Adaptive sliding mode neural net control for missile autopilot

An adaptive sliding mode neural net control algorithm is developed for a cruising missile with highly nonlinear aerodynamics. An adaptive switching gain based on the Lyapunov stability theory is proposed to suppress the chattering in the steady state. The gain varies with respect to the distance from the sliding surface. The controller with the adaptive switching gain also has robust characteristics on system model error and unexpected changes in aerodynamic coefficients. A control command generation method which uses the estimated actuator characteristics is proposed. The proposed method mitigates the control performance degradation caused by actuator fault or model disagreement. The effectiveness of the proposed control algorithm is verified through numerical simulations.