论文部分内容阅读
对离散时间大系统的最优控制问题，应用空间和时间分解，提出一个新的三层递阶控制的并行算法．平行变量尺度（PVM）法用作求解空间分解的松驰耦合子系统．其第二层协调变量用修正的牛顿方法校正，而低层更小子问题用推广微分动态规划（DDP）并行求解．
For the optimal control of discrete-time large-scale systems, a new parallel algorithm of three-level hierarchical control is proposed by applying space and time decomposition. The Parallel Variable Scale (PVM) method is used as a relaxation coupled subsystem to solve the spatial decomposition. The second-level coordination variables are corrected by a modified Newton method, while the lower-level sub-problems are solved in parallel by the use of differential-dynamic programming (DDP).