大系统优化控制中的非光滑函数优化问题

Nonsmooth Optimization in Optimizing Control for Large Scale Systems

分析了大系统递阶-协调控制方法之一——关联预估方法中的协调级非光滑函数最优化问题,确定了相应的目标函数对协调变量的子梯度,从减少优化控制对系统干预次数的角度,提出了一种基于子梯度的最速下降的优化协调策略。数字仿真结果表明了本算法的优越性。

Nonsmooth optimization in the coordination level is the main problem of interactive prediction coordination in hierarchical control systems. In this paper, attention is paid to the determination of the subgradient of the overall cost function with respect to coordination variables. A coordination strategy based on the subgradient steep decent method is developed for hierarchical optimizing control algorithm. Simulation results demonstrate that the number of coordination times and of the intervention times on the real processes are extremely reduced.