基于混合神经遗传算法的梁式结构损伤识别研究

DAMAGE IDENTIFICATION OF BEAM STRUCTURE BASED ON HYBRID NEURAL GENETIC ALGORITHM

由于裂纹对结构动态行为具有重要影响,为识别结构裂纹的位置和深度,引入混合神经遗传算法。该方法将遗传算法(GA)与神经网络相结合,利用遗传算法全局寻优的特点对BP网络进行优化。通过有限元分析得到裂纹梁的固有频率,并将其作为神经网络的输入,裂纹的位置和深度作为网络的输出。首先,利用遗传算法(GA)优化网络的权重和阈值;然后将优化结果作为三层BP神经网络的初始值,经过样本数据的训练得到合适的网络;最后以裂纹梁固有频率作为测试值,得到裂纹参数的预测结果,通过理论值与预测值的比较,结果证明了该算法能够对结构损伤进行准确的识别。

It has established that a crack has an important effect on the dynamic behavior of a structure.To identify the location and depth of a crack on a structure,a method is presented which uses hybrid neuro-genetic technique.This method combines genetic algorithm（GA） with neural network,and uses genetic algorithm to optimize the BP network.Though the finite element analysis,acquire the first three natural frequencies of the beam,which used as inputs into neural network models for damage assessment.Genetic algorithm（GA） optimizes the weights and thresholds of the network,and using the optimization results as the initial value of the three layer BP neural network.Though the training of the sample data,then establish the suitable network.Finally,using the theoretical natural frequencies as the test values,obtain the predicted results of crack parameters.Comparing the theoretical value with the predicted value,the results show that this algorithm can identify the damage structure accurately.