Dynamic Programming and Approximation in Policy Space for Optimal Feedback Control of Nonlinear Systems

摘要

Dynamic programming is used to obtain optimal feedback control policies for time-invariant nonlinear systems with fixed final time. The policy is obtained in the form of a table of the control variable as a function of the state of the system and the time remaining in the process. The recursive relation necessary to obtain the control table is derived for nonlinear systems in general and then used to calculate the optimal temperature schedules for a first order and for a second order batch reactor system. It is shown that approximation in policy space is not needed for the numerical solution of the problem. (Author)