Research On Autonomous Flight Control Method Of Unmanned Helicopter

Unmanned helicopter has a broad application protect in both military and civil areas. Identifying the high precision mechanism model and studying its autonomous flight control become the key technologies of the unmanned helicopter. In this paper, a general small unmanned helicopter with single rotor and tail rotor is the research object. Its motion equation of all variables is established based on Newton’s second law and rigid body’s moment of momentum theorem. The nonlinear system identification method is studied, and the model of unmanned helicopter is identified by BP neural network method. Design nonlinear model predictive control algorithm based on a neural network model for unmanned helicopter’s attitude control. Design the actual flight test to verify the effectiveness of the proposed autonomous flight control method.Firstly, the unmanned helicopter and coordinate system are detailed described. Then make aerodynamic analysis of the unmanned helicopter, and establish its full equations of motion.Secondly, considering strongly coupled and nonlinear characters of unmanned helicopter, the nonlinear system identification method is studied. The unmanned helicopter’s model in hovering state is identified by BP neural network. As it will get the local optimal solution by BP neural network, we train the BP neural network by using genetic algorithm (GA).It is used mostly by multi-loop cascade control way to realize the autonomous flight control of unmanned helicopter. Attitude control as the inner loop can affect the velocity and position. Design the NN-MPC of unmanned helicopter in hovering state, and realize the attitude control.Select the training and validation sets from the actual flight data which have been pretreated by low-pass filter and interpolation. Design the model predictive controller based on BP neural network as the prediction model. Check the method’s feasibility by actual turning flight test in hovering state.