网格任务调度属于一个 NP完全问题,传统遗传算法很难将这一多对象问题求得最优解。通过生成节点性能评估函数及构建任务动态调度模型,经由函数参数权重值调节,可实现将多对象问题转化为单一对象问题,并对遗传算法的杂交算子和变异算子进行优化,以实现全局最优解的求解。%Grid task scheduling belongs to a np-complete problem,the traditional genetic algorithm (ga)is difficult to the obj ect more optimal solutions are obtained.In this paper,by generating node performance evaluation function and build tasks dynamic scheduling model,through the function parameter weights adjustment will be more objects into a single ob-j ect,and the cross operator and mutation operator of genetic algorithm optimization,implementation solution of the global optimal solution.
展开▼