Lateral Automatic Landing Control System for Cooperating to Suppress Risk and Deviations Based on MPC and NN

The model predictive control (MPC) and neural network (NN) are utilized to establish an automatic carrier landing system (ACLS) guidance law in the lateral plane, which achieves the objective of eliminating the lateral landing risk and lateral state deviations, and compensating the deck motion. For this motivation, the nonlinear lateral kinetic function is transformed into a polytopic model based on the landing state deviations. The NN is used to construct the lateral deck motion compensation (DMC) so that the proposed method can track the dynamic lateral desired glide slope. The NN outputs the controlled variables, which compensate the aileron and rudder manipulations in the MPC algorithm. The previous risk model provides the current landing risk numerical value that are introduced into the MPC performance. The landing state deviations, landing risk, and deck motion is considered in the solution of several linear matrix inequalities (LMIs). The input and output constraints are transformed into LMIs to maintain the feasible of the lateral actuators of the aircraft. The off-line design strategy of weight matrices is adopted to enhance the calculation speed and robust. Finally, three simulation examples are executed on a semi-physical platform based on the original nonlinear landing model, which verify the excellent multiple constraints optimization performance of the proposed method.