基于捕食搜索策略的网格任务调度算法

摘要

In order to solve the problem of grid task scheduling, we designed a grid task scheduling algorithm imitating animals' predatory search strategy (SAPS). First the algorithm determines the area to be searched,and divides the area into several sub-areas,then it searches the sub-areas in turn. During the search process,when a good solution is found in a certain sub-area, then the precise seatrch will be carried out, but if no good solution has been found, then another sub-area will be searched, until all the sub-areas are traversed. SAPS has considerable good abilities in global search and local search, it overcomes the drawbacks of the Min-min algorithm in purely pursuing the local optimum while lacking global consciousnees. Test results demonstrate that this algorithm can more effectively resolve the grid task scheduling problem.%为解决网格任务调度难题,设计了一种模仿动物捕食策略的网格任务调度算法SAPS(Scheduling Algorithm Based on Predatory Search).该算法首先确定待搜索区域,将待搜索区域划分为若干子区域,然后对子区域依次进行区域搜索,在搜索过程中如在某个子区域发现较优解,则对此子区域进行精密搜索,如未发现较优解,则转到下一个子区域,直至遍历所有子区域.SAPS算法具有较好的全局搜索和局部搜索的能力,克服了Min-min算法单纯追求局部最优而缺少全局意识的缺点.试验结果表明,该算法能更有效地解决网格任务调度问题.