Approximation-based control of uncertain helicopter dynamics using MLNN with inputs saturation and saturation compensator

Practically the control of actuators is limited and saturations can occur, so we are interested to investigate the control of the unmanned helicopter dynamics with bounded control inputs, actuator saturation compensation scheme and a nonlinear saturated observer for estimate the velocities without measuring them. The altitude and yaw angle tracking with saturated torque inputs is considered for an unmanned model helicopter dynamic with the nonlinearities such as high model uncertainties and dynamic coupling. Multi-Layer Neural Network (MLNN) approximators have been employed to deal with the uncertainties and the saturation compensation designing without velocity measurements. The boundedness of the NN weights, the closed-loop system performance and Semi-Global Uniform Ultimate Boundedness (SGUUB) of tracking errors are guaranteed based on the Lyapunov stability synthesis. The designed observer- adaptive Multi-Layer Neural Network controller ensures that the altitude and the yaw angle track the reference signals under the input saturations. This paper completes the reference [4] by (1) considering the saturation problem of actuators, (2) Limit the error signals by utilizing the hyperbolic tangent function and (3) compensate the actuator saturation by MLNN controller

Keywords:

actuator saturation , output feedback control , model uncertainty , trajectory tracking , saturated observer