Astragalus membranaceus harvester and control system thereof

Abstract

The invention relates to the technical field of agricultural machinery, in particular to an astragalus membranaceus harvester and a control system thereof, comprising a rack, a shovel plate integrated with a soil hardness sensor, a connecting frame with an industrial camera, and a screen frame driven by an eccentric wheel to vibrate and provided with a compression roller. The control system comprises six core modules: a digging execution and soil information sensing module controls digging and obtains soil hardness information; the mining parameter regulation and control module generates a mining depth and speed regulation and control instruction according to the information; the separation execution module controls the screen frame and the compression roller to perform soil and root separation; the visual information sensing module collects separation process images through an industrial camera; the visual analysis decision module analyzes the image to obtain a skin damage rate and a soil block residual state, and generates a regulation instruction of the screen frame and the compression roller; and the separation parameter regulation and control module receives and executes the instruction to realize self-adaptive adjustment. Closed-loop intelligent control over the harvesting process is achieved, root system damage is remarkably reduced, and the harvesting purity rate is increased.

Description

Technical Field

[0001] This invention relates to the field of agricultural machinery technology, and in particular to an Astragalus harvester and its control system. Background Technology

[0002] Astragalus, an important traditional Chinese medicine, is increasingly being cultivated on a large scale. Traditional harvesting of astragalus relies primarily on manual labor or single-function mechanized harvesting equipment. Existing technologies include some harvesting machinery for root crops, whose basic structure typically includes a digging shovel, conveyor chain, and vibrating screen, enabling preliminary soil-root separation. Some more advanced equipment is also attempting to integrate simple sensors to achieve basic stability in digging depth. However, the design focus of these existing technologies is generally concentrated on improving the efficiency of purely mechanical operations, lacking a systematic solution for the coordinated intelligent control of the core indicators affecting the commercial value and economic benefits of astragalus during harvesting—root integrity rate and harvest purity rate.

[0003] A significant drawback of existing technologies is that the soil-root separation process is an open-loop, fixed-parameter mechanical process that cannot be adaptively adjusted based on real-time operational results. Specifically, key parameters such as the frequency and amplitude of the vibrating screen and the pressure of the rollers remain constant after being preset before operation. However, the moisture, viscosity, and clumping state of the soil in the field, as well as the thickness and toughness of the Astragalus root system itself, all fluctuate in real time. This means that the fixed separation parameters, under varying working conditions, can easily cause large-area mechanical damage to the root epidermis or fail to effectively break up and separate adhering soil clods. Existing equipment lacks the ability to perceive and provide feedback on separation quality and operational results in real time, and therefore cannot autonomously optimize working parameters to consistently maintain the harvesting target of "low damage and high purity" under complex working conditions. Summary of the Invention

[0004] To overcome the above shortcomings, this invention provides an Astragalus harvester and its control system, which aims to improve the problem that existing equipment lacks the ability to perceive and provide feedback on separation quality and operation effect in real time, and therefore cannot autonomously optimize working parameters to maintain the harvesting target of low damage and high purity under complex working conditions.

[0005] In a first aspect, the present invention provides the following technical solution: an Astragalus harvester, comprising a frame, a shovel plate installed on one side of the frame, a soil hardness sensor integrated on the surface of the shovel plate, a drive seat symmetrically installed on the top of the frame, a connecting frame fixedly installed on the top surface of the drive seat, an industrial camera installed on the bottom surface of the connecting frame, symmetrical moving grooves opened on both sides of the frame, two sets of screen frames provided on the inner side of the frame, with both sides of the screen frames extending into the inner side of the moving grooves, screen meshes fixedly installed on the inner side of the two sets of screen frames, two sets of pressure rollers installed on the inner side of the frame, with the two sets of pressure rollers respectively positioned above the two sets of screen meshes, an eccentric wheel provided on the inner side of the moving grooves, and a connecting belt connected between the eccentric wheel and the drive seat via a pulley, and a telescopic rod provided on the inner side of the moving grooves, with a spring installed inside the telescopic rod.

[0006] Preferably, sliders are fixedly installed on the upper and lower sides of the screen frame, and a sliding groove is provided on the inner wall of the moving groove, with the sliders disposed inside the sliding groove.

[0007] Preferably, a fixing plate is fixedly installed on one side of the frame, a traction component is fixedly installed on the top surface of the fixing plate, support rods are fixedly installed on both sides of the frame, and a moving wheel is installed at the bottom end of the support rod.

[0008] Secondly, the present invention provides the following technical solution: a control system for an Astragalus membranaceus harvester, the system comprising the following modules: The excavation execution and soil information sensing module is used to control the excavation operation and, during the excavation operation, to acquire the soil hardness information of the operation area in real time through the soil hardness sensor. The excavation parameter control module is used to receive the soil hardness information, generate control commands for excavation depth and travel speed through a built-in closed-loop control algorithm, and send them to the excavation execution and soil information sensing module. The separation execution module is used to separate the soil roots of the harvested material by controlling the operation of the screen frame and pressure roller after the excavation operation is completed. The visual information perception module is used to collect image information of the harvested material in real time through an industrial camera deployed on the bottom surface of the connecting frame during the soil separation operation. The visual analysis and decision-making module is used to receive the image information, analyze the surface damage rate and soil residue status of the harvested material in real time through the built-in visual recognition model, and generate corresponding screen frame working parameter control instructions and pressure roller control instructions based on the analysis results. The separation parameter control module is used to send the received screen frame working parameter control command and the pressure control command of the pressure roller to the separation execution module respectively, so as to control the operation of the screen frame and the pressure roller to perform corresponding adjustments.

[0009] Preferably, the workflow of the excavation execution and soil information sensing module includes: Receive an external operation start command and control the shovel to descend to the initial working depth to start the excavation movement; During the excavation process, the soil hardness sensor integrated on the surface of the shovel plate continuously collects the original soil hardness signals of the working area in front of and below the shovel plate at a preset sampling frequency. The collected raw soil hardness signal is filtered and converted from analog to digital to obtain real-time digital soil hardness information. The digitized soil hardness information is sent to the excavation parameter control module at a preset communication cycle. While continuing the excavation operation, repeat the above steps in a loop until an external operation stop command is received.

[0010] Preferably, the workflow of the mining parameter control module includes: Receive soil hardness information sent in real time by the excavation execution and soil information sensing module; The received soil hardness information is compared with the preset soil hardness threshold range in real time. When the soil hardness information is lower than the preset lower limit threshold of soil hardness, a first control command is generated. The first control command includes commands to increase the travel speed and maintain the current excavation depth. When the soil hardness information is higher than the preset upper limit threshold of soil hardness, a second control command is generated. The second control command includes commands to reduce the travel speed and increase the digging depth. When the soil hardness information is within the preset soil hardness threshold range, a third control instruction is generated, which includes an instruction to maintain the current travel speed and the current digging depth. The generated first control command, second control command, or third control command is sent to the excavation execution and soil information sensing module to control the execution of the excavation operation.

[0011] Preferably, the workflow of the separate execution module includes: When the excavation execution and soil information sensing module completes the excavation work in the current area, the separation execution module receives a start command; The separation execution module drives the first drive device installed in the drive seat to operate. The first drive device drives the eccentric wheel to rotate through the connecting belt. The rotational motion of the eccentric wheel is converted into the reciprocating vibration of the screen frame along the moving groove. While the screen frame is reciprocating, the separation execution module drives the second drive device installed inside the frame to operate, and the second drive device drives the two sets of pressure rollers to rotate in opposite directions. The harvested material, conveyed to the screen frame by the digging shovel, first undergoes the rotational crushing of the pressure roller, and then falls onto the surface of the screen mesh, which is in a reciprocating vibration state. The harvested material is thrown up, tumbled and conveyed forward under the reciprocating vibration of the screen, while soil particles are separated and fall through the mesh of the screen. The separation execution module continues to execute the above steps until the current batch of harvested material completes soil-root separation and leaves the screen, or receives an external stop command.

[0012] Preferably, the workflow of the visual information perception module includes: When the separation execution module starts the operation of the screen frame and pressure roller, the visual information perception module simultaneously triggers the industrial camera deployed on the bottom of the connecting frame to enter the working state. During the soil separation operation, the industrial camera continuously captures images of the harvested material below the screen surface at a preset frame rate to obtain the original image sequence. The original image sequence is preprocessed to obtain a stable image to be analyzed; The image to be analyzed is marked with a timestamp and the corresponding working status information of the sieve frame, and packaged to form the image information of the harvested item; The packaged image information is sent to the visual analysis and decision-making module in real time. The visual information perception module continues to execute the above steps until the separate execution module stops operating.

[0013] Preferably, the workflow of the visual analysis decision module includes: Receive image information sent in real time by the visual information perception module, and process the image to be analyzed in the image information; The harvest outline region and soil clod impurity region in the image to be analyzed are identified and segmented. Within the harvest outline region, the visual recognition model calculates the ratio of the area of ​​damaged pixels to the total area of ​​pixels on the surface as the real-time surface damage rate. Within the soil clod impurity area, the visual recognition model calculates the total projected area of ​​the residual soil clods as a real-time quantitative indicator of the residual soil clod status. The calculated skin damage rate is compared with a preset damage rate threshold, and the quantified soil clod residual state is compared with a preset impurity rate threshold. Based on the comparison results, the visual analysis and decision-making module generates corresponding screen frame working parameter control instructions and pressure roller pressure control instructions.

[0014] Preferably, the workflow of the separation parameter control module includes: The system receives the screen frame working parameter adjustment command and the pressure roller pressure adjustment command sent by the vision analysis and decision module, and parses the screen frame working parameter adjustment command to obtain the target adjustment value of the screen frame vibration frequency. Based on the target adjustment value of the vibration frequency of the screen frame, a first control signal for controlling the rotational speed of the first drive device is calculated and generated; The pressure control command for the pressure roller is analyzed to obtain the target adjustment value of the pressure roller pressure; Based on the target adjustment value of the pressure roller pressure, a second control signal is calculated and generated for controlling the output torque or hydraulic pressure of the second drive device; The first control signal and the second control signal are respectively sent to the first drive device and the second drive device; The first drive device responds to the first control signal, changes the rotation speed, and drives the vibration frequency of the screen frame to the target value through the connecting belt and the eccentric wheel; The second drive device responds to the second control signal by changing the output torque or hydraulic pressure to drive the pressure roller to adjust the pressure to the target value.

[0015] The present invention has the following beneficial effects: 1. In this invention, through the real-time quality monitoring closed loop composed of the visual information perception module and the visual analysis decision module, the system can immediately detect an increase in the surface damage rate or excessive soil residue, and automatically adjust the vibration parameters of the screen frame and the pressure of the pressure roller. This enables the machine to dynamically adapt to changes in soil and crop conditions, transforming the separation process from fixed parameters to target-oriented, thus solving the problems of high damage rate and unstable impurity content in traditional equipment.

[0016] 2. In this invention, through the excavation execution, soil information perception module and excavation parameter control module forming an adaptive closed loop, the machine can optimize its travel speed and excavation depth in real time according to the soil hardness ahead. It speeds up the operation in soft soil and slows down and digs deeper appropriately in hard and compacted soil to ensure the cleanliness rate. This achieves maximum efficiency while ensuring harvesting quality and overcomes the efficiency loss or incomplete excavation problems caused by the traditional machinery maintaining constant parameters.

[0017] 3. In this invention, complex harvesting tasks can be decomposed into independent functional modules such as digging control, separation control, visual perception, and analysis and decision-making, and coupled through clear signal interfaces. This architecture not only makes the system logic clear and easy to debug and maintain, but also provides convenience for functional upgrades and multi-crop adaptation. At the same time, the combination of local real-time control and cloud data services ensures the stable operation of core functions and the potential for continuous optimization. Attached Figure Description

[0018] Figure 1 This is a three-dimensional structural schematic diagram of an Astragalus harvester proposed in this invention; Figure 2 This is a schematic diagram of the sieve frame structure of an Astragalus harvester proposed in this invention; Figure 3 This is a schematic diagram of the spring position structure of an Astragalus harvester proposed in this invention; Figure 4 This is a schematic diagram of the industrial camera position structure of an Astragalus harvester proposed in this invention; Figure 5 This is a schematic diagram of the system architecture of the control system for an Astragalus harvester proposed in this invention.

[0019] The components include: 1. Frame; 2. Drive base; 3. Connecting frame; 4. Shovel plate; 5. Moving groove; 6. Eccentric wheel; 7. Connecting belt; 8. Pressure roller; 9. Screen frame; 10. Screen mesh; 11. Fixing plate; 12. Traction component; 13. Support rod; 14. Moving wheel; 15. Telescopic rod; 16. Spring; 17. Slider; 18. Industrial camera. Detailed Implementation

[0020] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Example 1: In a first embodiment of the present invention, an Astragalus harvester is provided, such as... Figures 1-4As shown, the device includes a frame 1, a shovel 4 fixed on one side of the frame 1, a soil hardness sensor integrated on the surface of the shovel 4, a drive seat 2 symmetrically mounted on the top of the frame 1, a connecting frame 3 fixedly mounted on the top surface of the drive seat 2, an industrial camera 18 mounted on the bottom surface of the connecting frame 3, symmetrical moving grooves 5 on both sides of the frame 1, two sets of screen frames 9 on the inner side of the frame 1, with the sides of the screen frames 9 extending into the inner side of the moving groove 5, screen mesh 10 fixedly mounted on the inner side of the two sets of screen frames 9, two sets of pressure rollers 8 mounted on the inner side of the frame 1, and the two sets of pressure rollers 8 respectively positioned above the two sets of screen mesh 10, an eccentric wheel 6 on the inner side of the moving groove 5, and a connecting belt 7 connected between the eccentric wheel 6 and the drive seat 2 via a pulley, and a telescopic rod 15 on the inner side of the moving groove 5, with a spring 16 installed inside the telescopic rod 15.

[0022] Specifically, the drive base 2 and the frame 1 are respectively equipped with a first drive device and a second drive device for driving the eccentric wheel 6 and the pressure roller 8 to rotate. In a preferred embodiment, the first drive device is composed of a dual-shaft motor. The two drive ends of the dual-shaft motor are respectively connected to the pulley at one end of the connecting belt 7 through a shaft. When the dual-shaft motor is driven, it can drive the connecting belt 7 to drive the eccentric wheel 6 connected to the pulley at the other end of the connecting belt 7 to rotate. When the eccentric wheel 6 rotates, it can abut against the screen frame 9 placed in the moving groove 5, and thus drive the screen frame 9 to move back and forth inside the moving groove 5. The back and forth moving screen frame 9 can screen the harvest placed on the top of the screen frame 9, that is, the top surface of the screen 10, so as to facilitate the separation of the harvest from the soil. In a preferred embodiment, the second drive device is composed of a motor. The output end of the motor is connected to the pressure roller 8, and thus can drive the pressure roller 8 to rotate to crush the harvest placed on the top surface of the screen 10, thereby increasing the separation effect of the harvest from the soil. One end of the telescopic rod 15 is connected to the screen frame 9. When the screen frame 9 moves inside the moving groove 5, it will press the telescopic rod 15 inward. When the telescopic rod 15 is pressed, it can press the spring 16 inside the telescopic rod 15. At the same time, the spring 16 can drive the telescopic rod 15 to reset through the reverse force, and then push the screen frame 9 to reset. This ensures that when the eccentric wheel 6 pushes the screen frame 9, the screen frame 9 can move back and forth normally through the elastic potential energy of the spring 16. The two sets of upper frames are set up one above the other. The harvested material is first scooped up by the shovel plate 4 and falls onto the upper screen frame 9. Through the back and forth sieving and transportation of the upper screen frame 9, the harvested material will fall into the lower screen frame 9, which can perform secondary sieving of the crops and ensure the accuracy of sieving.

[0023] Please see the appendix Figure 2The upper and lower sides of the screen frame 9 are fixedly installed with sliders 17. The inner wall of the moving groove 5 is provided with a sliding groove, and the sliders 17 are located inside the sliding groove. A fixed plate 11 is fixedly installed on one side of the frame 1. A traction component 12 is fixedly installed on the top surface of the fixed plate 11. Support rods 13 are fixedly installed on both sides of the frame 1. A moving wheel 14 is installed at the bottom end of the support rod 13.

[0024] Specifically, the upper and lower inner walls of the moving groove 5 are provided with sliding grooves, and the slider 17 is movably installed inside the sliding groove. Thus, when the screen frame 9 moves back and forth inside the moving groove 5, the slider 17 can slide inside the sliding groove, thereby ensuring the directionality and stability of the screen frame 9 when moving inside the moving groove 5. The traction member 12 can be used to easily connect the frame 1 to an external traction vehicle, and the moving wheel 14 can be used to facilitate the overall movement of the frame 1.

[0025] Example 2: Existing equipment lacks the ability to perceive and provide feedback on separation quality and operational effectiveness in real time, thus failing to autonomously optimize operating parameters to maintain the harvesting target of "low damage and high purity" under complex conditions. To address these issues, this invention provides an image recognition system, the structure of which is as follows: Figure 5 As shown. The specific implementation process of this system is as follows: The excavation execution and soil information sensing module is used to control the excavation operation and to acquire soil hardness information of the work area in real time through a soil hardness sensor during the excavation operation.

[0026] Furthermore, the workflow of the excavation execution and soil information sensing module includes: Receive external operation start command and control the shovel 4 to descend to the initial working depth to start digging; During the excavation process, the soil hardness sensor integrated on the surface of the shovel plate 4 continuously collects the original soil hardness signals in the working area in front of and below the shovel plate 4 at a preset sampling frequency. The collected raw soil hardness signal is filtered and converted from analog to digital to obtain real-time digital soil hardness information. The digitized soil hardness information is sent to the excavation parameter control module at a preset communication cycle. While continuing the excavation operation, repeat the above steps in a loop until an external operation stop command is received.

[0027] Specifically, the excavation execution and soil information sensing module receives an external operation start command. The external operation start command originates from the start signal triggered by the driver through the vehicle human-machine interface. After receiving the command, the excavation execution and soil information sensing module immediately sends a control signal to the hydraulic control system or electric push rod system to drive the shovel 4 to descend from the transport position to the preset initial working depth. The initial working depth is preset according to the average root length of the Astragalus variety, for example, it is set to 50 centimeters. At the same time, the module sends a signal to the whole machine walking drive system to make the frame 1 start moving along the field ridge at an initial speed, thereby starting the excavation operation. Simultaneously, the module collects raw soil hardness signals in real time. During the process of the frame 1 driving the shovel 4 to move and perform digging operations, the soil hardness sensor integrated on the surface of the shovel 4 starts to work. The sensor continuously collects data at a preset sampling frequency, which is set according to the working speed, for example, 100 times per second. The sensor simultaneously collects the hardness signal of the soil that the shovel 4 is about to cut into and the hardness signal of the soil that is being worked on below the shovel 4, obtaining a continuous raw soil hardness signal. The module contains a signal processing unit, which receives the raw soil hardness signal from the soil hardness sensor in real time. The signal processing unit first filters the raw soil hardness signal to eliminate noise introduced by mechanical vibration and electromagnetic interference. Then, the signal processing unit performs analog-to-digital conversion on the filtered analog signal, converting it into a digital signal that the digital controller can directly process. After the above processing, real-time digital soil hardness information is obtained. The module encapsulates the obtained digitized soil hardness information into data packets according to a preset communication cycle and sends them to the excavation parameter control module through the vehicle communication network. The preset communication cycle and sampling frequency are coordinated and set, for example, once every 0.1 seconds, to ensure the real-time control. During the continuous excavation operation, the excavation execution and soil information perception module continuously executes the above steps in a loop, thereby realizing the continuous perception and reporting of soil hardness information in the work area. When the module receives an external operation stop command from the driver or the upper control system, it immediately stops the loop and sends a command to the drive system to raise the shovel 4 and stop moving, ending the excavation operation process.

[0028] The excavation parameter control module receives soil hardness information and generates control commands for excavation depth and travel speed through a built-in closed-loop control algorithm, which are then sent to the excavation execution and soil information sensing module.

[0029] Furthermore, the workflow of the parameter control module includes: Receive soil hardness information sent in real time by the excavation execution and soil information sensing module; The received soil hardness information is compared with the preset soil hardness threshold range in real time. When the soil hardness information is lower than the preset lower limit threshold of soil hardness, a first control command is generated. The first control command includes commands to increase the travel speed and maintain the current excavation depth. When the soil hardness information is higher than the preset upper limit threshold of soil hardness, a second control command is generated. The second control command includes commands to reduce the travel speed and increase the digging depth. When the soil hardness information is within the preset soil hardness threshold range, a third control instruction is generated. The third control instruction includes instructions to maintain the current travel speed and the current digging depth. The generated first, second, or third control command is sent to the excavation execution and soil information sensing module to control the execution of the excavation operation.

[0030] Specifically, the excavation parameter control module continuously monitors and receives data packets sent by the excavation execution and soil information sensing modules according to a preset communication cycle via the vehicle controller LAN bus or Ethernet. The excavation parameter control module parses the data packets and extracts the real-time digital soil hardness information contained therein. The module stores a preset soil hardness threshold range, which is pre-calibrated through experiments based on the typical soil conditions of Astragalus membranaceus planting fields. This range includes a lower soil hardness threshold and an upper soil hardness threshold. The module compares the received current soil hardness information with the preset soil hardness threshold range in real time to determine whether the current soil hardness information is below the lower soil hardness threshold, above the upper soil hardness threshold, or within the range formed by both the lower and upper soil hardness thresholds. Based on the comparison results from the second step, the excavation parameter control module executes different instruction generation logic. Specifically, when the current soil hardness information is determined to be below the lower limit threshold, indicating that the soil is too soft, the module generates a first control instruction. This first control instruction contains two parts: first, it sends an instruction to the machine's walking drive system to increase the travel speed, for example, by 10%; second, it sends an instruction to the shovel blade 4 attitude control system to maintain the current excavation depth. When the current soil hardness information is determined to be above the upper limit threshold, indicating that the soil is too hard and compacted, the module generates a second control instruction. This second control instruction contains two parts. The module has two main functions: first, it sends a command to the machine's walking drive system to reduce the travel speed, for example, by 20%; second, it sends a command to the shovel plate 4 attitude control system to increase the digging depth, for example, by 5 centimeters from the current depth, to ensure that the roots are completely dug up. When the current soil hardness information is determined to be within the range formed by the lower and upper limits of soil hardness, it indicates that the soil hardness is suitable. At this time, the module generates a third control command, which includes two specific parts: first, it sends a command to the machine's walking drive system to maintain the current travel speed; second, it sends a command to the shovel plate 4 attitude control system to maintain the current digging depth. The module encapsulates the generated first, second, or third control instructions into a control command frame and sends it in real time to the excavation execution and soil information sensing module via the vehicle communication network. The excavation execution and soil information sensing module receives and parses the control command frame and, based on the instructions contained therein, drives the corresponding walking drive system and shovel 4 attitude control system to adjust the speed and depth. At the same time, the excavation parameter control module continuously and cyclically executes the above steps, thereby achieving adaptive closed-loop control of the entire excavation operation process based on the dynamic changes in soil hardness.

[0031] The separation execution module is used to separate the soil roots of the harvested material by controlling the operation of the screen frame 9 and the pressure roller 8 after the excavation operation is completed.

[0032] Furthermore, the workflow of the separate execution module includes: When the excavation execution and soil information sensing modules complete the excavation work in the current area, the separation execution module receives the start command; The separation execution module drives the first drive device installed in the drive base 2 to operate. The first drive device drives the eccentric wheel 6 to rotate through the connecting belt 7. The rotational motion of the eccentric wheel 6 is converted into the reciprocating vibration of the screen frame 9 along the moving groove 5. While the screen frame 9 is reciprocating, the separation execution module drives the second drive device installed inside the frame 1 to run, and the second drive device drives the two sets of pressure rollers 8 to rotate in opposite directions. The harvested material, conveyed to the screen frame 9 by the digging shovel 4, first undergoes the rotational crushing of the pressure roller 8, and then falls onto the surface of the screen 10, which is in a reciprocating vibration state. The harvested material is thrown up, tumbled and conveyed forward under the reciprocating vibration of the screen 10, while soil particles are separated and fall through the mesh of the screen 10. The separation execution module continues to execute the above steps until the current batch of harvested material completes soil-root separation and leaves the screen 10, or receives an external stop command.

[0033] Specifically, when the excavation execution and soil information sensing module completes the excavation work in the current travel area and transports the harvested material to the separation station, the separation execution module receives a start command from the upper-level scheduling system or the preset program logic. After receiving the start command, it first sends a start signal to the first drive device installed in the drive base 2. The first drive device is a dual-axis motor. The first drive device starts to rotate, and its output shaft transmits power to the rotating shaft of the eccentric wheel 6 located inside the moving groove 5 through the transmission of the pulley and the connecting belt 7, driving the eccentric wheel 6 to rotate. The rotational motion of the eccentric wheel 6 is converted into the reciprocating linear vibration of the screen frame 9 along the moving groove 5 on the frame 1 through its contact with the side wall of the screen frame 9. The upper and lower sides of the screen frame 9 are engaged with the sliding groove of the inner wall of the moving groove 5 through the slider 17 to ensure the stability of the vibration trajectory. At the same time as or within a very short time interval after the screen frame 9 begins to reciprocate, the separation execution module sends a start signal to the second drive device installed inside the frame 1. The second drive device is an electric motor. The second drive device drives the two sets of pressure rollers 8 to rotate in opposite directions through gear or chain transmission or direct transmission. That is, the rotation direction of the two sets of pressure rollers 8 makes the linear velocity direction of their contact surface face the surface of the screen 10. The mixture of Astragalus membranaceus and soil, which is transported and fallen by the digging shovel plate 4, is sent to the upper entrance of the screen frame 9. The harvested material first falls into the bottom of the pressure roller 8 and is subjected to the rotation and crushing of the pressure roller 8. The crushing action breaks up the large pieces of compacted soil and loosens the adhesion between the soil and the Astragalus membranaceus roots. Subsequently, the harvested material falls onto the surface of the screen 10, which is in a reciprocating vibration state, under the push of gravity and material flow. The harvested material falling onto the surface of the screen 10 is continuously thrown up, tumbled, and conveyed forward along the tilt angle of the screen 10 under the high-frequency reciprocating vibration of the screen 10. During this process, the broken or loosened soil particles are separated and fall into the field through the mesh of the screen 10 under the action of vibration and gravity. The Astragalus root system with a size larger than the mesh and the large soil clods that do not pass through remain on the surface of the screen 10 and continue to move forward. The continuous vibration of the screen 10 causes the Astragalus root system to further rub and separate from the residual soil clods. The pure Astragalus root system is finally conveyed to the end of the screen 10 and collected. The separation execution module controls the first drive device and the second drive device to run continuously, so that the screen frame 9 vibrates and the pressure roller 8 rotates continuously until it is determined that the current batch of harvested material has completely passed through the screen 10 and completed the separation, or a stop command is received from the outside. After receiving the stop command, the separation execution module sends a stop signal to the second drive device and the first drive device in sequence to end the separation operation process.

[0034] The visual information perception module is used to collect image information of the harvested material in real time through an industrial camera 18 deployed on the bottom surface of the connecting frame 3 during the soil separation operation.

[0035] Furthermore, the workflow of the visual information perception module includes: When the separation execution module starts the operation of the screen frame 9 and the pressure roller 8, the visual information perception module simultaneously triggers the industrial camera 18 deployed on the bottom surface of the connecting frame 3 to enter the working state. During the soil separation operation, the industrial camera 18 continuously captures images of the harvested material on the surface of the screen 10 at a preset frame rate to obtain the original image sequence. The original image sequence is preprocessed to obtain a stable image to be analyzed; The images to be analyzed are marked with timestamps and corresponding working status information of sieve frame 9, and packaged to form image information of the harvested items; The packaged image information is sent to the visual analysis and decision-making module in real time. The visual information perception module continues to execute the above steps until the separate execution module stops operating.

[0036] Specifically, when the separation execution module starts the operation of the screen frame 9 and the pressure roller 8, the visual information perception module triggers the industrial camera 18 deployed on the bottom of the connecting frame 3 to enter the working state through electrical interlock or by receiving the start signal from the separation execution module. The industrial camera 18 is powered on and initialized, and its lens is vertically or tilted to the working area in the middle and rear section of the screen 10 below. During the soil and root separation operation, the industrial camera 18 continuously captures images at a preset frame rate. The preset frame rate is set according to the vibration frequency of the screen frame 9 and the speed of material movement, for example, thirty frames per second, to ensure that the material state in the dynamic process can be clearly captured. The industrial camera 18 continuously captures images of the harvested material on the surface of the screen 10 below, including Astragalus root system, soil particles and their mixture, to obtain the original image sequence arranged in chronological order. The visual information perception module contains an image processing unit that processes the raw image sequence from the industrial camera 18 frame by frame. The preprocessing first performs illumination compensation, adjusting the grayscale value according to the overall brightness of the image to eliminate interference caused by changes in ambient natural light. Then, image stabilization is performed, using algorithms to identify and compensate for image blurring or inter-frame displacement caused by the vibration of the frame 1 and the screen frame 9, ensuring that the image is clear and stable. After preprocessing, a stable image content that can be used for analysis is obtained. The visual information perception module adds data tags to each frame of the image to be analyzed. First, it adds a precise timestamp to each frame. Second, it obtains the current working status information of the screen frame 9 from the separation execution module or sensor, including but not limited to the real-time vibration frequency of the screen frame 9. Finally, the images to be analyzed with timestamps and screen frame 9 working status information are packaged into a complete data structure, which is defined as the image information of the harvested object. The visual information perception module packages the image information and sends it to the visual analysis and decision-making module in real time via the vehicle's high-speed data bus. The transmission process ensures low latency to meet the needs of subsequent real-time analysis and decision-making. The visual information perception module controls the industrial camera 18 to run continuously and execute the above steps in a loop, thereby realizing continuous visual monitoring of the separation operation quality. When the separation execution module stops operating, the visual information perception module receives the corresponding stop signal, and then controls the industrial camera 18 to stop shooting and end the entire image acquisition and transmission process.

[0037] The visual analysis and decision-making module is used to receive image information, analyze the surface damage rate and soil residue status of the harvested material in real time through the built-in visual recognition model, and generate corresponding control instructions for the working parameters of the sieve frame 9 and the pressure control instructions for the pressure roller 8 based on the analysis results.

[0038] Furthermore, the workflow of the visual analytics decision-making module includes: It receives image information sent in real time by the visual information perception module and processes the image to be analyzed in the image information; The model identifies and segments the harvest outline region and soil clod impurity region in the image to be analyzed. Within the harvest outline region, the visual recognition model calculates the ratio of the area of ​​damaged surface pixels to the total area of ​​surface pixels as the real-time surface damage rate. Within the soil clod impurity area, the visual recognition model counts the total projected area of ​​the remaining soil clods as a quantitative indicator of the real-time soil clod residue status. The calculated skin damage rate is compared with the preset damage rate threshold, and the quantified soil block residual state is compared with the preset impurity rate threshold. Based on the comparison results, the visual analysis and decision-making module generates corresponding control instructions for the working parameters of the screen frame 9 and the pressure control instructions for the pressure roller 8.

[0039] Specifically, the visual analysis and decision-making module receives image information sent by the visual information perception module in real time through a data interface. The module parses the image information, extracts the image to be analyzed, and the corresponding timestamp and screen frame status information. The visual analysis and decision-making module calls its built-in, pre-trained visual recognition model to process the extracted image to be analyzed. This visual recognition model is a deep learning-based image semantic segmentation model, which performs the following specific analysis tasks: First, the model identifies and segments different target regions in the image to be analyzed. The model classifies and segments all pixels in the image into at least two regions: one is the harvest outline region, i.e., the pixel region occupied by the Astragalus root system; the other is the soil clod and impurity region, i.e., the region containing soil clods and impurities other than the root system. The model first identifies the pixel areas occupied by soil clumps, stones, and other impurities. Then, within the segmented harvest outline area, the visual recognition model further identifies the healthy and damaged parts of the root epidermis. The model calculates the total area of ​​all pixels identified as epidermal damage, i.e., the epidermal damage pixel area. At the same time, the model calculates the total area of ​​pixels enclosed by the outer contour of the entire harvest outline area, i.e., the total epidermal pixel area. Finally, the model calculates the ratio of the epidermal damage pixel area to the total epidermal pixel area and outputs this ratio as the real-time epidermal damage rate. Meanwhile, within the segmented soil clump impurity area, the visual recognition model counts the total area of ​​all pixels identified as residual soil clumps, i.e., the total projected area of ​​the residual soil clumps. This total area is output as a real-time quantitative indicator of the residual soil clump status. The visual analysis and decision-making module internally stores preset damage rate thresholds and preset impurity rate thresholds. These two thresholds are pre-set experimentally based on the requirements for Astragalus membranaceus product grading and subsequent processing standards. The module compares the real-time epidermal damage rate obtained from the visual recognition model with the preset damage rate thresholds. Simultaneously, the module compares the real-time quantitative index of soil residue state obtained from the visual recognition model with the preset impurity rate thresholds. Based on the comparison results from the third step, the visual analysis and decision-making module executes decision logic and generates corresponding control instructions. When the comparison results show that the real-time epidermal damage rate is higher than the preset damage rate threshold, the module will initiate a control mechanism. When the damage rate threshold is reached, the module determines that the current separation operation is causing excessive damage to the root system. At this point, the module generates commands to adjust the working parameters of the screen frame 9 and the pressure of the pressure roller 8. These commands reduce the vibration frequency of the screen frame 9 and the applied pressure of the pressure roller 8 to weaken the separation intensity. If the comparison results show that the real-time quantitative index of soil residue is higher than the preset impurity rate threshold, the module determines that the purity of the current separation operation is insufficient. At this point, the module generates commands to adjust the working parameters of the screen frame 9 and the pressure of the pressure roller 8. These commands increase the vibration frequency of the screen frame 9 and the applied pressure of the pressure roller 8 to enhance the separation intensity. When neither of the two indicators exceeds its respective threshold, the module determines that the current separation quality is qualified and generates an instruction to maintain the current working parameters of the screen frame 9 and the pressure roller 8. The visual analysis and decision-making module packages the above-generated screen frame 9 working parameter control instructions containing specific adjustment content and pressure control instructions of pressure roller 8, along with necessary timestamp information, and sends them to the separation parameter control module. The visual analysis and decision-making module continuously executes the above steps in a loop, performs real-time analysis on the image information stream continuously sent by the visual information perception module, and dynamically generates control instructions, thereby realizing closed-loop quality control and adaptive parameter optimization of the soil root separation process.

[0040] The separation parameter control module is used to send the received working parameter control commands of the screen frame 9 and pressure control commands of the pressure roller 8 to the separation execution module to control the operation of the screen frame 9 and the pressure roller 8 to perform corresponding adjustments.

[0041] Furthermore, the workflow of the separate parameter control module includes: Receive the working parameter adjustment command of screen frame 9 and the pressure adjustment command of pressure roller 8 sent by the vision analysis and decision module, and parse the working parameter adjustment command of screen frame 9 to obtain the target adjustment value of the vibration frequency of screen frame 9. Based on the target adjustment value of the vibration frequency of the screen frame 9, a first control signal for controlling the rotational speed of the first drive device is calculated and generated; The pressure control command of pressure roller 8 is analyzed to obtain the target adjustment value of pressure roller 8; Based on the target adjustment value of the pressure roller 8, a second control signal is calculated and generated to control the output torque or hydraulic pressure of the second drive device. The first control signal and the second control signal are respectively sent to the first drive device and the second drive device; The first drive device responds to the first control signal, changes the rotation speed, and drives the vibration frequency of the screen frame 9 to the target value through the connecting belt 7 and the eccentric wheel 6. The second drive unit responds to the second control signal, changes the output torque or hydraulic pressure, and drives the pressure roller 8 to adjust the pressure to the target value.

[0042] Specifically, the separation parameter control module receives the screen frame 9 working parameter control command and the pressure roller 8 pressure control command sent by the vision analysis and decision module through the communication interface. The module first parses the screen frame 9 working parameter control command, which clearly includes the target adjustment value of the screen frame 9 vibration frequency. This value is a specific frequency value or an adjustment range relative to the current frequency. The module also parses the pressure roller 8 pressure control command, which clearly includes the target adjustment value of the pressure roller 8 pressure. This value is a specific pressure value or an adjustment range relative to the current pressure. Based on the target adjustment value of the screen frame 9 vibration frequency obtained from the parsing, the separation parameter control module combines the current speed of the first drive device and the transmission ratio relationship of the mechanical transmission system composed of the eccentric wheel 6 and the transmission belt. Through the built-in calculation model, it calculates the speed setting value of the first drive device required to achieve the target vibration frequency. Subsequently, the module generates the corresponding first control signal based on the speed setting value. If the dual-axis motor of the first drive device is an AC motor, the first control signal is the frequency control command of the frequency converter. If it is a DC motor, it is the pulse width modulation duty cycle command. The separation parameter control module calculates the set value of the output torque of the second drive device or the hydraulic system pressure required to achieve the target pressure based on the target adjustment value of the pressure roller 8 obtained from the analysis and the feedback value of the current pressure sensor, through the built-in pressure control algorithm. Then, the module generates a corresponding second control signal based on the set value. The second drive device is a motor, and the second control signal is a torque or current control command. The separation parameter control module sends the first control signal generated above to the first drive device that drives the screen frame 9 to vibrate through the corresponding drive circuit or controller local area network bus. At the same time, the module sends the second control signal generated above to the second drive device that drives the pressure roller 8 through the corresponding drive circuit or controller local area network bus. The first drive unit receives the first control signal, and its built-in controller or external frequency converter or servo driver changes its own rotation speed according to the requirements of the first control signal. The change in rotation speed is transmitted to the eccentric wheel 6 through the connecting belt 7, causing the rotation speed of the eccentric wheel 6 to change. Since there is a direct linear transmission relationship between the rotation speed of the eccentric wheel 6 and the reciprocating vibration frequency of the screen frame 9, the vibration frequency of the screen frame 9 is adjusted to the target adjustment value. The second drive unit receives the second control signal, and its controller changes the output torque of its output shaft according to the requirements of the second control signal. The change in output torque is directly transmitted to the two sets of pressure rollers 8, so that the linear pressure or extrusion force applied by the pressure rollers 8 to the harvest is adjusted to the target adjustment value. The separation parameter control module continuously listens to and receives control instructions from the visual analysis and decision module, and executes the above steps in a loop, thereby achieving dynamic and closed-loop adjustment of the vibration frequency of the screen frame 9 and the pressure of the pressure roller 8, ensuring that the soil separation operation always operates under the guidance of visual feedback, towards the optimal goal of low damage and high purity.

[0043] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An Astragalus harvester, comprising a frame (1), characterized in that, A shovel plate (4) is installed on one side of the frame (1), and a soil hardness sensor is integrated on the surface of the shovel plate (4). A drive seat (2) is symmetrically installed on the top of the frame (1), and a connecting frame (3) is fixedly installed on the top surface of the drive seat (2). An industrial camera (18) is installed on the bottom surface of the connecting frame (3). Moving slots (5) are symmetrically opened on both sides of the frame (1). Two sets of screen frames (9) are provided on the inner side of the frame (1), and the two sides of the screen frames (9) extend into the moving slots. Inside the 5), the inner sides of the two sets of screen frames (9) are fixedly installed with screens (10), the inner sides of the frame (1) are installed with two sets of pressure rollers (8), and the two sets of pressure rollers (8) are respectively placed above the two sets of screens (10). The inner side of the moving groove (5) is provided with an eccentric wheel (6), and the eccentric wheel (6) is connected to the drive seat (2) by a connecting belt (7) through a pulley. The inner side of the moving groove (5) is provided with a telescopic rod (15), and the inside of the telescopic rod (15) is equipped with a spring (16).

2. The Astragalus harvester according to claim 1, characterized in that, The upper and lower sides of the screen frame (9) are fixedly installed with sliders (17), the inner wall of the moving groove (5) is provided with a sliding groove, and the sliders (17) are set inside the sliding groove.

3. The Astragalus harvester according to claim 1, characterized in that, A fixing plate (11) is fixedly installed on one side of the frame (1), a traction component (12) is fixedly installed on the top surface of the fixing plate (11), and support rods (13) are fixedly installed on both sides of the frame (1). A moving wheel (14) is installed at the bottom end of the support rod (13).

4. A control system for an Astragalus membranaceus harvester, characterized in that, The system includes the following modules: The excavation execution and soil information sensing module is used to control the excavation operation and, during the excavation operation, to acquire the soil hardness information of the operation area in real time through the soil hardness sensor. The excavation parameter control module is used to receive the soil hardness information, generate control commands for excavation depth and travel speed through a built-in closed-loop control algorithm, and send them to the excavation execution and soil information sensing module. The separation execution module is used to perform soil root separation operation on the harvested material by controlling the operation of the screen frame (9) and the pressure roller (8) after the excavation operation is completed; The visual information perception module is used to collect image information of the harvested material in real time through an industrial camera (18) deployed on the bottom surface of the connecting frame (3) during the soil separation operation. The visual analysis and decision module is used to receive the image information, analyze the surface damage rate and soil residue status of the harvested material in real time through the built-in visual recognition model, and generate corresponding screen frame (9) working parameter control instructions and pressure roller (8) pressure control instructions based on the analysis results. The separation parameter control module is used to send the received working parameter control instructions of the screen frame (9) and pressure control instructions of the pressure roller (8) to the separation execution module respectively, so as to control the operation of the screen frame (9) and the pressure roller (8) to perform corresponding adjustments.

5. The control system for an Astragalus harvester according to claim 4, characterized in that, The workflow of the excavation execution and soil information sensing module includes: Receive external operation start command and control the shovel (4) to descend to the initial working depth to start excavation; During the excavation process, the soil hardness sensor integrated on the surface of the shovel plate (4) continuously collects the original soil hardness signals of the working area in front of and below the shovel plate (4) at a preset sampling frequency. The collected raw soil hardness signal is filtered and converted from analog to digital to obtain real-time digital soil hardness information. The digitized soil hardness information is sent to the excavation parameter control module at a preset communication cycle. While continuing the excavation operation, repeat the above steps in a loop until an external operation stop command is received.

6. The control system of the Astragalus harvester according to claim 4, characterized in that, The workflow of the mining parameter control module includes: Receive soil hardness information sent in real time by the excavation execution and soil information sensing module; The received soil hardness information is compared with the preset soil hardness threshold range in real time. When the soil hardness information is lower than the preset lower limit threshold of soil hardness, a first control command is generated. The first control command includes commands to increase the travel speed and maintain the current excavation depth. When the soil hardness information is higher than the preset upper limit threshold of soil hardness, a second control command is generated. The second control command includes commands to reduce the travel speed and increase the digging depth. When the soil hardness information is within the preset soil hardness threshold range, a third control instruction is generated, which includes an instruction to maintain the current travel speed and the current digging depth. The generated first control command, second control command, or third control command is sent to the excavation execution and soil information sensing module to control the execution of the excavation operation.

7. The control system for an Astragalus harvester according to claim 4, characterized in that, The workflow of the separate execution module includes: When the excavation execution and soil information sensing module completes the excavation work in the current area, the separation execution module receives a start command; The separation execution module drives the first drive device installed in the drive seat (2) to run. The first drive device drives the eccentric wheel (6) to rotate through the connecting belt (7). The rotational motion of the eccentric wheel (6) is converted into the reciprocating vibration of the screen frame (9) along the moving groove (5). While the screen frame (9) is reciprocating, the separation execution module drives the second drive device installed inside the frame (1) to run, and the second drive device drives the two sets of pressure rollers (8) to rotate in opposite directions; The harvested material, which is conveyed to the screen frame (9) by the digging shovel (4), first bears the rotation and crushing of the pressure roller (8), and then falls onto the surface of the screen (10) which is in a reciprocating vibration state after being crushed. The harvested material is thrown up, tumbled and conveyed forward under the reciprocating vibration of the screen (10), while soil particles are separated and fall through the mesh of the screen (10); The separation execution module continues to execute the above steps until the current batch of harvested material completes soil-root separation and leaves the screen (10), or receives an external stop command.

8. The control system for an Astragalus harvester according to claim 4, characterized in that, The workflow of the visual information perception module includes: When the separation execution module starts the operation of the screen frame (9) and the pressure roller (8), the visual information perception module synchronously triggers the industrial camera (18) deployed on the bottom surface of the connecting frame (3) to enter the working state. During the soil separation operation, the industrial camera (18) continuously captures images of the harvested material on the surface of the screen (10) below at a preset frame rate to obtain the original image sequence; The original image sequence is preprocessed to obtain a stable image to be analyzed; The image to be analyzed is marked with a timestamp and the corresponding working status information of the sieve frame (9), and packaged to form the image information of the harvested item; The packaged image information is sent to the visual analysis and decision-making module in real time. The visual information perception module continues to execute the above steps until the separate execution module stops operating.

9. The control system for an Astragalus harvester according to claim 4, characterized in that, The workflow of the visual analysis decision module includes: Receive image information sent in real time by the visual information perception module, and process the image to be analyzed in the image information; The harvest outline region and soil clod impurity region in the image to be analyzed are identified and segmented. Within the harvest outline region, the visual recognition model calculates the ratio of the area of ​​damaged pixels to the total area of ​​pixels on the surface as the real-time surface damage rate. Within the soil clod impurity area, the visual recognition model calculates the total projected area of ​​the residual soil clods as a real-time quantitative indicator of the residual soil clod status. The calculated skin damage rate is compared with a preset damage rate threshold, and the quantified soil clod residual state is compared with a preset impurity rate threshold. Based on the comparison results, the visual analysis decision module generates corresponding control instructions for the working parameters of the sieve frame (9) and control instructions for the pressure of the pressure roller (8).

10. The control system for an Astragalus harvester according to claim 4, characterized in that, The workflow of the separation parameter control module includes: Receive the screen frame (9) working parameter adjustment command and the pressure roller (8) pressure adjustment command sent by the vision analysis decision module, and parse the screen frame (9) working parameter adjustment command to obtain the target adjustment value of the screen frame (9) vibration frequency; Based on the target adjustment value of the vibration frequency of the sieve frame (9), a first control signal for controlling the rotation speed of the first drive device is calculated and generated; The pressure control command of the pressure roller (8) is analyzed to obtain the target adjustment value of the pressure roller (8); Based on the target adjustment value of the pressure of the pressure roller (8), a second control signal for controlling the output torque or hydraulic pressure of the second drive device is calculated and generated; The first control signal and the second control signal are respectively sent to the first drive device and the second drive device; The first drive device responds to the first control signal, changes the rotation speed, and drives the vibration frequency of the screen frame (9) to the target value through the connecting belt (7) and the eccentric wheel (6); The second drive device responds to the second control signal, changes the output torque or hydraulic pressure, and drives the pressure roller (8) to adjust the pressure to the target value.

Images