基于BP网络的涡流探伤裂纹信号特征提取的研究

The Research of Crack Signal Character drawing in Eddy Current Flaw Detection Based on BP Network

用变频扫描励磁方法进行涡流定量检测以提取裂纹信号,利用三层结构的反向传播网络(Backpropagationnetwork,简称BP网络)建立了网络模型,并利用实际数据拟合参数,针对提离实验数据进行了计算仿真.结果表明:该网络模型算法收敛速度较快,输出精确,为实现板材裂纹深度的定量标定打下了理论基础.

We study quantitative eddy detection by scanning drive magnetism in changed frequency to collect signal of crack. We form the multi-layer back propagation network model, and simulate experiment date at computer. The simulating result indicate that the algorithm is rapid and the result is precision. This work is the fundamental theory in quantitative indication of crack depth on steel material surface by eddy current flaw detection.