并行分布式遗传算法的研究

The research of parallel and distributed genetic algorithm

传统遗传算法在面对一些搜索空间巨大的复杂问题时,其表现往往难以令人满意。作者针对传统遗传算法解决高维多峰值问题时可能会出现的困难进行了分析,然后根据困难出现的原因,基于PVM设计了并行分布式遗传算法,并对适应度评估、交叉、变异算子做了一些改进,旨在加强算法的全局搜索能力,提高算法的收敛速度。为了验证算法多项措施的有效性,对一多峰函数在高维条件下进行多方面的测试,实验结果表明这几项措施是有效的。

In the face of complex problems with a large number of search spaces, traditional genetic algorithm＇s performance is often difficult to satisfactory. The author analyzes those possible difficulties for solving high-dimensional multimodal problems by traditional genetic algorithm. Then according to the cause that difficulties occur, we design parallel and distributed genetic algorithm based on PVM, and make some improvements to the evaluation method of fitness, the crossover and mutation operator. These improvements aim at strengthening the global search ability and improving the rate of convergence of the algorithm. In order to verify the measures＇ effectiveness of the algorithm, we take a test to a multi-modal functions in many aspects under the condition of high dimension. The experimental results show that the several measures are effective.