农田作业车的RBF-PID横向路径跟踪控制研究OA北大核心

In response to the demand for path tracking of agricultural vehicles in large agricultural environments under the background of agricultural modernization,the traditional vehicle lateral path tracking algorithm based on incremental PID was optimized using RBF neural network self-tuning PID control parameters.A 3-6-1 structure RBF neural network was established,with target path position,actual driving position,and previous cycle steering wheel angle as input variables,and 6 neurons were set in the hidden layer.The output was the steering wheel angle,and the PID parameters were adjus-ted in real-time using the gradient descent method.Finally,CarSim/Simulink was used to model and simulate a farmland soil sampling vehicle modified from the John Deere 825i Gator model.Under a low speed(10 km/h)U-shaped(serpen-tine)path,the average error was 3.89 cm,the maximum error was 16.61 cm,and the standard deviation was 5.99 cm.The tracking effect was superior to traditional incremental PID control,with good robustness,and can meet the operational requirements of common agricultural vehicle path tracking conditions.