Data-Driven Design for Model-Referenced Model-Free Controller

This paper presents the data-driven design of model-free control (MFC) based on an ultra-local model to realize model-matching. MFC can realize the desired system response even for nonlinear systems; however, the control performance relies on design parameters. The data-driven tuning approach has attracted attention because it provides optimized controller parameters without the system model to be controlled. In this paper, the control law of MR-MFC is derived by extending the traditional MFC based on the ultra-local model and introducing the data-driven design method to optimize the MFC parameters from single-experiment time-series data without repeated experiments and a system model to be controlled. Additionally, the processing cost is extremely low because the controller parameters can be obtained by using least square methods. The effectiveness of the proposed method was verified through simulations with a linear parameter varying (LPV) system.