| Tractor automatic navigation technology has been widely used in land tillage,sowing,fertilization,plant protection and harvest,and other agricultural production links.Practice at home and abroad has proved that adopting tractor automatic navigation technology can significantly improve the operational efficiency and quality of agricultural production,reduce labor intensity and production cost,and is one of the key technologies for agricultural automation and intelligent production.Key Laboratory of Key Technology on Agricultural Machine and Equipment of South China Agricultural University of Ministry of Education successfully developed the GNSS based automatic navigation system.The operation accuracy of the navigation system reaches 2.5cm,and the operation is simple and reliable,and the system has been carried on the large-scale promotion application in Xin Jiang region.The traditional linkage angle sensor is difficult to be installed,the linkage is easy to be loose and deformed,which affects the stability of the navigation system and increases the maintenance cost of the navigation system.For this reason,this paper proposes the front wheel steering angle measurement scheme based on the information fusion of GNSS/MEMS gyroscope,which replaces the traditional connecting rod angle sensor measurement scheme.Field experiments show that this scheme can track the linear path and satisfy the requirement of automatic navigation of farm machinery.Main research contents and achievements are as follows:(1)A gyro random drift Kalman filter based on AR model is designed.Based on the basic principle of time series analysis,the stationarity of the original drift data of CRS03 was analyzed,and was modeled based on the autoregressive model(AR model).The model order was determined by using AIC criteria,and the autoregressive coefficient was calculated by the least square method,and established the autoregressive mathematical model.Kalman filter is designed based on this model to filter the static random drift of CRS03 gyro.The experiment results show that the standard deviation of the original drift observation data before filtering is 0.38 ?/s,and the standard deviation after filtering is0.05?/s,the effect is obvious,indicating that this scheme can effectively inhibit the random drift of the gyro.(2)A Kalman filter based on GNSS/MEMS gyroscope information fusion was designed to calculate the steering angle of the front wheel.MEMS gyroscope is installed on the left steering axle of the agricultural engine and directly under the vehicle seat.The difference between the output value of MEMS gyroscope on the wheel shaft and the output value of MEMS gyroscope on the vehicle body is calculated.The calculation results are integrated to obtain the steering angle.The Kalman filter was constructed,and the front wheel rotation angle calculated by the two-wheel agricultural machine kinematics model was used as the observation value of the Kalman filter to calibrate the gyro integral to calculate the zero deviation of the rotation angle.Simulation experiments show that the linear driving situations,angle measurement error of the mean is 0.12°,the curve driving situations,the front wheel steering angle measuring error of the mean is 0.25°,the filter filtering effect is good,meet the requirements of navigation system for precision of angle.(3)An adaptive Kalman filter based on angular acceleration invariant model is designed.In order to solve the problem of large angular rate noise in the course differential calculation of double antennas,according to the practical situation that the course angle changes less when the automatic navigation of tractor is driven in straight line,a method of the value filtering of the course differential calculation by Kalman filter based on the constant angular acceleration model is proposed.Considering the mismatch of the model in the process of online automatic navigation,in this study,the mathematical relationship between position deviation and covariance matrix Q was established,and an adaptive Kalman filter was designed based on the position deviation size variable covariance matrix Q.Simulation results show that the average angular rate error is reduced from0.91 ?/ s before filtering to 0.59 ?/ s after filtering when driving in a straight line,and the average angular rate error is reduced from 1.57 ?/ s before filtering to 1.38 ?/ s after filtering when driving in a curve.The filter is effective in removing noise and the filtering effect meets the measurement demand of automobile body angle rate.(4)Field experiment was carried out on the Lovol M904-D tractor.Considering different application scenarios and cost control of automatic navigation of farm machinery,two test schemes of steering angle measurement of front wheel were designed.The first scheme is the measurement scheme based on the position information of the GNSS board card,the MEMS gyroscope of the front wheel steering shaft and the MEMS gyroscope information of the agricultural vehicle body.The second scheme is the measurement scheme based on the position and heading information of the GNSS board card and the MEMS gyroscope of the front wheel steering shaft.When the online test was conducted at a distance of 2.5m and1.5m.The first scheme angle measurement error of the mean is 0.4°and 0.26°respectively,online process time is 17.9 s and 16.2 s respectively.The second scheme angle measurement error of the mean is 1.13°and 0.87°respectively,online process time is 29.4 s and 23.5 s respectively.Under the linear navigation test of 4km/h,7km/h and 10km/h,The first scheme angle measurement error of the mean is 0.31°and 0.35°and 0.31°respectively,horizontal position deviation from the mean is 1.2 cm,1.2 cm and 1.6 cm.The second scheme angle measurement error of the mean is 0.44°and 0.49°and 0.44°respectively,the horizontal position deviation from the mean is 1.8 cm,2.5 cm and 3.8 cm.The first scheme is stable and reliable with no vibration.The second scheme has a long time to go online,but it meets the needs of navigation operation within acceptable range. |
PDF Full Text Request