论文部分内容阅读
基于在反演过程中对初始模型依赖性强、易陷入局部极值等问题 ,本文引入小波分析 ,提出多尺度地震波形反演方法 ,从而将参数反演问题转化到小波域中重要系数优化问题 .利用多尺度之间的内在联系及小波域中重要系数的稀疏性 ,有效改进了局部极值、计算量等问题 .并对几种多尺度反演策略进行了比较讨论 .基于波动方程正演及褶积模型的两种反演方法的数值实例结果显示了本方法良好的效果 .
Based on the problems of strong dependence on the initial model and easy falling into local extremum in the inversion process, this paper introduces wavelet analysis and proposes a multi-scale seismic waveform inversion method, which transforms the parameter inversion problem into the important coefficient optimization problem in the wavelet domain Using the inherent relations among many scales and the sparsity of the important coefficients in the wavelet domain, the problems of local extremum and computation are effectively improved, and several multi-scale inversion strategies are compared and discussed. And numerical results of two inversion methods of convolution model show the good effect of this method.