基于改进互补集合经验模态分解的自适应小波熵阈值地震随机噪声压制算法

Adaptive Wavelet Entropy Threshold Seismic Random Noise Suppression Algorithm Based on Improved Complementary Ensemble Empirical Mode Decomposition

针对互补集合经验模态分解(complementary ensemble empirical mode decomposition,CEEMD)后不易有效区分有用信号和噪声的问题,以及传统小波去噪阈值选取的不足,提出基于改进CEEMD的自适应小波熵阈值地震随机噪声压制算法。将地震信号进行CEEMD后,基于互信息熵和互相关系数获取高频含噪本征模态函数(intrinsic mode function,IMF);对含噪IMF进行多尺度小波分解,将高频小波系数等分为若干区间计算各区间小波熵,在此基础上得到不同尺度的自适应阈值,同时设计了改进阈值函数进行小波阈值去噪。仿真实验中,去噪残差和频谱分析表明,算法能在保留有用信号的同时有效去除随机噪声,实现保幅去噪。实际地震资料处理表明,相比其他去噪算法,算法能有效提高信噪比(signal-to-noise ratio,SNR)1 dB以上,降低均方误差(root mean square error,RMSE),具有良好的去噪能力。

To solve the problem that it is difficult to distinguish useful signal from noise effectively after complementary ensemble empirical mode decomposition(CEEMD),and to overcome the deficiency of threshold selection in traditional wavelet de-noising algorithm,an adaptive wavelet entropy threshold seismic random noise suppression algorithm based on improved CEEMD was proposed.First,the seismic signal was decomposed by CEEMD,then the noised intrinsic mode functions(IMF)with high frequency were obtained according to the mutual information entropy and cross-correlation coefficient.After that,the noised IMFs were decomposed by multi-scale wavelet,and the high frequency wavelet coefficients were divided into several intervals to calculate the wavelet entropy of each interval.Then adaptive thresholds of different wavelet scales were obtained based on the calculated wavelet entropy.At the same time,an improved threshold function was designed for wavelet threshold de-noising.In the simulation experiment,the residual noise and spectrum analysis show that the proposed algorithm can remove random noise effectively while retaining useful signal and achieve preserved-amplitude de-noising.Actual seismic data processing show that,compared with other de-noising algorithms,the proposed algorithm can improve the signal-to-noise ratio(SNR)effectively by more than 1 dB and reduce root mean square error(RMSE),which have good de-noising ability.