Research On Tractor Path Tracking Control Based On Adaptive Model Predictive Control

China is in the process of rapidly advancing from traditional agriculture to modern agriculture,and all aspects of modern agricultural production are in urgent need of intelligent upgrading.Agricultural intelligence has become the new direction of China’s agricultural development,the development of intelligent agriculture has become the inevitable path of national development.Unmanned tractors help to improve production efficiency,reduce labor intensity and ensure operational accuracy,in line with the development needs of modern agriculture in China.Path tracking control is one of the core technologies of unmanned tractor,so become the research hot spot in the field of agricultural intelligence.This paper took the Dongfanghong LF1104-C tractor as the test platform,took path tracking control as the research object,built the software and hardware platform of the control system,and realized the unmanned tractor.The main research contents are as follows.(1)In order to meet the requirements of speed monitoring accuracy for path tracking,the advantages and disadvantages of different kinds of speed sensors were analyzed in depth,and the multi-sensor information fusion speed measurement method was designed on the basis of comprehensive consideration of cost and practicality.The Extended Kalman Filter(EKF)algorithm was used to fuse the Global Navigation Satellite System(GNSS)system and Inertial Measurement Unit(IMU)data for accurate measurement of tractor operating speed,solving the limitations of single sensor speed measurement.The test results show that the method can effectively reduce the speed measurement error and also the speed measurement accuracy meets the requirements of field operations.(2)The path tracking control algorithm based on adaptive time-domain MPC was proposed for the problem that the model predictive control in fixed time domain cannot adapt to the complex operating conditions.Based on the current speed of the tractor and the reference path curvature,the optimized prediction time domain and control time domain are determined by a fuzzy control algorithm,which is combined with the MPC algorithm to control the tractor to track the reference path.A joint simulation platform was built using Casim and Simulink,and the simulation was verified.The simulation results show that the designed controller can guarantee the lateral error within 5 cm for different working conditions,and the mean value of the absolute lateral error does not exceed 1.5 cm;moreover,it does not require the controller to maintain large time domains,which improves the calculation speed.(3)The adaptive feedforward controller was designed to address the problem of slow upload speed when the path tracking error is large.Based on the improved particle swarm optimization(PSO)algorithm,the front wheel compensation rotation angle is determined according to the current speed of the tractor and the lateral error.The results of the simulation tests on the adaptive feedforward controller using the joint simulation platform show that the MPC controller with adaptive feedforward takes at least 18 %less time to track the convergence of the reference path compared to the MPC controller.(4)The Dongfanghong LF1104-C tractor was transformed into an electrified tractor and the hardware and software platform of the unmanned control system was built.For the field operation requirements,relevant test conditions were designed and real-vehicle field tests were conducted.From the test data it can be seen that: the average value of the absolute lateral error of the tractor path tracking is less than 3 cm;the lateral error of the straight line section can be kept within ±7 cm;the absolute lateral error of the cross-line curve section does not exceed 15 cm;compared with the MPC-based path tracking system,the path tracking system designed in this paper can improve the on-line speed of tractor tracking reference path by at least 12 %.The test results show that the designed path tracking controller can ensure that the tractor can accurately and quickly track the pre-set reference path.