基于GA-BP和GA-RBF网络的结构损伤识别

Comparison of the Structure Damage Analysis between GA-BP Neural Network and GA-RBF Neural Network

首先指出了当人工神经网络算法解决结构工程实践问题时,网络结构本身所面临的缺陷;然后描述了人工神经网络和遗传算法的概念,从理论和实例上说明了运用遗传算法优化和改进神经网络结构的可行性,以结合二者的长处解决工程实践问题;接着详细阐述了如何利用遗传算法优化或改进BP(B ack P ropagation)网络模型和RBF(R ad ia l B as is Function)网络模型,以及如何利用遗传优化BP网络和遗传优化RBF网络模型分析结构损伤,进而比较遗传BP网络和RBF网络在结构损伤分析方面的性能。

When applying artificial neural network model to the identification of structure damage, some disfigurements of the network model itself are pointed out. Then the concept and relationship of genetic algorithm and artificial neural network are introduced. It is stated that how to optimize BP （Backpropagation） neural network model and RBF （Radial Basis Function） neural network model and how to apply GA-BP neural network and GA-RBF neural network to analyze structure damage. According to the identification results, performance of GA-BP neural network and GA- RBF neural network in analyzing structure damage are finally compared.