The Key Technology Of Auxiliary Navigation System For Crawler Rice Combine Harvester

Rice is the first cereal crop in China.Crawler combine harvester is the major equipment for rice harvesting.In order to balance harvested quality and operation efficiency,the forward direction,speed,header height,wheel positions and working parts operating parameters of combine harvester need to be adjusted according to crop growth condition during harvesting in real time.Combine harvester need to be operated by skilled driver with heavy laboring intensity.The navigation system can plan and track working path autonomously,improving harvest efficiency and quality,reducing labor intensity,extending working time.To improve the adaptability of navigation system to crawler rice combine harvester,aiming at“full swath operation”requirement and combining with the environmental characteristics of paddy soil which moisture content is high and bearing capacity is weak,the research on key technology of auxiliary navigation system for crawler rice combine harvester was carried out.It mainly included navigation information acquisition and processing method,rice pending harvest area visual recognition and harvesting path extraction method,tracing algorithm and control strategy,auxiliary navigation system actuator.The key technologies were analyzed and verified by experiments.The research in more detail including:（1）Based on comprehensively and systematically comparative analysis of the research progress and development status of agricultural equipment positioning and navigation technology in the domestic and overseas,the main techniques and devices in crawler rice harvester auxiliary navigation system were clarified,including navigation information acquisition and data processing technology,pending harvest area identification and working path extraction technology,harvesting path tracking control technology and end effector mechanism.（2）The basic theory of crawler combine harvester field driving and influential factors in paddy field were analyzed.The theoretical steering parameters correction model of crawler combine harvester in paddy field was constructed.（1）The mechanical state analysis model of crawler-soil coupling system and crawler combine harvester steering kinematics analysis model on soil were established based on ground mechanics and kinematical theory.Soil physical parameters,crawler combine harvester structure parameters and crawler kinematic parameters were main factors that affect driving condition in soil.（2）The physical and mechanical properties of paddy field soil at harvest time were determined:moisture content 23.8～31.2%,plastic limit 25.9～27.8%,liquid limit34.8～36.9%,bulk density 1.35～1.68g·cm^-3 and porosity 18.3～26.4%.The measurement results showed that the soil had high moisture content,large porosity,loose texture and in a plastic state.The soil tend to deformation under external force,causing relative displacement between crawler and soil,thus driving state of crawler combine harvester was affected.（3）Crawler combine harvester paddy field steering kinematic characteristic test was carried out.The results showed that in paddy field,crawler combine harvester actual steering radius was greater than theoretical value,actual steering angular velocity was less than theoretical value.The actual steering radius change trend was contrary to its theoretical value,actual steering angular velocity variation trend was same as its theoretical value but changing rate was less than theoretical value.（4）Based on the multiple function model fitting degree of steering radius correction coefficient,steering angular velocity correction coefficient,and forward speed were compared,the steering kinematics parameters correction model in paddy field was constructed.The relation between steering radius correction coefficient and the forward speed was quadratic function,simulated for Kρ=0.751v_c²-0.392v_c+1.819;The relationship between steering angular velocity correction coefficient and forward velocity was exponent function,simulated for K_ω=0.9187e^-0.745vc.The research could provide theoretical basis and data support for the design of crawler combine harvester auxiliary navigation system.（3）Crawler combine harvester auxiliary navigation system was designed,and the main functions and implementation methods of auxiliary navigation system were determined.The“three units and one mechanism”overall structure of auxiliary navigation system was built according to the functional requirements,which could meet the practical demand of rice harvest of crawler combine.（1）The overall structure of crawler rice combine harvester auxiliary navigation system was designed,which include navigation information acquisition and processing unit,harvest image processing and working path extraction unit,harvesting path tracking control unit,hydraulic steering end effector mechanism.（2）Aiming at single sensor was susceptible to field surface relief and crawler combine harvester vibration,the Extend Kalman Filter for fusing RTK-GNSS date and IMU data was designed,which could output locator data,heading data,forward speed and other key navigation information.The experiment showed that after fusing RTK-GNSS date and IMU data by Extend Kalman Filter,the standard deviation of the heading monitoring value was 0.039 rad and 0.045 rad was reduced compared with before,the fluctuation of monitoring data was reduced.The distance error calculated by locator data was 0.021 m and steering angle error calculated by heading data was 0.36°.The locator data and heading data was accuracy.（3）Series and parallel combination hydraulic steering system of crawler combine harvester was developed as auxiliary navigation system actuator,which could switch freely between manual steering and auxiliary navigation control.Take hydraulic steering cylinder thrust more than 600 N and response time less than 0.1 s as design specifications,the working parameters of main hydraulic components were determined.Experiments showed that the crawler combine harvester offset distance of 50 m straight driving was0.25 m,offset rate was 0.5%and line excursion offset was small.The steering response delay,angular velocity standard deviation and overshoot was 0.2 s,0.017 rad/s and less than 3.8%,respectively.The response delay was low and steering process was stable,which could meet the demand of crawler rice combine harvester auxiliary navigation.（4）A method for rice pending harvest area visual recognition and harvesting path extraction was proposed and applied.（1）The pre-processing method of rice harvest original image was designed.The inverse transformation correction model of distorted image was constructed.Based on maximum likelihood estimation and Levenberg-Marquardt algorithm,the camera internal parameter,external parameter and distortion parameter could be calibrated quickly in field.Two-dimension gaussian filter with 5×5 pixel rectangular template was designed to reduce image noise by pixel convolution operation within the window.Experiments showed that the average pixel error of planar target image feature point recognition was less than 0.17 pixels,focal length calibration error was less than 0.34 mm,the calibration results of camera parameters were accurate and could be used to correct image distortion,completing rice harvest image pre-preprocessing.（2）The parameters distribution of rice harvest images in HSV,HSI and RGB color space models were analyzed and compared.A binarization synthesis threshold algorithm for rice harvest images was proposed considering the super red feature 2R-G-B and was used to divide rice pending harvest area.Based on expansive-corrosion morphological closed operation reconstruction,particle noise after segmentation was reduced and the boundary of the pending harvest area was enhanced.（3）Based on the image column pixels statistical characteristics analysis of harvested area and pending harvest area,ROI was decided dynamically to enhance harvesting path fitting efficiency.According to the correlation analysis between pixel row gray value function and step function,pending harvest area boundary points were decided.Cubic B-spline curves were used to fit boundary points as harvest target path.（4）The coordinate transformation matrix of pixel position and space position was constructed,the relational mapping between visual path and spatial path was realized.Experiments showed that the visual system distance recognition average error was 9.6mm and error rate was 1.92%,angle recognition average error was 0.77°and error rate was 2.7%,spatial distance and angle identification were accurate.（5）Pending harvest area boundary line extraction experiments for Zhonggeng798and Lindao20 under front light,backlighting,highlight and low light showed that,the boundary extraction error of Zhongjing798 was 23.9～38.7 mm,the error was minimum under highlight and maximum under backlighting;the boundary extraction error of Lindao20 was 38.9～55.8 mm,the error was minimum under backlighting and maximum under low light.The average processing time of a single frame was 38 ms,which was suitable for rice harvesting paths rapid extraction in various light environments.（5）The tracing algorithm and control strategy for crawler combine harvester were proposed.（1）The path deviation analysis model of crawler combine harvester was established,the relative position geometry between crawler combine and target was derived,and the position and orientation error state matrix of crawler combine harvester were constructed.The steering transient response characteristic of crawler combine harvester was studied,the relation between control period and response curve interval was analyzed and the influence of steering signal excitation time on response curve interval was explored.Forward viewing distance dynamic adjustment strategy was determined based on the principle that input-response linear interval was greater than 90%.（2）Crawler rice combine harvester tracking path discrete time recurrence equation was derived,arc-tangent tracking path model was constructed based on the equation.Simulation analysis showed that path rectification maximum overshoot,rise time and regulation time decreased by 44.2%,16.3%and 28.0%respectively compared with the pure pursuit model without considering the actual steering characteristics in paddy field.It was conductive to reduce the overshot of crawler combine harvester rectification,improve path tracking convergence speed and reduce control model error.（3）Fuzzy controller for crawler rice combine harvester path tracking was designed and the parameter influence on control effect was analyzed.Controller parameter optimizing tuning algorithm was built based on particle swarm optimization.Simulation analysis showed that the overshoot of steering rate deviation was reduced from 14.2%to4.6%,rising time decreased from 2.1 s to 1.7 s and adjustment time reduced from 5.6 s to2.1 s after particle swarm optimization tuning.The response speed and stability of crawler combine harvester path tracking controller were improved.（4）Crawler combine harvester steering feature recognition method based on LS-SVM was proposed,and the model between control signal duty ratio and actual steering rate was established by online regression,fixing duty ratio output by fuzzy controller.Regression model parameters 2 factors 3 level full factor experiment was carried out and the result showed that when penalty coefficient was 10 and kernel function parament was 20,the fitting correlation coefficient of the model to the test set data was 0.9608,root-mean-square error was 0.0040 and insensitive to abnormal value.The steering rate error compared expected value could be reduced to 0.29°in paddy field,which 0～55.7%was reduced compared with before.The adaptability of auxiliary navigation system to paddy field was improved by this method.（5）Auxiliary navigation control system path tracking experiments showed that,the straight-line tracking average deviation was 4.3～5.8 cm,maximum deviation was10.6～17.2 cm and root-mean-square deviation was 1.92～3.60 cm respectively within the forward speed range 2.16～3.60 km/h.Path rectification rise time,steady-state regulation time,maximum overshoot and average steady-state error was 7.5 s,14.7 s,14.8 cm and6.4 cm respectively under the experimental condition that initial deviation was 1 m and forward speed was 2.52 km/h.The proposed crawler combine harvester auxiliary navigation path tracking controller convergence was stable.The deviation correction and stability tracking of harvest path could be realized by the proposed control algorithm.（6）The mean swath deviation,maximum swath deviation and swath ratio were determined as the evaluation indexes of auxiliary navigation system.Field experiment of rice harvesting with crawler combine harvester was carried out,operation affect under different forward speed was compared.Results showed that when forward speed was in the range of 2.45～4.03 km/h,the mean swath was 1.98～2.05 m,mean swath deviation was 0.15～0.21 m,maximum swath deviation was 0.29～0.49 m,swath ratio was90.1～93.1%.With the increase of the forward speed,the mean and maximum swath deviation increased,and swath ratio decreased.The pending harvest boundary could be recognized by auxiliary navigation system and harvest path could be adjusted according to the boundary without miss harvesting,which could meet the requirement of rice harvesting.