Filtering solution of nonlinear stochastic optimal control problem in discrete-time with model reality differences

Abstract

In this paper, we propose an efficient algorithm for solving a nonlinear stochastic optimal control problem in discrete-time, where the true filtered solution of the original optimal control problem is obtained through solving a linear model-based optimal control problem with adjustable parameters iteratively. The adjustments of these parameters are based on the differences between the real plant and the linear model that are measured. The main feature of the algorithm proposed is the integration of system optimization and parameter estimation in an interactive way so that the correct filtered solution of the original optimal control problem is obtained when the convergence is achieved. For illustration, a nonlinear continuous stirred reactor tank problem is studied. The simulation results obtained demonstrate the efficiency of the algorithm proposed.