摘要
研究任务调度优化系统问题。任务调度问题的主要难点在于复杂度太高,传统的基于任务调度Q学习算法更新收敛速度慢。针对协同工作中的任务调度实际问题,提出了一种基于模拟退火的改进的Q学习算法。算法首先建立任务调度目标模型,在分析了Q学习算法的基础上,通过引入模拟退火算法,同时结合贪婪策略,以及在状态空间上的筛选判断,并给出了任务调度的整个过程。仿真结果表明,与单一的Q学习任务调度算法相比,改进的算法显著地提高了收敛速度,缩短了执行时间。从而验证了改进算法的有效性。
The task of scheduling problems was studied.Q-learning algorithm based on the traditional learning algorithms for task scheduling has slow convergence speed,and the paper presented a simulated annealing-based Q-learning algorithm to improve the convergence speed.Task scheduling algorithm for object model was established,and the analysis of the Q-learning algorithm based on simulated annealing algorithm was introduced.Combined with the greedy strategy and the filtrating and determining in the state space,and the whole process of scheduling was given.Simulation results show that Q-learning tasks with a single scheduling algorithm significantly improves the convergence rate and shortens the execution time.
出处
《计算机仿真》
CSCD
北大核心
2011年第12期212-214,共3页
Computer Simulation
关键词
任务调度
学习策略
模拟退火算法
Task scheduling
Learning strategies
Simulated annealing algorithm
