Data-driven cascade control system: Response estimation and controller design

In this study, we propose a data-driven design method for a cascade control system with inner and outer control loops. First, the input-output response of a controlled plant, which varies with the controller parameters of a fixed-structure inner-outer control law, is estimated directly from open-loop input-output data. Next, based on the estimated response, the controller parameters are optimized to minimize the difference between a reference model with a controlled closed-loop system. In the proposed method, the response of a fictitious reference input, which varies with the controller parameters, is estimated, and then the closed-loop response is estimated. Therefore, a closed-loop input-output data is not required, and the controller parameters are determined directly from an open-loop input-output data. Furthermore, the time constant of a reference model is also optimized so as to reduce the control error. The proposed method is compared with both conventional single-loop and cascade data-driven methods by means of numerical examples.