关键词: 非线性多参数优化, 遗传算法, 并行计算

Abstract: The authors introduce the improvement strategies of each process stage systematically and comprehensively in genetic algorithm. Firstly, the authors create various initial colonies and introduce quinary coding strategy to improve searching speed. Then, they adopt random selection mechanism without reimbursement to prevent premature convergence. Twopoint crossover and multipoint mutation are adopted to extend searching range of model space and to keep individual diversity. In addition, they use generation gap technique to ensure the convergence. Finally, the authors bring forward the description of the ameliorated parallel genetic algorithm according to PCAM principle. Thus seismic attribute optimization of nonlinear multiparameter and multiextremum is achieved obviously by running the mended parallel genetic algorithm.

Key words: nonlinear multiparameter optimization, genetic algorithm, parallel computing