A Feasible Control Strategy for LQG Control Problem with Parameter and Structure Uncertainties

Abstract

Previous efforts have mainly been made on Linear Quadratic Gaussian (LQG) control problem with parameter uncertainties. However, in practice the change of system environment and parameters usually leads to the change of system structure. On the basis of a receding horizon strategy forth LQG control problem with unknown parameters, this paper provides a solution to LQG control problems which involve both parameter and structure uncertainties. When system parameters are estimated and updated gradually, this method considers the impact of the change of system structure on the performance index. It realizes parameters estimation based on posterior probability by Bayesian theorem, eliminates the correlation between system structures by changing the weighted symmetric matrices, obtains control gain minimizing the performance index and learns the future information at the same time. Finally, simulation results illustrate the effectiveness and accuracy of the proposed method.