【摘要】 在可重构系统任务调度过程中,配置预取可以有效隐藏任务的配置时间从而提高系统执行性能.然而调度算法需要额外的任务配置策略,这不但增加调度问题的复杂度,而且导致算法时间开销大,影响系统的实时性.为解决该问题,提出一种带有预生成策略的离散粒子群优化算法应用于任务调度问题中.首先,描述可重构系统任务调度问题模型,将该问题转化为最优化问题求解;其次,设计调度方案的编解码形式,将离散粒子群优化策略应用于调度问题中;最后,提出预生成策略提高算法的可靠性和收敛速度.实验结果相比自适应蚁群算法和混合遗传算法,求解质量分别提高13.2%和32.3%.该算法生成的调度方案能够满足系统要求,并有效提高调度方案质量.更多还原

【Abstract】 In the process of task scheduling in reconfigurable system,configuration prefetching can effectively hide the reconfigurable time to improve system performance. However,the scheduling algorithm needs an extra prefetching strategy. This not only lead to the complexity of the scheduling problem but increased the execution time. To solve this problem,we proposed a Particle Swarm Optimization algorithm with pre-generation strategy to solve task scheduling problems. First of all,the reconfigurable system module and task scheduling Model are described,transforming this problem into the problem of optimal solution. Second,we encode the scheduling solution and put forward a discrete Particle Swarm Optimization algorithm. Finally,we propose the pre-generation strategy to improve the reliability and the rate of convergence. The experimental results show that the performance increased by 13. 2% and 32. 3% compared with Adaptive Ant Colony Optimization and Genetic Algorithm. The proposed algorithm inherits low complexity of the heuristic method and has high reliability and quality compared with other algorithms.更多还原