基于小波分析的脉冲涡流信号去噪方法研究

摘要

应用小波多尺度分解与重构技术和小波包分析技术对脉冲涡流检测信号进行处理,将信号中的噪声分离出来并去除.采用离散小波变换的强制消噪法、软阈值消噪法和不同熵准则的小波包消噪法,对实测的脉冲涡流信号进行消噪.在保留信号中有用成分的前提下,结合信号的信噪比和均方根误差,对消噪方法进行比较和优选.实验结果表明,小波包分析中的Threshold熵准则法效果最好.%This paper utilizes multi-scale decomposition and reconstruction technology of wavelet and wavelet packet analysis (WPA) technology to do preprocessing for signals, separates noises from pulsed eddy current (PEC) testing signals and removes them. Discrete wavelet transform (DWT) and wavelet packet analysis (WPA) de-noising methods are used for PEC testing sig-nals. Under the pre-condition of retaining useful component of the signals, combining with SNR and RMSE to compare the meth-ods and the best de-noising method is recommended. The results show that threshold entropy principle method of WPA is better than others.