Adaptive control design for a class of nonlinear systems based on BP neural networks with scalers and saturators

Abstract

In this paper, a novel extended-neural-system (ENS) is presented for a class of single-input-single-output (SISO) uncertain nonlinear systems. The ENS consists of one Back Propagation (BP) neural network, two scalers and one saturator. By combining the ENS with the Lyapunov stability analysis, we propose an adaptive control scheme to guarantee that all the signals in the closed-loop system are uniformly ultimately bounded (UUB). In the steady control process, the weight parameters of BP neural network are trained in advance via offline data. Subsequently, one scaler parameter and the estimate values of the BP neural networks approximate accuracies are adjusted by the adaptive design laws. It means that the number of adaptive parameters is effectively reduced compared with the conventional weight parameters of BP neural network and the convergence rate of the system is improved. Finally, simulation example is performed to demonstrate and verify the effectiveness of the proposed scheme.