Modeling and Fitting of Aircraft Fly-by-wire Control System Based on Improved BP Neural Network

In this paper, the modeling and fitting of the longitudinal and lateral fly-by-wire control systems of the flight parameter data are realized through the analysis of the control laws and manipulation response types of the aircraft fly-by-wire control system. Firstly, outliers are eliminated and corrected based on the polynomial fitting method, and the angular velocity, attitude angle, overload and other flight parameter data are digitally filtered by the median filter based on the least square method. Then the appropriate flight parameters are selected as the input and output, which are sent to the improved BP neural network based on genetic algorithm for training. Finally, the modeling and fitting of aircraft longitudinal and lateral fly-by-wire control systems are realized in the three flight stages of climb, cruise, and descent/landing.