变化为特征参数,建立结构损伤识别模型,提出用遗传算法来调整神经网络的权值,并对一个框架模型进行了损伤数值模拟计算,即基于遗传算法-神经网络方法的结构损伤识别的研究。该方法弥补了传统的神经网络 BP 网络收敛速度慢,易陷入局部极小点等缺陷. 结构表明,该方法具有收敛速度快和识别精度高的特点。
Taking the variation of natural frequency as characteristic parameter, the damaged structures can be diagnosed by technique of artificial neural networks(ANN).The article works out the method that using genetic algorithm to optimize network’s weights. A frame structural damage based on numerical simulation is diagnosed. It is the research of the structure damage identification based on genetic algorithm-neural network method .The method remedies the defect on the traditional neural met work’s BP network restraining speed slowly, and fall into the partly extreme minimum value easily and so on. The results are more precise and faster than with traditional one.
出处
《辽宁工程技术大学学报（自然科学版）》
CAS
北大核心
2004年第5期636-637,共2页
Journal of Liaoning Technical University (Natural Science)
关键词
损伤识别
神经网络
神经网络权值
遗传算法
damage identification
artificial neural network
ANN’s weights
genetic algorithm
