论文部分内容阅读
地震数据反演是一个典型的非线性逆问题。其目标函数都是多极值的函数，因此，传统的迭代优化方法常常会遇到局部收敛性的限制。本文提出的同伦神经优化理论（HNOT）及其算法（HNOA）能将非线性多极值目标函数较快地收敛于全局极值，是一种有效的反演方法。本文将该反演方法与相邻道互相关技术和层位信息约束有机地结合起来，实现了地震数据控制下的井资料高分辨率岩性参数联合反演。理论模型与实际资料的处理结果表明，本文提出的方法是可行的。
Seismic data inversion is a typical nonlinear inverse problem. The objective function is a multi-extremal function, therefore, the traditional iterative optimization methods often encounter the limitations of local convergence. HNOT and its algorithm (HNOA) proposed in this paper can converge the non-linear multi-polar objective function to the global maximum quickly, which is an effective inversion method. In this paper, the inversion method is combined with the adjacent channel cross-correlation and horizon information constraints to realize the joint inversion of high resolution lithologic parameters of well data under the control of seismic data. The theoretical model and actual data processing results show that the proposed method is feasible.