基于GA的DCS中任务的容错优化调度

Optimal and Fault-Tolerant Scheduling for Tasks in Distributed Control System Based on GA

分布式控制系统(DCS)中的实时任务必须在其时限前完成,否则会出现灾难性后果,因此必须为DCS提供一定的容错能力。该文基于EDF算法和版本复制技术给出了DCS的容错调度算法。在此基础上采用启发式任务分配算法分配任务,通过遗传算法对基版本任务时限进行优化,以提高处理器的利用率。仿真结果表明该算法是有效的。

In Distributed Control System（DCS）, real-time tasks must be finished before their deadlines, otherwise the catastrophes occur. Therefore, fault-tolerance must be offered for DCS. Based on EDF algorithm and task duplication technique, fault-tolerant scheduling algorithm is presented for DCS. And a heuristic static algorithm for assigning tasks to processors is given. Genetic Algorithms（GA） is adopted to optimize the deadlines of primary copies in order to improve utility of processor. The simulation experiments show that the algorithm is effective.