地震波阻抗反演的蚁群算法实现

The ant colony optimization for the seismic impedance inversion

（1.中国石油大学（北京）CNPC物探重点实验室，北京 102249；2.中国石化科技部，北京 100029）

CNPC Key Lab in Geophysics, China Petroleum of University, Beijing 102249, China

波阻抗反演属于求解最优化问题,其优化目标函数可能包含多个在一定范围上的连续变量，传统的优化手段存在有些函数难以优化，容易陷入局部解，收敛速度较慢等问题。蚁群算法具有正反馈性、分布式计算和贪婪式启发搜索的特点，克服了传统优化算法的缺陷。通过分析蚁群算法的原理，提出了地震道非线性反演中蚁群算法的实现方法。利用模型对该方法进行了检验，在无噪情况下，反演结果与模型一致；在加入5%，10%和30%噪声的情况下，反演结果与模型的相似系数分别为98.76%，97.85%和86.42%。

For solving the non-linear problem in the seismic data inversion, the ant colony optimization is a new method with the characters of forward feedback, distributing computation and greedy enlighten search. The ant colony optimization is applied to the processing in the seismic impedance inversion. Under analysis the principle of the ant colony optimization, an improved ant colony optimization in solving the seismic data inversion problem is proposed. The algorithm integrates the three characters of the ant colony optimization. Analysis the result of the seismic model data inversion shows the algorithm is very effective in finding global optimizations, and it is a good algorithm for the seismic data non-linear inversion.