混杂生产系统变时域滚动优化生产控制

为降低求解生产优化控制策略的复杂性,提出了递阶变时域滚动优化生产控制策略.通过对考虑切换费用及随机故障情况的混杂生产系统模型分析,把原始命题简化为有限时域下随机动态规划问题,并给出了求解该动态规划问题的Bellman方程.利用准时制思想,在期望意义下把求解多维随机最优生产控制问题递阶为求解确定系统的最优生产控制子问题,并通过在一维方向上的变时域滚动优化来逼近原始命题最优解.仿真结果表明,该策略快速有效,避免了直接求解偏微分方程带来的困难,使得基于该策略的算法可行,提高了控制的精确性和实时性.

Varying receding-horizon based production control for hybrid production systems

To simplify the complexities of the production optimum control problem,a new policy composed of dynamic programming and varying receding-horizon optimization was initiated.After analyzing the model of the hybrid production system involving setup cost and stochastic failure-prone,the original problem was simplified into a stochastic dynamic programming problem over finite horizon,and the associated Bellman equation was also proposed.By introducing the idea of just-in-time,the stochastic optimal production control problem of multi-dimension vector was reduced to a limited sub-problem of the system under expectation meaning,and the solutions of the simplified problem were made close to that of the original proposition by varying receding-horizon optimization on one dimension.The simulation results show that the proposed policy can avoid solving partial differential equations directly,and makes the receding algorithm feasible,more accurate and realtime.