Model-free subspace approach to NLQG controller design using bilinear model

Abstract

In this paper, Linear Quadratic Gaussian (LQG) controller extended to a class of nonlinear systems based on subspace matrices using bilinear model. LQG controller design based on subspace matrices provides directly from system input output data. Therefore it is more useful for systems that their models are not available. Since the most practical systems are nonlinear, LQG controller design based on linear subspace model is reflected to a weak control performance or even instability. To overcome this problem, LQG controller design based on bilinear subspace model is presented. Simulation results and comparison studies are provided to show the effectiveness of proposed method.