Research on UAV Time Delay Controller Based on Improved Smith Training Neural Network

摘要

In the UAV flight process, when subjecting to strong airflow, attitude adjustment is delayed, resulting in the poor flight stability. Therefore, an improved Smith UAV controller is proposed based on neural network training. And the establishment of UAV flight control system model with parameter uncertainties is taken in the environment of disturbances and modeling errors, the improved Smith neural network is used for training, attitude adjustment and time delay compensation of UAV flight. Given the steady-state error caused by time delay, the state feedback adaptive adjustment is used to adjust the actuator input for flight attitude values in strong airflow. The adjustment of the UAV center speed is carried out, the center of gravity is moved down. In multilayer neural network, the Smith training is used for approximation of multivariate function, reduce the delay of the UAV flight attitude adjustment to achieve optimal control. The simulation results show that the method has the advantages of less time delay for aircraft attitude adjustment, which has steady state tracking performance of attitude parameters, improving the stability of flight control. Besides, the real-time performance of control command receiving and sending is improved to ensure the stability of UAV flight.