一种基于自适应遗传算法优化BP神经网络的漏磁缺陷量化框架

MFL defect quantification framework based on combination of adaptive genetic algorithm and BP neural network

由于传统遗传算法在优化BP神经网络应用中训练速度慢、拟合效果不好,所以神经网络对管道漏磁缺陷的量化精度差。将自适应遗传算法引入到BP神经网络中,进行漏磁缺陷的尺寸反演,根据实测漏磁缺陷的数据特点,自适应设定交叉算子和变异算子的交叉率和变异率,进而优化BP网络的初始权值;采用不同尺寸的缺陷特征库训练网络,实现对缺陷长度、宽度、深度的量化。

In order to overcome the disadvantages of BP neural network based on traditional genetic algorithm,such as slow training speed,low quantitative accuracy,based on improving the computing method of the rate of crossover and mutation,also taking current stage of the population′s evolution into consideration,an improved adaptive genetic algorithm is introduced.This new algorithm is used to optimize BP network′s initial weights,and the defect features from magnetic flux leakage are used to train the network,so as to achieve the quantification of length,width,depth of the defects.With this model,the training time of net work can be saved and the computing accuracy can be improved as well.