Observer-based model-free adaptive sliding mode predictive control

This paper proposes a new observer-based model-free adaptive sliding mode predictive control method (MFASPC) for discrete-time nonlinear systems. This scheme first equates the discrete-time nonlinear system to a linear form using a data-driven compact form dynamic linearization (CFDL) technique, establishes a data model consisting of only the pseudo partial derivatives (PPD), input data and output data, designs adaptive observers to achieve the estimation of the unknown PPD. The controller design part uses integral sliding mode control (SMC) to ensure the system's robustness. In contrast, with its constraint characteristics, the model predictive control (MPC) replaces the traditional switching control of SMC. The closed-loop control quantities are obtained by solving a rolling optimization problem in the finite time domain to provide dynamic optimal control action. The theoretical derivation of the Lyapunov function is used to demonstrate the system's stability. In order to verify the effectiveness of the proposed algorithm, numerical simulations and Photovoltaic power generation system simulation experiments are conducted, respectively, and the results show that the proposed control algorithm has a very reliable tracking capability and control accuracy.