基于混合微粒群算法的网格服务工作流调度

A grid service-workflow schedule using hybrid particle swarm

采用微粒群优化的种群搜索方式,融合了局部搜索和全局搜索,引入了模拟退火算法和遗传算法思想,利用模拟退火随机概率来避免陷入局部最优,提出了一种混合微粒群优化算法,以便更好地满足用户期望的服务质量,解决网格服务工作流调度问题.网格仿真试验结果显示:对于具有全局QoS约束条件的Web服务选择,在执行效率上混合微粒群优化算法明显优于其他混合遗传算法,可在较短时间内获得较好的解,是求解多目标网格服务工作流调度问题的有效方法.

Workflow schedule consisting of grid service is a NP problem. QoS-aware was introduced in grid workflow. Particle swarm optimization （PSO） is discussed, which combines local search and global search. An easily implemented hybrid particle swarm optimization algorithm （HPSOA） is presented for the multi-objective grid service-workflow scheduling problem by using simulated annealing （SA） and genetic algorithm. Experiment results show that this algorithm is available and better than some traditional hybrid genetic algorithms （HGA）, and that it is a viable and effective approach to the multi-objective grid service-workflow scheduling problem.