Novel Data and Neural Network-Based Nonlinear Adaptive Switching Control Method

We propose an adaptive nonlinear control method for a discrete-time dynamical system. First, the nonlinear term is decomposed into a previous sampling instant term and an unknown increment term, which are determined using an intelligent estimation algorithm based on adaptive fuzzy neural networks. The problem of obtaining accurate input data due to the unknown current control signal in unmodeled dynamics using conventional estimation algorithms is addressed, and the conservativeness is reduced. Furthermore, historical data of the controlled plant are leveraged, and the data in the nonlinear term containing repeated estimation information are disregarded. Then, we apply the proposed decomposition method for the nonlinear term to design nonlinear switching controllers. One linear and two nonlinear adaptive controllers are designed, all with compensation of the nonlinear term at the previous sampling instant and increment estimation. These three adaptive controllers coordinately operate the plant by switching rules to guarantee the stability of the controlled plant and to improve the system performance. The stability and convergence of the system are analyzed and verified. Finally, simulation examples are used to verify the effectiveness of the proposed method and compare it with existing methods to confirm its superior performance.