大规模时滞系统的动态规划模型与优化算法  

DYNAMIC PROGRAMMING MODEL OF LARGE-SCALE SYSTEM WITH DELAY AND ITS OPTIMIZATION ALGORITHM

作　　者：段鹰[1] 段文泽[2] 

机构地区：[1]重庆大学机械工程学院,重庆400044 [2]重庆大学电气工程学院,重庆400044

出　　处：《机械工程学报》2007年第4期217-223,共7页Journal of Mechanical Engineering

摘　　要：工业工程中的一些大系统往往具有时滞效应,其最优决策一直是个难题。基于生产计划和生产调度的优化研究,归纳出一类大规模时滞系统的动态规划模型,研究其优化算法。提出条件最优化原理,使动态规划推广到时滞系统的优化中。推导时滞型微分动态规划的递推算法公式,在一定条件下证明了其收敛性。将这一算法用于水厂最优递阶控制的协调级,仿真验证它的可行性和有效性,获得的结果对于水厂节能具有重要意义。所提出的模型与算法可用于更为广泛的场合。On the basis of the optimization study of the production scheme and control, the dynamic programming model of a kind of large-scale system with delay is summarized and it＇s optimization algorithm is studied. The principle of optimality under condition is proposed, so the dynamic programming is extended to the optimization of the system with delay. The recursive algorithm of delay dynamic differential programming is presented and it＇s convergence under some conditions is proved. This algorithm is used in the coordinate-lever of the optimal hierarchical control of the water plant and the simulation demonstrates its availability and efficiency. The result obtained is important for energy-saving. The model and the algorithm mentioned may be used in a more wide range.

关 键 词：大规模时滞动态系统 条件最优化原理 时滞型微分动态规划 最优递阶控制 

分 类 号：O221.3[理学—运筹学与控制论]