Abstract: The grid task scheduling is the typical NP（Non-Polynomial） difficulty problem，based on Particle Swarm Optimization，which is an intelligent algorithm，this paper proposes one kind of new task scheduling algorithm，the goal is completing all tasks as soon as possible.Firstly，the algorithm stochasticly produces a swarm of particles and then unceasingly iterates the particle’s position and speed and obtains the task scheduling plan which can be accepted，secondly chooses a superior solution which has the highest trusted degree，using the taboo search algorithm to achieve the optimist solution.The result of simulation experiment indicates that compared with the genetic algorithm the TTPSO algorithm is more suitable for solving big-scale grid task scheduling problem.

Key words: grid task scheduling, particle swarm optimization algorithm, taboo search algorithm, trusted degree

摘要： 网格任务调度是典型的NP（Non-Polynomial）难题，基于粒子群优化智能算法，提出一种新的任务调度算法，目标是使所有任务整体完成时间最小。算法首先随机产生一群粒子，然后对粒子的位置和速度不断迭代，获得可以接受的任务调度方案，并从中选择一组信任度最高的作为较优解，再利用禁忌搜索算法由较优解获得最优解。仿真实验结果表明，与遗传算法相比更适合于求解规模较大的网格任务调度问题。

关键词: 网格任务调度, 粒子群优化算法, 禁忌搜索算法, 信任度