波阻抗反演中的全局寻优策略

针对常规基于模型的波阻抗反演方法严重依赖于初始模型的选择和易陷入局部最优等局限性，提出了一种新的全局寻优策略。该反演策略是在常规测井约束反演的框架中加入一定次数的遗传全局搜索机制。如果迭代过程一旦陷入局部最优解；就可以把已经形成的局部最优解作为寻优的新起点，在此基础上进行全局遗传寻优，搜索更优的反演结果。通过模型数值计算和实际资料处理，表明该反演策略具有跳出初始模型控制的优良性能和良好的实际反演效果。

A GLOBAL OPTIMIZATION STRATEGY FOR IMPEDANCE INVERSION

In this paper, the authors describe a new global optimization strategy to solve some problems such as the initial model sensitivity, frequent local traps for the conventional impedance inversion based on model in the method searching model based on genetic algorithm is added into the calculating flow of logging constraint inversion. When the searching process falls into local optimum in the iteration, a genetic algorithm is used to search for the optimal inversion result using the local optimum as the new initial value. The results of synthetic and real data demonstrate that the method is efficient for escaping from the control of initial model and getting the satisfying high precision inversion results.