| Traditional modes of agricultural production are being reformed and developed based on precision agriculture with complete and accurate information on farmland and agricultural conditions.Crop growth and development of real-time information collection in which is crucial,spectral data as the qualitative and quantitative analysis of plant physiological and biochemical state,guiding plant cultivation,planting,field management,the basis of advanced analysis methods,but the hilly complex environment crop spectral information acquisition need to mobile acquisition equipment,low efficiency of information collection.Therefore,reducing the intensity of manual labor,improving the efficiency of data collection,and providing effective data support for the development of precision agriculture are practical issues.Motivated by this requirement,in this paper,we focus on omnidirectional self-balancing platforms and design an automated horizontal regulation platform for carrying crop information data acquisition devices in hilly areas.In this paper,we complete the con Fig.uration design,control system design,embedded programming,prototype production,experimental and spectroscopic data acquisition with the omnidirectional self-balancing platform,as well as the analysis,modeling and prediction of plant nitrogen content.The main line of research is as follows.(1)Referring to domestic and international studies on levelling platforms,stabilizing platforms,and spectral prediction of nitrogen content,we analyze and determine the design requirements of an omnidirectional self-balancing platform suitable for spectroscopic data acquisition on hills and mountains.The conversion relation between each coordinate system is derived and the relation between the adjustment angle and the vehicle offset angle is obtained.A parallel mechanism was designed,whose two rotational degrees of freedom were derived and verified using screw theory.The length of this component is then calculated by kinematic analysis to ensure the working angle range of the plateau.The load torque of the push-rod motor is selected and checked to ensure the load capacity of the platform;Finite element analysis is performed to ensure that components do not fail under maximum load.(2)We design a control system for an omnidirectional self-balancing platform and analyze the principles of the platform.The STM32F103 series chip was selected as the main control chip,the closed-loop stepper motor actuator was used as the driving element,and the MPU6050 was used as the attitude detection sensor.Based on microcontroller control,I2C communication protocol is used to communicate with pose sensors and DMP function is enabled to collect platform pose information.The closed-loop stepper motor is automatically controlled independently and a two-axis linear insertion is implemented to ensure the fastest leveling speed.A straight-line insert is made by driving a two-axis step-and-push motor to an equilibrium position.The attitude angle control is a large closed loop to ensure that the equilibrium position can be reliably reached,and the magnetic encoder feeds the position information to the actuator to ensure that the step size is a small closed loop to ensure control accuracy.(3)Prototypes of the platform’s mechanical structure and control system were produced on a trial basis,and indoor and outdoor tests were conducted.The laboratory tests used different angles and loads to test the angle,load,speed and accuracy of the platform.The maximum adjustment time of pitch angle is 8.94 seconds,and the maximum absolute error is 0.2°.The maximum adjustment time of roll angle is 8.42seconds,and the maximum absolute error is 0.27°.In the outdoor test,3BD-350 crawler power platform produced by Sampu Agricultural Machinery Co.,LTD was used as the carrier,and the self-balancing platform was carried out in situ rotation leveling test at the 8°slope of hilly and mountainous terrain,and the leveling test was carried out at the0-18°slope change.The maximum error of rolling angle in the process was-2°,and the average error was 0.02°.The maximum absolute error of pitch Angle is-2.77°,and the average error is-0.14°.During rotation,the maximum error of roll Angle is 1.41°,the average error is 0.41°,the maximum error of pitch Angle is-1.34°,the average error is-0.22°.The designed and manufactured platform was equipped with an imaging spectrometer for spectroscopic data acquisition,and 100 samples were obtained for nitrogen measurements.Black-and-white correction,region of interest labelling,and spectral data extraction were performed,and the nine processed spectral data were obtained using wavelet transform,multiple scattering correction,first-order differentiation,and other pre-processing methods.Dimension reduction algorithms such as UVE,SPA,CARS and GA were used,and finally Unscrambler X software was used to build the prediction model with PLSR algorithm and cross-validation method.The modeling path is SP-DWT-MSC-FD-CARS-PLSR with a correction set determination coefficient R~2of 0.7988.The prediction set determined the coefficients R~2,RMSEC and RMSEV to be 0.6355,2.0035 and 2.7304,respectively.The modeling path is the SP-MSC-FD-CARS-PLSR modeling approach with coefficient determination of 0.7583for the correction set R~2,0.704 for the prediction set R~2,RMSEC of 2.1964 and RMSEV of 2.4638.The effectiveness and availability of spectroscopic data collection equipment were verified. |
PDF Full Text Request