Abstract:

An adaptive neural network control method is proposed for a class of control systems with uncertain actuator nonlinearity. A model for the nonlinear actuator is developed which includes the characteristics of dead zone, backlash and “backlash-like” hysteresis. By combining the developed model and the Nussbaumgain technique, the problem of uncertain actuator nonlinearity is solved perfectly. The proposed scheme does not require the prior knowledge on the bounds of parameters of the motioned characteristics, and the slopes of dead zone can be time variant when dead zone nonlinearity is concerned. The adaptive compensation term is adopted to minify the influence of modeling error and external disturbance. Simulation results are presented to demonstrate the effectiveness of this method.