一类分布式控制系统的容错优化调度算法

在分析同步采样分布式控制系统基础上,给出了一种新的容错调度算法。该算法首先在一个周期内形成一个任务调度序列(包括基版本和副版本),然后采用启发式算法将任务分配倒各个处理器上,且同一任务的基版本和副版本分配到不同的处理器上。同一处理器上的任务按照其在调度序列中的先后次序执行。由于控制系统的性能与任务的采样周期和控制延迟有关系,因而采用以整数编码为基础遗传算法对调度序列进行优化。仿真实验表明,这种调度算法在保证实时任务容错可调度的情况下,可以极大地提高控制系统地的性能。

Fault-tolerant and Optimal Scheduling Algorithm for a Class of Distributed Control System

On the basis of analysis on the character of the synchronous sampling distributed control system, a novel fault-tolerant scheduling algorithm was given. Firstly, a scheduling sequence (including primary copies and backup copies) was designed in a sampling period. Then heuristic static algorithm for assigning tasks to processors has been adopted and the primary copy and backup copy of a real-time task are assigned to different processors. The tasks assigned to a processor are executed according to their orders in scheduling sequence. Because the performance of control system is influenced on by sampling period and control delay, genetic algorithm based on integer coding was applied to optimized scheduling sequence. Simulation result show the scheduling algorithm can guarantee that real-time tasks are fault-tolerant schedulable and the performance of control system is improved.