为了解决基于启发式算法的资源分配和任务调度过程中由于没有考虑任务间的相互依赖关系而出现的任务死锁问题,提出了一种基于蚁群系统的改进算法.首先阐述了如何将分配调度问题映射到任务资源分配图的优化选择问题上和如何将信号量机制引入到最优任务资源分配图中来解决死锁问题.其次说明了基于蚁群系统如何利用网格信息素系统模型实现该算法,涉及任务资源分配图的构造,以及通过蚁群的正反馈和分布式并行计算机制优化任务资源分配图.最后模拟试验结果说明所提出的算法可以有效地解决网格中任务死锁问题. In order to solve the problem of deadlock of tasks due to the lack of consideration of the interdependencies between tasks in resource allocation and task scheduling based on heuristic algorithm, an improved algorithm based on ant colony system is proposed.First, The problem is mapped to the optimization selection problem of task resource allocation map and how to introduce the semaphore mechanism into the optimal task resource allocation map to solve the deadlock problem.Secondly, based on the ant colony system how to use the grid pheromone system model to achieve the Algorithm, which involves the construction of task resource allocation map, as well as the optimization of resource assignment map by ant colony positive feedback and distributed parallel computing mechanism.The simulation results show that the proposed algorithm can effectively solve the deadlock problem in grid.