人工神经网络BP算法的改进及其在无损检测中的应用

摘要

采用多层感知器(MLP)与误差反向传播算法(error back-propagationalgorithm)构造与监督训练人工神经网络，采用了改进的非线性激励函数与学习率的误差反向传播算法。超声无损检测的计算机模拟与实验结果表明，改进的BP算法收敛速度较之常规BP算法明显加快。%The model of multilayer perceptron (MLP) and back-propagation(BP) training algorithm in artificial neural network are employed.New ideas are proposed to improve BP algorithm in aspects of nonlinear activation function and learning rate.Simulation and experiment results are also presented to demonstrate the effect of improvement in ultrasonic nondestructive testing.