云控制系统并行任务分配优化算法与并联控制

Parallel Task Assignment Optimization Algorithm and Parallel Control for Cloud Control Systems

利用Petri网模拟云控制系统的并行处理过程,引入并行处理系统的时钟周期、吞吐率和任务完成时间性能指标,运用极大–加代数方法分析和优化云控制系统并行处理性能.采用子过程细分的优化方式,通过求解一类最优控制问题,设计并行任务分配优化方案,以保证任务完成时间最短,并给出计算最短任务完成时间的有效算法.同时,采用重复设置多套瓶颈段并联的方式提高并行处理能力,并运用Petri网实现瓶颈子过程的并联控制,且给出并联控制在协同云控制系统中的一个应用.

We use Petri nets to simulate the parallel processing arising in cloud control systems. We introduce perfor- mance indexes called clock period, through-put rate and task completion time, and use the max-plus algebra to analyze and optimize the parallel processing performance of cloud control systems. By using the method of segmenting sub-processes and solving the optimal control problem, we design tile optimization scheme for parallel task assignment to minimize tile completion time, and develop an effective algorithm to compute such a ininimum time. Computer performances can be improved through parlel connection of bottle-neck roads. We use the Petri nets to realize the parallel control of bottle-neck sub-processes and present an application of the parallel control in cooperative cloud control systems.