基于回卷恢复的数控系统实时容错调度策略

Fault tolerant real-time scheduling strategy for NC system based on rollback recovery

针对数控系统对时间确定性和可靠性的要求,建立混合关联任务集描述模型,提出了基于回卷恢复机制和最佳优先(BF)算法的容错调度策略.该策略对数控系统关键任务的执行过程进行周期性分段验证,通过保存正确的检查点状态,使得出现瞬时故障时任务可以从检查点开始继续执行,从而以较低的计算损失达到容错的效果.与传统的实时容错调度算法相比,基于回卷恢复机制的容错调度策略能很好地适用于数控系统多类型任务并存和任务间相互约束的特点,同时,通过降低错误恢复所需的时间开销,有效地提高了系统的资源利用率.该算法应用于数控系统的有效性在实例分析和算法模拟研究中得到验证.

A model for hybrid dependent task set was reliability of numerical control （NC） system, and a established to guarantee the time determinability and fault tolerant real-time scheduling strategy based on rollback recovery and best-first （BF） algorithm was proposed to remove system transient faults. The running result of critical task in NC system was verified by this fault tolerant strategy section by section, and every correct state of checkpoint Was saved to restore task from checkpoint with low cost when error was detected. Comparing to traditional fault tolerate scheduling algorithm, the strategy based on rollback recovery is applicable to schedule mixed and dependent tasks of NC system, and raises the level of resources utilization by reducing the cost of fault disposal. The validity of this proposed approach applied to NC system was verified in case study and algorithm simulation.