Data-driven set-point control for nonlinear nonaffine systems

Optimal set-point control is important in maintaining good control performance of practical industrial processes. Considering the challenges when modeling a complex process, this work presents a data-driven optimal set-point control (DDOSC) scheme for nonlinear nonaffine systems bypassing modeling steps. The outer loop of the proposed DDOSC adopts an ideal nonlinear set-point control function which is theoretically achievable to obtain a perfect tracking. Then, a dynamic linearization (DL) is used to transfer the ideal nonlinear set-point control law into a linear parametric one such that it is implementable through a parameter updating law. To address the nonlinear nonaffine system within a data-driven framework, the DL method is again used to obtain its linear data model which is further updated by a parameter-adaptive law. With a proportional feedback controller in the inner loop, the convergence of the DDOSC method is shown. In addition, the results are further extended by considering a proportional-integral-derivative feedback controller in the inner loop. The simulation results on a numerical example and a car suspension system verify the effectiveness of the proposed DDOSC in improving the performance of the local feedback controller.