Abstract:

Genetic algorithm is not suitable in solving some complicated problems due to its shortcomings. Some improved strategies are adopted for genetic operators to ensure convergence and the effective search of the model space. Search boundaries are set up to stabilize the solutions. In order to improve computational efficiency, coarse-grained parallel genetic algorithm is used, which combines high-speed concurrency of parallel computer with inherent one of genetic algorithm, and select appropriate migrating topological architecture and migrating strategy. Flow chart is designed and detailed algorithm description of modified Parallel Genetic Algorithm(PGA) are offered. Prestack seismic inversion is made with it, which obtains good results.

Key words: nonlinear multiparameter optimization, genetic operators, Parallel Genetic Algorithm (PGA), prestack seismic inversion

摘要： 遗传算法在处理一些复杂问题时效果不理想。该文在保证算法收敛和最大限度地搜索模型空间的基础上，对遗传算子采取相应策略进行改进，并通过界约束增加解的稳定性。为了提高计算效率，采用并行遗传算法，将并行计算机的高速并行性和遗传算法固有的并行性相结合，选择合适的迁移拓扑结构和迁移策略，构建并行模型。给出了改进后并行遗传算法(PGA)的设计流程图及详细算法描述，在叠前地震反演的实际应用中，取得了良好的效果。

关键词: 非线性多参数优化, 遗传算子, 并行遗传算法, 叠前地震反演

CLC Number: 

• TP391