Jointly design of data-driven 2-DOF controllers with BIBO stability based on response estimation

This research presents a new data-driven design approach for two-degree-of-freedom (2-DOF) control systems. First, obtain the initial input/output data from an experiment. Then, estimate the system response when updating the controller parameters using only the initial data set. Last, update the controller parameters by optimizing the cost function using the estimated data. The optimization problem explicitly considers the bounded-input bounded-output (BIBO) stability of the closed-loop system. The proposed method simultaneously designs the feedforward and feedback controllers by minimizing a time-domain performance index rather than model-matching. Specifically, the proposed controller design scheme does not require the reference model, and the simultaneous design algorithm is more straightforward than the traditional divide-step design approach. Moreover, our proposed approach can predict the system response before applying updated controllers to practical systems, avoiding machine wear or damage caused by inappropriate controllers, and saving on experimental costs. We additionally consider noise and address it by introducing a signal projection method. Finally, a three-tank liquid-level system simulation verifies the validity and feasibility of the proposed approach.