非调制大容量气枪信号处理中反褶积方法的对比研究

Comparative Study of Deconvolution Methods in Signal Processing of Untuned Large-Volume Airgun

利用云南宾川主动源资料,对比分析了阻尼系数法、水准法、最小二乘法、时间域迭代法以及维纳法去除气枪信号震源响应的效果。结果表明:①在参数选择正确时,以上方法得到的波形差异不大。其中阻尼系数法、水准法参数选取难度最大,最小二乘法虽然能够自动找到最优解,但步长设置过小时,计算效率低,步长设置过大时,易出现假极值点。时间域迭代法及维纳法的参数对结果的影响相对较小。②对于高信噪比叠加信号,时间域迭代法得到的信号信噪比最高,计算时间对比其他方法增加不大,是一种较好的反褶积方法。对于单枪低信噪比信号,当只需要到时信息时,可使用时间域迭代法,但需波形信息时,从波形的可信度及计算效率上考虑,维纳法是更加折中的方法。

Using airgun data from Binchuan in Yunnan Province,we compare the calculation effects of damping factor method,water level method,least squares method,time domain iteration method and Wiener method on the removal of the time function. The results show that:① The waveforms obtained by the above methods have little difference when the parameters are correctly selected. The damping coefficient method and water leveling method are the most difficult to choose. The least squares deconvolution can automatically find the optimal solution,but the calculation efficiency is too low if the step size is set too small,and false extreme points are likely occurred when the step size is set too large. The parameters of the time domain iteration and Wiener method have relatively little influence on the results.② For high signal-to-noise ratio(SNR) stacked signals,the SNR obtained by the time domain iterative deconvolution method is the highest,and the computation time is within the accepted range,so it is a good deconvolution method. For single-gun low SNR signals,the time domain iterative method can be chosen when we just need to pick the arrival time. But when waveform information is needed,the Wiener method is more effective in terms of the reliability of the waveform and the calculation efficiency.