Closed-Loop Subspace Identification with Prior Information

Adding prior information to open-loop subspace identification has been proven to be effective in obtaining state-space models with improved accuracy. The purpose of this study is to improve the accuracy of close-loop subspace identification by using prior information. The proposed method uses a parsimonious model formulation to remove the correlation between future input and past innovation, which is the significant challenge in closed-loop subspace identification. In addition, every row is considered as an optimal multi-step ahead predictor and the constrained least square (CLS) approach is used to incorporate prior information. The mathematical analysis and simulation results verify the feasibility and effectiveness of the proposed method.