电缆终端搪铅涡流检测信号小波去噪参数选择方法

Wavelet Denoising Parameter Selection Method for Eddy Current Detection Signals of Lead Sealings of Cable Terminal

摘要:

电缆终端搪铅质量直接影响高压电缆的安全稳定运行，有必要对其进行脉冲涡流无损检测，以排查是否存在缺陷。在实际检测时，涡流检测信号容易受到白噪声干扰，小波去噪是一种有效的白噪声滤除手段。然而，以往的研究通常根据经验选取小波参数，忽略了参数变化对涡流检测信号实际去噪效果的影响。因此，提出了基于粒子群优化算法的电缆终端搪铅涡流检测信号小波去噪参数选择方法，以归一化相关系数作为适应度函数，通过量化去噪涡流检测信号在峰值附近的失真程度，获得了最优小波去噪参数为Sym25小波基、10层分解层数和中位数阈值函数，保证了降噪涡流检测信号峰值大小与峰值时间的相对误差小于3%。研究成果可为确定不同噪声干扰下的涡流检测信号小波去噪参数自动选择提供方法参考。

Abstract:

The quality of lead sealings of cable terminal has a direct impact on safe and stable operation of high-voltage cables, so it is necessary to carry out the pulse eddy current non-destructive testing to avoid the defects. But the eddy current detection signals can be impacted by noise in actual engineering, and wavelet denoising is an effective measure to filter out the white noise. However, the previous researches on wavelet denoising have typically selected wavelet parameters based on experiences, ignoring the impact of parameter variations on actual denoising effect. Hence, a wavelet denoising parameter selection method for eddy current detection signals is proposed based on particle swarm optimization algorithm, which adopts an evaluation index called normalized correlation coefficient (NCC) as fitness function. Through quantifying distortion degree of denoising eddy current detection signals near the peak, the optimal wavelet denoising parameters are obtained, that is, Sym25 wavelet, ten-layer decomposition layer and median threshold function, which makes the relative error between peak value of eddy current detection signals after denoising and peak time be less than 3%. The research results can provide a method reference for automatically determining the optimal wavelet denoising parameters for eddy current detection signals under different noise interference.