基于进化规划的多Agent系统任务调度

Agent任务调度是多Agent系统研究的重要内容之一.调度方法直接影响调度方案的优劣与否和系统的执行效率.进化规划是近年来兴起的一种进化计算方法,具有对实数直接操作及全局寻优能力.将之用于Agent任务调度,建立了Agent任务调度模型,设计了进化规划调度算法.采取多个体竞争策略有效地解决了进化规划的早熟问题.实例验证了这种方法的可行性及性能,进一步的研究是在资源及优先度限制的条件下,如何用进化规划算法解决Agent任务调度问题.

Scheduling of multi-agent tasks based on evolutionary programming

Scheduling of agent tasks is one of important research on multi-agent system. Scheduling strategy influences directly on dispatch scheme which is excellent or not and system efficiency. Evolutionary programming is an evolutionary computation method developed recently years, it has abilities of operating to real directly and searching optimal scheme in global. Using it in scheduling ofagent tasks, agent scheduling model is created and evolutionary programming algorithm is designed. Strategy of multi-individual competition solves the problem of premature convergence validly. Scheduling instance indicates its feasibility and performance, all procedures runs successfully in Matlab 6.1, Further research is how to solve the problem of agent task scheduling with evolutionary programming constrained resource and priority.