| Due to natural disasters(such as typhoons,hail,etc.)and improper management of the planting process,mature rice is prone to lodging.Rice lodging not only affects the harvesting efficiency of the combine harvester,but also increases harvesting losses.When a combine harvester harvests lodging rice,it is necessary to adjust the operating parameters such as header height,reel position,and operating speed in a timely manner based on the morphological characteristics of the lodging rice.In order to reduce header loss and realize adaptive adjustment of header working parameters during harvesting of lodging rice,this paper proposes a characterization method of lodging rice morphology features,constructs a recognition system of lodging rice morphology features based on machine vision,proposes a header height control strategy of combine harvester based on inverse linear quadratic form robust feedback linearization,and constructs a control model of reel position and rotation speed according to lodging rice morphology feature parameters,Realized adaptive adjustment of header working parameters when the combine harvester harvests collapsed rice,providing technical support for achieving unmanned driving of the combine harvester.The main research content and achievements of this article are as follows:(1)By analyzing different rice lodging scenarios,it is proposed to characterize the morphological characteristics of rice lodging using information such as lodging angle,lodging direction,lodging height difference,and lodging area.An online recognition system for rice lodging morphology features was constructed based on machine vision.The recognition system mainly consists of depth cameras,imaging devices,information processing units,parameter display units,etc.The image of lodging rice is obtained online through a binocular camera,and then imaged by the imaging device and transmitted to the information processing unit.Using Deeplab V3+to extract image feature attributes,construct a training dataset;Using Deeplab V3+to segment lodging areas,no lodging areas,and harvested areas,achieving lodging area judgment and visual output of results.In order to solve the problem that conventional two-dimensional images cannot obtain three-dimensional depth information,a low-cost depth camera imaging method is proposed to obtain detailed features of the lodging area.RGB color coding is used to achieve unified horizontal lodging information extraction.A rice lodging morphology feature recognition calculation model is established based on depth data information collection,and the lodging angle,lodging direction,and the difference in lodging height and lodging area information shall be sent to the display unit for display.(2)Analysis of the influencing factors of lodging crops on the working parameters of the combine harvester header.Starting from the influence of the position,speed,and operating speed of the combine harvester’s reel on the harvesting of lodging rice,this paper analyzes the force between the reel and the crop during harvesting,and provides the relationship between the reel speed and operating speed.The reel index was introduced to reveal the correlation between the radius,rotation speed,and operating speed of the reel.A model for adjusting the position and speed of the reel in response to lodging rice harvesting was established based on the influence of header parameters on the missed cutting of rice,which falls within the range of 5～60 ° in the clockwise direction.(3)Taking the rice lodging height difference as the adjustment basis of header height,this paper analyzes the physical model of the combine header under two degrees of freedom and three degrees of freedom,discusses the linearization method of the physical model,and puts forward the method of transforming the uncontrollable original system into an extended controllable affine system by redefining the state and state equation,and transforms the height control system of the combine header into a linear quadratic form problem,A header height control strategy based on inverse linear quadratic form robust feedback linearization is proposed,which makes the control system have feedback characteristics and realizes the adaptive control of the header height of the combine when harvesting rice lodging.(4)To build a hardware and software system for adaptive adjustment of the working parameters of the cutting table for inverted crop harvesting,the trained Deeplab V3+ model weights on the PC are deployed to NVIDIA’s Jetson Xavier NX platform,and the software design of the cutting table control system is completed using SIMATIC S7-1200 PLC platform.The measurement and control software mainly includes: rice image acquisition and imaging,inverted rice morphological feature recognition,original image and recognition result display,paddle wheel position and speed adjustment for inverted rice harvesting,cutting table height adaptive control program module,human-computer interaction interface,etc.(5)Conduct experimental research on the identification of the morphological characteristics of inverted rice and the regulation of the working parameters of the cutting table of the combine harvester.The developed device for identifying the morphological features of lodging rice and regulating the working parameters of the combine harvester header was integrated into the rice combine harvester.Field tests were conducted on the single functions of the control execution mechanisms of the combine harvester header height,reel speed,and reel position.The results showed that the maximum relative errors of header height,reel height,reel front and rear positions,and reel speed were 6.4%,2.8%,2.2% and 9.9%,respectively.A feature recognition experiment was conducted on different lodging rice regions,and the automatic recognition results were compared with manual measurement results.The results showed that the maximum relative errors of lodging angle,lodging height difference,and lodging area in different regions were 14.5%,8.6%,and 16.8%,respectively;An adaptive control experiment was conducted on the working parameters of the header in different areas of lodging rice with different lodging directions.From the results after harvesting,it can be seen that the stubble height can automatically change from the lodging area to the normal area.The stubble height when fully lodging is 10-12 cm,and the stubble height in the normal area can be automatically adjusted to the preset height of 30 cm.The position and speed of the reel can be automatically adjusted according to the lodging morphology,There has been basically no missed cutting phenomenon,proving that the system constructed in this article can perform adaptive harvesting under different lodging conditions,and verifying the effectiveness of the control method for the working parameters of the combine harvester based on the recognition of lodging rice morphology features. |
PDF Full Text Request