基于WOA-VMD的瞬变电磁探测信号降噪方法

Denoising method of transient electromagnetic detection signal based on WOA-VMD algorithm

为有效降低工频干扰和随机噪声对矿山采空区瞬变电磁精准探测与定位的影响,基于鲸鱼优化算法(WOA)提出一种改进变分模态分解(VMD)的瞬变电磁信号噪声识别和分离的方法。首先,根据排列熵能够检测时间序列随机性和复杂程度的特性,选择分解结果的最小排列熵作为优化目标;然后,利用鲸鱼优化(WOA)算法对变分模态分解(VMD)算法中参数组合分解个数K和二次惩罚因子α进行全局搜优,根据寻优结果确定最佳组合(K,α),并利用参数优化后的变分模态分解算法对瞬变电磁信号进行自适应分解,得到一系列具有带限特性的本征模态函数(IMF)分量;最后,结合相关系数和IMF时域变化曲线共同判定噪声分量,选取有效分量重构得到降噪后的瞬变电磁信号。根据瞬变电磁信号特征构建仿真信号,并采用WOA-VMD进行降噪处理,与集成经验模态分解(EEMD)降噪结果进行对比。研究结果表明:在不同信噪比下,WOA-VMD算法能显著改善瞬变电磁中晚期信号质量,且能获得更高质量的瞬变电磁响应电动势数据。运用WOA-VMD算法对采空区进行现场试验,发现本文提出的降噪方法能有效去除信号中的工频干扰和随机噪声,有利于后期反演电阻率的生成从而确定采空区位置,提高了瞬变电磁法对矿山采空区的解释精度。

In order to effectively reduce the influence of power frequency interference and random noise on the precise detection and location of mine goafs by transient electromagnetic method(TEM), an improved variational mode decomposition(VMD) method was proposed for noise identification and separation of TEM signal based on whale optimization algorithm(WOA). Firstly, according to the characteristics that permutation entropy can detect the randomness and complexity of time series, the minimum permutation entropy of decomposition results was selected as the optimization objective. Secondly, WOA was used to globally optimize the parameter combination decomposition number K and the quadratic penalty factor α of VMD. According to the optimization result, the best parameter combination(K, α) was set, TEM signal was adaptively decomposed by the parameter-optimized VMD algorithm, and a series of band-limited intrinsic mode functions(IMF) were obtained. Finally, the correlation coefficient and time-domain change curve of IMFs were used to jointly determine the noise component, and the effective component was selected to reconstruct TEM denoised signal. The simulation signal was constructed according to the characteristics of transient electromagnetic signal, and the simulation signal was denoised by WOA-VMD. The denoising results were compared with the ensemble empirical mode decomposition(EEMD).The results show that WOA-VMD algorithm performs significantly in improving the quality of the middle and late stages of TEM signal and higher quality transient electromagnetic response electromotive force data can be obtained under different SNR conditions. The results of WOA-VMD algorithm which is used to conduct field test on the goaf show that the proposed method can effectively remove the power frequency interference and random noise in the signal. It is favorable for later inversion resistivity generation to determine goaf location and improve the interpretation accuracy of the transient electromagnetic method for mined-out areas of mines.