【摘要】 遗传算法作为通用而有效的全局搜索算法已在图像处理、自动控制等众多领域获得应用,但其计算量大、极耗计算资源,运行效率直接影响到复杂的非线性和多维空间寻优问题的求解效率。在分析OpenMP并行技术特点的基础上,针对主从式并行模型没有充分利用遗传算法内在并行性的问题,提出了一种改进的主从式并行遗传算法,并应用OpenMP编程模型在多核计算环境下实现。利用旅行商问题进行的实验表明,改进的并行遗传算法有更好的计算效率、扩展性,可在求解大规模TSP问题上有更广泛和高效的应用。更多还原

【Abstract】 As a universal and effective global searching algorithm,genetic algorithm has been applied in the field of image processing,automatic control and so on.However,it has huge workloads and consumes great computation resources,and its operating efficiency has a direct effects on solution efficiency of complicated nonlinear and multi-dimension problems.Based on the analysis of characteristics of OpenMP parallel technology,in view of the problem that master-slave parallel model doesn ’t make full use of the internal parallelism of genetic algorithms,this paper proposes a modified master-slave parallel genetic algorithm and implements it using OpenMP programming model under the multi-core computing environment.The experiment on travelling salesman problem shows that the modified parallel genetic algorithm has better computation efficiency and extensibility.Therefore,it can be applied to solve the large-scale TSP problem more widely and efficiently.更多还原