Abstract: A parallel hybrid genetic algorithm framework is proposed aiming at improving the computing efficiency and the solution quality.This algorithm framework consists of three parts which are genetic algorithm，niche operation and simplex search.The first two parts，genetic algorithm and niche operation，are executed in serial way and the other is in distributed parallel way.The distributed parallel computing environment is constructed with four computers connected by a switch.A dynamic task scheduling solution is introduced into the proposed algorithm.The experimental result of a classical engineering example demonstrates the validity of the new algorithm and good speedup and computing efficiency in the distributed parallel computing environment.

Key words: genetic algorithm, niche, simplex, distributed parallel computing, task scheduling

摘要： 为提高混合遗传算法的计算效率和求解质量，提出一个并行混合遗传算法框架。该框架主要由遗传算法、小生境操作和单纯形3部分组成，遗传算法和小生境操作采用串行执行方式，单纯形采用分布式并行执行方式。分布式并行计算环境由4台计算机通过交换机连接构成，并设计了一个动态任务调度方案。一个典型工程算例验证了新算法的有效性，并且在分布式并行环境下取得了较好的加速比和并行效率。

关键词: 遗传算法, 小生境, 单纯形, 分布式并行计算, 任务调度