小波神经网络在电磁无损检测中的应用

摘要

Based on orthogonal wavelet transform wavelet neural network model of structure, this paper puts forward a new algorithm of wavelet neural network on the basis of selecting B-wavelet for wavelet base. The method of selecting typical training samples is used to enhance the wavelet neural network, and the wavelet neural network is used to implement the compression and reconstruction for the wire rope electromagnetic nondestructive testing signals. The experimental results indicate that the wavelet neural network can perform the intelligentized detection for defects signals. Compared to traditional identification methods, the accuracy rate of the wavelet neural network algorithm is significantly improved.%在基于正交小波变换的小波神经网络模型构造的基础上，选取B-小波为小波基提出了一种小波神经网络新算法。通过选取典型训练样本集的方法提高了该小波神经网络诊断的准确性，并运用该小波神经网络对钢丝绳电磁无损检测信号进行了压缩与重构，试验结果证明了小波神经网络能够较为理想地完成缺陷信号的智能化检测，采用小波神经网络算法比传统的断丝识别方法准确率提高了很多。