An intelligent subarea and subzone variable pesticide application machine for corn-soybean strip interplanting and a control method thereof

Abstract

A smart zoned and variable-rate sprayer and its control method for corn-soybean strip intercropping belong to the field of agricultural machinery technology. This invention solves the technical problems of existing integrated boom sprayers, which cannot achieve crop zone differentiation and are prone to pesticide drift and cross-zone mis-spraying during operation. A zoned spraying mechanism is installed on the suspension frame, including a transverse main beam, multiple independent boom units, an intelligent identification module, and a zoned variable-rate control module. The central control system receives the identification results output by the intelligent identification module and issues task commands to the zoned variable-rate control module. The zoned variable-rate control module drives the displacement of each independent boom unit and outputs variable-rate spraying control commands to the independent atomizing components according to the weed distribution. Flexible isolation covers are installed on the outside of the independent boom units; the flexible magnetic strips of adjacent flexible isolation covers can adhere to each other and seal, forming an isolation barrier. This solves the problem of pesticide damage in corn-soybean intercropping.

Description

Technical Field

[0001] This invention relates to an intelligent zoned and variable-rate spraying machine and its control method for corn-soybean strip intercropping, belonging to the field of agricultural value-preserving machinery technology. Background Technology

[0002] The corn-soybean strip intercropping model can fully utilize land, light, heat, and space resources, effectively increasing the land multiple cropping index and grain and oil yield per unit area. It is a core cultivation model for achieving stable and increased grain and oil production and improving arable land utilization efficiency. However, corn is a monocotyledonous grass crop, while soybean is a dicotyledonous broadleaf crop. The two crops have fundamentally different herbicide tolerance characteristics: corn can tolerate broadleaf herbicides but is highly sensitive to grass herbicides; soybean can tolerate grass herbicides but is highly sensitive to broadleaf herbicides.

[0003] In field chemical weeding operations, traditional integrated boom sprayers cannot perform strip-specific spraying, leading to pesticide drift and accidental spraying across strips. This can easily cause herbicide damage to corn and soybeans, resulting in crop burn, yield reduction, or even crop failure, severely hindering the large-scale and industrialized promotion and application of corn-soybean strip intercropping. Currently, existing strip-specific spraying equipment mostly relies on passive isolation structures such as physical partitions and simple circuit switches, which have substantial defects such as poor isolation effect, lack of variable spraying performance, and weak field adaptability and versatility.

[0004] In summary, existing integrated boom sprayers have technical problems such as inability to differentiate crop zones during operation, and the tendency for pesticide drift and accidental spraying across crop zones. Summary of the Invention

[0005] This invention addresses the technical problems of existing integrated boom sprayers, which cannot perform crop strip differentiation operations and are prone to pesticide drift and mis-spraying across strips during operation. Therefore, it proposes an intelligent strip-differentiated variable-rate pesticide application machine for corn-soybean strip intercropping. It includes a self-propelled chassis, a suspension frame, a pesticide tank, and a central control system. The suspension frame is installed on the front of the self-propelled chassis, and the pesticide tank is installed on the upper rear of the self-propelled chassis. A strip-differentiated pesticide application mechanism is mounted on the suspension frame.

[0006] The strip-type spraying mechanism includes a transverse main beam, multiple sets of independent spray bar units, an intelligent identification module, and a zoned variable control module; the multiple sets of independent spray bar units are arranged and installed along the transverse main beam.

[0007] Multiple independent spray bar units can be independently slidably arranged along the length of the transverse main beam. Each independent spray bar unit can be moved to a designated position and correspond to a crop planting strip in the field by means of transverse sliding and / or vertical extension under the drive of the partition variable control module. Each independent spray bar unit is equipped with an independent atomizing component.

[0008] The central control system is electrically connected to the intelligent identification module and the partition variable control module, respectively.

[0009] The intelligent recognition module is used to collect field images and three-dimensional point cloud data in real time, identify crop planting zone type, crop planting zone boundary and weed distribution, and output the recognition results to the central control system.

[0010] The central control system is used to receive the recognition results output by the intelligent recognition module and issue task instructions to the partition variable control module according to the recognition results; the partition variable control module drives each group of independent spray bar units to move according to the task instructions and outputs variable application control instructions to the independent atomizing components according to the weed distribution.

[0011] A flexible isolation cover is installed on the outside of the independent spray bar unit. The side edge of the flexible isolation cover is embedded with a flexible magnetic strip. The flexible isolation cover can be lowered to the working position along with the independent spray bar unit. The flexible magnetic strips of adjacent flexible isolation covers can attract and seal each other to form an isolation barrier.

[0012] As another improvement of the present invention, a slide rail is provided on the front side of the transverse main beam, and a rack is provided on the front side of the slide rail. The independent spray bar unit includes a slider, a stepper motor and a gear. The slider is slidably connected to the slide rail. The slider is provided with a hollow channel. The gear is arranged in the channel and meshes with the rack. The stepper motor is arranged on the upper side of the slider. The top surface of the slider is provided with a mounting hole. The output end of the stepper motor is connected to the input end of the gear through the mounting hole. The stepper motor drives the independent spray bar unit to slide along the transverse main beam by driving the gear to mesh with the rack.

[0013] As another improvement of the present invention, the independent spray bar unit also includes an electric push rod, which is installed on the front side of the slider, and the independent atomizing component is installed at the bottom of the electric push rod. A connecting rod is installed on the lower part of the electric push rod, and the flexible isolation cover is installed on the outside of the independent spray bar unit through the connecting rod.

[0014] As another improvement of the present invention, the intelligent recognition module includes a high-definition visual camera and a lidar; the lidar is used to collect three-dimensional point cloud data, and the high-definition visual camera is used to collect field images. The two work together to identify the crop planting zone type, boundary and weed distribution through a multi-source sensor fusion algorithm.

[0015] As another improvement of the present invention, the medicine box includes a corn-specific medicine box and a soybean-specific medicine box, which are set independently; the strip-type application mechanism has two independent supply pipelines, which are respectively connected to the corn-specific medicine box and the soybean-specific medicine box; the independent atomization component includes an atomizing nozzle and a controllable solenoid valve group; the controllable solenoid valve group has two independent valves, which are respectively connected to the two independent supply pipelines; the zone variable control module controls the opening and closing of the two independent valves, and only one valve is opened at any given time, so that the atomizing nozzle sprays the special medicine for the corresponding crop planting zone.

[0016] As another improvement of the present invention, the variable application control command output by the partition variable control module to the independent atomizing component is a PWM pulse width modulation signal; the independent atomizing component adjusts the spraying flow rate of the atomizing nozzle according to the duty cycle of the PWM pulse width modulation signal.

[0017] As another improvement of the present invention, the intelligent identification module calculates the weed coverage rate ρ and determines the weed density level based on the ρ value:

[0018] When ρ < 5%, it is level 0, and the duty cycle of the PWM pulse width modulation signal is 0%.

[0019] When 5%≤ρ<20%, ​​it is level 1, and the duty cycle of the PWM pulse width modulation signal is 30%-50%;

[0020] When 20%≤ρ<50%, it is level 2, and the duty cycle of the PWM pulse width modulation signal is 50%-80%;

[0021] When ρ≥50%, it is level 3, and the duty cycle of the PWM pulse width modulation signal is 80%-100%.

[0022] The present invention also provides a control method based on an intelligent zoned and variable-rate spraying machine for corn-soybean strip intercropping as described in any one of claims 1 to 7, comprising the following steps:

[0023] S1. Pre-operation calibration and path generation: The planting parameters of the field to be operated are entered through the central control system, the sprayer is controlled to travel to the starting position of the field, the boundary of the first row of crops is calibrated, and an automatic operation path is generated.

[0024] S2. Real-time perception and multi-source fusion identification: During the process of the sprayer moving along the automatic operation path, the intelligent identification module collects field images and three-dimensional point cloud data in real time, and identifies the crop planting zone type, crop planting zone boundary coordinates and weed distribution based on the multi-source sensing fusion algorithm.

[0025] S3. Automatic row alignment and vertical adaptive adjustment: The central control system issues row alignment instructions to the partition variable control module based on the crop planting zone boundary coordinates identified by S2; the partition variable control module drives each group of independent spray bar units to slide along the horizontal main beam to directly above the corresponding crop planting zone, and drives each vertical lifting and vertical extension according to the crop canopy height, so that the bottom of the flexible isolation cover maintains a preset distance from the crop canopy or the ground.

[0026] S4. Magnetic Sealing Safety Interlock: When each group of independent spray bar units descends to the working position, the flexible magnetic strips on the edges of adjacent flexible isolation covers attract and adhere to each other, forming a continuous and sealed physical isolation barrier that spans the distance between adjacent crop planting belts. At the same time, the control system detects the sealing status through Hall sensors located on the edges of the flexible isolation covers.

[0027] S5. Dynamic calculation of weed coverage and variable application of pesticides: If a sealed barrier is detected to have been formed, the central control system dynamically calculates the weed coverage ρ in each crop planting zone based on the weed distribution identified in S2.

[0028] The corresponding PWM duty cycle is determined based on the ρ value, and a variable drug delivery control command containing this PWM duty cycle is issued to the partition variable control module; the partition variable control module drives the independent atomizing component to deliver the drug according to the command; wherein:

[0029] When ρ < 5%, it is level 0, the duty cycle of the PWM pulse width modulation signal is 0%, and no pesticide is applied;

[0030] When 5%≤ρ<20%, ​​it is level 1, the duty cycle of the PWM pulse width modulation signal is 30%-50%, and the dosage is low.

[0031] When 20%≤ρ<50%, it is level 2, the duty cycle of the PWM pulse width modulation signal is 50%-80%, and the dosage is medium.

[0032] When ρ≥50%, it is level 3, the duty cycle of the PWM pulse width modulation signal is 80%-100%, and high-volume application is allowed.

[0033] As another improvement of the present invention, the intelligent recognition module includes a high-definition visual camera and a LiDAR, which cooperate in recognition through a multi-source fusion strategy, wherein:

[0034] The recognition process of the high-definition visual camera includes:

[0035] Real-time field images are acquired, and an instance segmentation model is used for pixel-level image recognition. The model extracts multi-scale features from the images through a backbone network, embedding a coordinate attention mechanism during the feature extraction process to enhance the model's spatial sensitivity to the extension and width directions of crop strips and suppress background noise interference from soil, light, and shadow. The model outputs semantic segmentation, target detection, and instance segmentation branches in parallel through the instance segmentation head to distinguish between corn strips, soybean strips, and background areas, and outputs the location and outline of weed targets. The classification results of this model serve as the final basis for determining the crop strip type.

[0036] The identification process of the lidar includes:

[0037] The system acquires real-time 3D point cloud data of the work scene ahead, transforms the point cloud from the radar coordinate system to the vehicle world coordinate system, and performs downsampling processing; it calculates the ground clearance Hi of each point, and dynamically sets a height threshold Hth based on the current crop type fed back by the high-definition vision camera.

[0038] When the visual recognition identifies the current area as a corn belt, Hth is taken as 1 / 2 of the average height of the corn canopy in that area;

[0039] When the visual recognition identifies the current area as a soybean zone, Hth is taken as 1 / 2 of the average height of the soybean canopy in that area;

[0040] Points with Hi≥Hth are marked as candidate points for high-level crops, and points with 0.05m≤Hi<Hth are marked as candidate points for low-level background, thus achieving preliminary segmentation between the crop layer and the ground background. A clustering algorithm is used to extract single-row crop point cloud clusters from the segmented high-level point cloud, and the least squares method is used to fit the linear equation for each point cloud cluster, outputting the centerline and boundary line equations of each crop zone.

[0041] As another improvement of the present invention, the specific steps of the intelligent recognition module to calculate the weed coverage rate ρ include: mapping the two-dimensional image recognition result of high-definition vision and the three-dimensional point cloud data of lidar to a unified coordinate system through sensor joint calibration, and projecting the visual semantic label onto the three-dimensional point cloud to realize point cloud coloring and three-dimensional semantic reconstruction.

[0042] Based on the 3D semantic reconstruction results, the weed point cloud within each crop belt was statistically analyzed, and the weed coverage rate ρ was calculated using the following formula:

[0043]

[0044] Where, N total N represents the total point cloud count within a certain crop zone. weed This represents the number of point clouds marked as weeds within this range. The system determines the weed density level based on the ρ value.

[0045] The beneficial effects of this invention are:

[0046] 1. The flexible isolation cover, combined with the magnetic seal, forms a continuous closed cavity, which improves the cross-drift of pesticide solution from a physical structure perspective and solves the problem of pesticide damage in corn-soybean intercropping.

[0047] 2. Intelligent self-adaptation and reduced operation difficulty: Automatic identification of crop strips, automatic row alignment, automatic height adjustment, and automatic pesticide distribution significantly reduce the operator's workload.

[0048] 3. One machine, two pesticides, independent operation in different areas: The same machine can spray two completely different herbicides at the same time without interference or mixing, greatly improving work efficiency.

[0049] 4. Variable-rate pesticide application and environmentally friendly: Based on real-time variable-rate pesticide application according to weed density, it saves 20%-30% of pesticides compared to traditional uniform application.

[0050] 5. High versatility and adaptability to multiple modes: The independent spray bar unit can slide and adjust the distance freely, quickly adapting to various strip composite planting modes such as "2:2", "4:2", and "6:4". Attached Figure Description

[0051] Figure 1 This is a front view schematic diagram of an intelligent zoned and strip-variable spraying machine for corn-soybean strip intercropping according to the present invention.

[0052] Figure 2 This is a side view schematic diagram of an intelligent zoned and strip-variable spraying machine for corn-soybean strip intercropping according to the present invention.

[0053] Figure 3 This is a front view of the independent spray bar unit and its installation position.

[0054] Figure 4 This is a side view of the independent spray bar unit and its installation location.

[0055] Figure 5 yes Figure 4 Enlarged diagram of point A in the middle.

[0056] Figure 6 This is a schematic diagram of the installation of the flexible isolation cover.

[0057] Figure 7 This is a flowchart of the intelligent identification module and the partition variable control module. Detailed Implementation

[0058] 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 embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] Specific implementation method one: Combining Figures 1 to 7 This embodiment describes an intelligent zoned and variable-rate spraying machine for corn-soybean strip intercropping. It includes a self-propelled chassis 1, a suspension frame 2, a spray tank 3, and a central control system. The suspension frame 2 is installed on the front side of the self-propelled chassis 1, and the spray tank 3 is installed on the upper rear side of the self-propelled chassis 1. A zoned spraying mechanism is installed on the suspension frame 2.

[0060] The strip-type spraying mechanism includes a transverse main beam 4, multiple sets of independent spray bar units, an intelligent identification module, and a zoned variable control module; the multiple sets of independent spray bar units are arranged and installed along the transverse main beam 4.

[0061] Multiple independent spray bar units can be independently slidably arranged along the length of the transverse main beam 4. Each independent spray bar unit can be moved to a designated position and correspond to a crop planting strip in the field by means of transverse sliding and / or vertical extension under the drive of the partition variable control module. Each independent spray bar unit is equipped with an independent atomizing component.

[0062] The central control system is electrically connected to the intelligent identification module and the partition variable control module, respectively.

[0063] The intelligent recognition module is used to collect field images and three-dimensional point cloud data in real time, identify crop planting zone type, crop planting zone boundary and weed distribution, and output the recognition results to the central control system.

[0064] The central control system is used to receive the recognition results output by the intelligent recognition module and issue task instructions to the partition variable control module according to the recognition results; the partition variable control module drives each group of independent spray bar units to move according to the task instructions and outputs variable application control instructions to the independent atomizing components according to the weed distribution.

[0065] A flexible isolation cover 5 is installed on the outside of the independent spray bar unit. The side edge of the flexible isolation cover 5 is embedded with a flexible magnetic strip 6. The flexible isolation cover 5 can be lowered to the working position along with the independent spray bar unit. The flexible magnetic strips 6 of adjacent flexible isolation covers 5 can attract and seal each other to form an isolation barrier.

[0066] The "2 rows of corn + 4 rows of soybeans" strip intercropping model will be explained in further detail.

[0067] The field was planted with a 2-row corn + 4-row soybean strip intercropping system, with a total operating width of 2.7m. The system included: 2 sets of independent spray boom units specifically for corn and 4 sets of independent spray boom units specifically for soybeans; the pesticide tank system consisted of separate tanks for corn and soybeans, with independent supply lines for each.

[0068] Work process:

[0069] The central control system loads planting parameters: corn strip width 40cm, soybean strip width 90cm, strip spacing 70cm. After initial visual calibration, the system automatically identifies crop strip boundaries and heights. LiDAR measures the corn canopy height at 80cm and the soybean height at 30cm.

[0070] The electric actuator of the corn unit moves the flexible isolation cover 10cm away from the corn canopy; the electric actuator of the soybean unit moves the flexible isolation cover 8cm away from the soybean canopy. The flexible magnetic strips at the edges of adjacent isolation covers automatically close, forming three continuous and enclosed application spaces.

[0071] Application control: The broadleaf herbicide pipeline is opened in the corn strip unit, and the flow rate is adjusted according to the weed density using PWM variable flow rate; the grass herbicide pipeline is opened in the soybean strip unit, and the flow rate is adjusted according to the weed density using PWM variable flow rate.

[0072] Once the operation is complete, the electric push rod lifts, the flexible magnetic strip separates, and the machine is moved to another location.

[0073] Field tests showed no pesticide drift, no pesticide damage, and a weeding effect of over 95%, while reducing pesticide usage by approximately 25%.

[0074] Specific Implementation Method Two: Combining Figures 1 to 6 This embodiment differs from specific embodiment one in that a slide rail 7 is provided on the front side of the transverse main beam 4, and a rack is provided on the front side of the slide rail 7. The independent spray bar unit includes a slider 8, a stepper motor 9, and a gear 10. The slider 8 is slidably connected to the slide rail 7 and has a hollow channel. The gear 10 is arranged in the channel and meshes with the rack. The stepper motor 9 is located on the upper side of the slider 8, and a mounting hole is provided on the top surface of the slider 8. The output end of the stepper motor 9 is connected to the input end of the gear 10 through the mounting hole. The stepper motor 9 drives the independent spray bar unit to slide along the transverse main beam 4 by driving the gear 10 to mesh with the rack. Other components and connection methods are the same as in specific embodiment one.

[0075] Specific implementation method three: Combining Figures 1 to 6This embodiment differs from specific embodiment one in that the independent spray bar unit also includes an electric push rod 11. The electric push rod 11 is installed on the front side of the slider 8, and the independent atomizing component is installed at the bottom end of the electric push rod 11. A connecting rod 12 is installed on the lower part of the electric push rod 11, and the flexible isolation cover 5 is installed on the outside of the independent spray bar unit through the connecting rod 12. The installation method is simple and reliable, and easy to maintain. Other components and connection methods are the same as in specific embodiment one or two.

[0076] Specific implementation method four: Combination Figures 1 to 7 This embodiment differs from Specific Embodiment 1 in that the intelligent recognition module includes a high-definition visual camera and a lidar; the lidar is used to collect three-dimensional point cloud data, and the high-definition visual camera is used to collect field images. The two work together to identify the crop planting zone type, boundary, and weed distribution through a multi-source sensor fusion algorithm.

[0077] 1. LiDAR point cloud processing and crop strip identification

[0078] The lidar system acquires real-time 3D point cloud data of the work scene ahead. First, the system transforms the point cloud from the lidar coordinate system to the vehicle's world coordinate system based on a calibrated extrinsic parameter matrix, and then performs voxel filtering and downsampling. Subsequently, the ground clearance H of each point is calculated. i When the visual recognition of the current area is a corn belt, Hth is taken as 1 / 2 of the average height of the corn canopy in that area. When the visual recognition of the current area is a soybean belt, Hth is taken as 1 / 2 of the average height of the soybean canopy in that area. Points with Hi≥Hth are marked as candidate points for high-level crops, and points with 0.05m≤Hi<Hth are marked as candidate points for low-level background, thus achieving preliminary segmentation between the crop layer and the ground background.

[0079] For the segmented crop point clouds, a particle swarm optimization algorithm is used to extract single-row crop point cloud clusters. Then, the least squares method is used to fit a linear equation y=ax+b to each point cloud cluster, and the error function is minimized.

[0080]

[0081] Output the equations of the centerline and boundary line of each crop zone to achieve precise positioning of the crop zone.

[0082] 2. Deep learning-based visual recognition algorithms

[0083] Deep learning-based visual recognition algorithms

[0084] High-definition visual cameras acquire field images in real time. The system uses an instance segmentation model as its basic algorithm framework to perform end-to-end pixel-level recognition of images. The model extracts multi-scale features of the images through a backbone network and embeds a coordinate attention mechanism during the feature extraction process, embedding positional information into channel attention to enhance the model's spatial sensitivity to the extension and width directions of crop strips. At the same time, a spatial and channel reconstruction convolution module is introduced to adaptively suppress interference from background noise such as soil, light, and shadow.

[0085] The model outputs three task branches in parallel using the instance splitting head:

[0086] Semantic segmentation branch: Outputs a semantic segmentation mask for crop strips, distinguishing corn strips, soybean strips, and background regions at the pixel level;

[0087] Target detection branch: Outputs the bounding box and confidence score of the weed targets;

[0088] Instance segmentation branch: Outputs pixel-level segmentation masks of weed targets, enabling precise localization of the position and outline of individual weeds.

[0089] 3. Multi-source data fusion and weed density calculation

[0090] The system maps the 2D image recognition results of high-definition vision and the 3D point cloud data of LiDAR to a unified coordinate system through joint sensor calibration, and projects the visual semantic labels onto the 3D point cloud to achieve point cloud coloring and 3D semantic reconstruction.

[0091] Based on the 3D semantic reconstruction results, the weed point cloud within each crop belt was statistically analyzed, and the weed coverage rate ρ was calculated using the following formula:

[0092]

[0093] in, This represents the total point cloud count within a certain crop zone. This represents the number of point clouds marked as weeds within this range. The system determines the weed density level based on the ρ value.

[0094] Specific Implementation Method Five: Combining Figures 1 to 7 This embodiment differs from specific embodiment one in that the medicine tank 3 includes a corn-specific medicine tank and a soybean-specific medicine tank, which are set independently; the strip-type spraying mechanism has two independent supply pipelines, which are respectively connected to the corn-specific medicine tank and the soybean-specific medicine tank; the independent atomizing component includes an atomizing nozzle and a controllable solenoid valve group; the controllable solenoid valve group has two independent valves, which are respectively connected to the two independent supply pipelines; the zone variable control module controls the opening and closing of the two independent valves, opening only one valve at a time, so that the atomizing nozzle sprays the special medicine for the corresponding crop planting zone.

[0095] The outlet of the corn-specific pesticide tank is sequentially connected to the first filter, the first diaphragm pump, and the first pressure regulator before being connected to the main corn pesticide supply line; the outlet of the soybean-specific pesticide tank is sequentially connected to the second filter, the second diaphragm pump, and the second pressure regulator before being connected to the main soybean pesticide supply line. The corn and soybean pesticide supply lines are arranged along the transverse main beam, and at each independent spray bar unit, they are connected to the corresponding solenoid valve inlet in the independent atomizing component of that unit via flexible branch pipes.

[0096] Specific Implementation Method Six: Combination Figures 1 to 7 This embodiment differs from Specific Embodiment 1 in that the variable application control command output by the partition variable control module to the independent atomizing component is a PWM pulse width modulation signal; the independent atomizing component adjusts the spray flow rate of the atomizing nozzle according to the duty cycle of the PWM pulse width modulation signal.

[0097] Specific implementation method seven: Combination Figures 1 to 7 This embodiment differs from Specific Embodiment 1 in that the intelligent identification module calculates the weed coverage rate ρ and determines the weed density level based on the ρ value:

[0098] When ρ < 5%, it is level 0, and the duty cycle of the PWM pulse width modulation signal is 0%.

[0099] When 5%≤ρ<20%, ​​it is level 1, and the duty cycle of the PWM pulse width modulation signal is 30%-50%;

[0100] When 20%≤ρ<50%, it is level 2, and the duty cycle of the PWM pulse width modulation signal is 50%-80%;

[0101] When ρ≥50%, it is level 3, and the duty cycle of the PWM pulse width modulation signal is 80%-100%.

[0102] Specific implementation method eight: Combination Figures 1 to 7 This embodiment describes a control method, which includes the following steps:

[0103] S1. Pre-operation calibration and path generation: The planting parameters of the field to be operated are entered through the central control system, the sprayer is controlled to travel to the starting position of the field, the boundary of the first row of crops is calibrated, and an automatic operation path is generated.

[0104] S2. Real-time perception and multi-source fusion identification: During the process of the sprayer moving along the automatic operation path, the intelligent identification module collects field images and three-dimensional point cloud data in real time, and identifies the crop planting zone type, crop planting zone boundary coordinates and weed distribution based on the multi-source sensing fusion algorithm.

[0105] S3. Automatic row alignment and vertical adaptive adjustment: The central control system issues row alignment instructions to the partition variable control module based on the crop planting zone boundary coordinates identified by S2. The partition variable control module drives each group of independent spray bar units to slide along the transverse main beam 4 to directly above the corresponding crop planting zone, and drives each vertical lifting and vertical extension according to the crop canopy height, so that the bottom of the flexible isolation cover 5 maintains a preset distance from the crop canopy or the ground.

[0106] S4. Magnetic sealing safety interlock: When each group of independent spray bar units descends to the working position, the flexible magnetic strips 6 on the edges of adjacent flexible isolation covers 5 adhere to each other, forming a continuous sealed physical isolation barrier spanning the distance between adjacent crop planting belts. At the same time, the control system detects the sealing status through Hall sensors located on the edges of the flexible isolation covers 5.

[0107] S5. Dynamic calculation of weed coverage and variable application of pesticides: If a sealed barrier is detected to have been formed, the central control system dynamically calculates the weed coverage ρ in each crop planting zone based on the weed distribution identified in S2.

[0108] The corresponding PWM duty cycle is determined based on the ρ value, and a variable drug delivery control command containing this PWM duty cycle is issued to the partition variable control module; the partition variable control module drives the independent atomizing component to deliver the drug according to the command; wherein:

[0109] When ρ < 5%, it is level 0, the duty cycle of the PWM pulse width modulation signal is 0%, and no pesticide is applied;

[0110] When 5%≤ρ<20%, ​​it is level 1, the duty cycle of the PWM pulse width modulation signal is 30%-50%, and the dosage is low.

[0111] When 20%≤ρ<50%, it is level 2, the duty cycle of the PWM pulse width modulation signal is 50%-80%, and the dosage is medium.

[0112] When ρ≥50%, it is level 3, the duty cycle of the PWM pulse width modulation signal is 80%-100%, and high-volume application is allowed.

[0113] Specific Implementation Method Nine: Combining Figures 1 to 7 This embodiment differs from specific embodiment eight in that the intelligent recognition module includes a high-definition visual camera and a LiDAR, which collaborate in recognition through a multi-source fusion strategy.

[0114] The recognition process of the high-definition visual camera includes:

[0115] Real-time field images are acquired, and an instance segmentation model is used for pixel-level image recognition. The model extracts multi-scale features from the images through a backbone network, embedding a coordinate attention mechanism during the feature extraction process to enhance the model's spatial sensitivity to the extension and width directions of crop strips and suppress background noise interference from soil, light, and shadow. The model outputs semantic segmentation, target detection, and instance segmentation branches in parallel through the instance segmentation head to distinguish between corn strips, soybean strips, and background areas, and outputs the location and outline of weed targets. The classification results of this model serve as the final basis for determining the crop strip type.

[0116] The identification process of the lidar includes:

[0117] The system collects 3D point cloud data of the working scene in real time, transforms the point cloud from the radar coordinate system to the vehicle world coordinate system, and performs downsampling processing. It calculates the ground clearance Hi of each point and dynamically sets a height threshold Hth based on the current crop type fed back by the high-definition visual camera: when the visual recognition identifies the current area as a corn belt, Hth is taken as half the average height of the corn canopy in that area; when the visual recognition identifies the current area as a soybean belt, Hth is taken as half the average height of the soybean canopy in that area. Points with Hi≥Hth are marked as candidate points for high-level crops, and points with 0.05m≤Hi<Hth are marked as candidate points for low-level background, achieving preliminary segmentation between the crop layer and the ground background. A clustering algorithm is used to extract single-row crop point cloud clusters from the segmented high-level point cloud, and the least squares method is used to fit a straight line equation for each point cloud cluster, outputting the centerline and boundary line equations of each crop belt.

[0118] Specific Implementation Method Ten: Combining Figures 1 to 7 This embodiment differs from specific embodiment eight in that the specific steps of the intelligent recognition module in calculating the weed coverage rate ρ include: mapping the two-dimensional image recognition results of high-definition vision and the three-dimensional point cloud data of lidar to a unified coordinate system through joint calibration of sensors, and projecting the visual semantic labels onto the three-dimensional point cloud to realize point cloud coloring and three-dimensional semantic reconstruction.

[0119] Based on the 3D semantic reconstruction results, the weed point cloud within each crop belt was statistically analyzed, and the weed coverage rate ρ was calculated using the following formula:

[0120]

[0121] Where, N total N represents the total point cloud count within a certain crop zone. weed This represents the number of point clouds marked as weeds within this range. The system determines the weed density level based on the ρ value.

[0122] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent zoned and variable-rate sprayer for corn-soybean strip intercropping, comprising a self-propelled chassis (1), a suspension frame (2), a spray tank (3), and a central control system, wherein the suspension frame (2) is installed on the front side of the self-propelled chassis (1), and the spray tank (3) is installed on the upper rear side of the self-propelled chassis (1), characterized in that, The suspension frame (2) is equipped with a strip-type pesticide application mechanism; The strip-type spraying mechanism includes a transverse main beam (4), multiple sets of independent spray bar units, an intelligent identification module, and a zone variable control module; the multiple sets of independent spray bar units are arranged and installed along the transverse main beam (4); Multiple independent spray bar units can be independently slidably set along the length of the transverse main beam (4). Each independent spray bar unit can be moved to a designated position and correspond to a crop planting strip in the field by means of transverse sliding and / or vertical extension under the drive of the partition variable control module. Each independent spray bar unit is equipped with an independent atomizing component. The central control system is electrically connected to the intelligent identification module and the partition variable control module, respectively. The intelligent recognition module is used to collect field images and three-dimensional point cloud data in real time, identify crop planting zone type, crop planting zone boundary and weed distribution, and output the recognition results to the central control system. The central control system is used to receive the recognition results output by the intelligent recognition module and issue task instructions to the partition variable control module based on the recognition results; The partition variable control module drives each group of independent spray bar units to move according to the task instructions and outputs variable application control instructions to the independent atomizing components according to the distribution of weeds. A flexible isolation cover (5) is installed on the outside of the independent spray bar unit. A flexible magnetic strip (6) is embedded on the side edge of the flexible isolation cover (5). The flexible isolation cover (5) can be lowered to the working position along with the independent spray bar unit. The flexible magnetic strips (6) of adjacent flexible isolation covers (5) can attract and seal each other to form an isolation barrier.

2. The intelligent zoned and variable-rate spraying machine for corn-soybean strip intercropping according to claim 1, characterized in that, A slide rail (7) is provided on the front side of the transverse main beam (4), and a rack is provided on the front side of the slide rail (7). The independent spray bar unit includes a slider (8), a stepper motor (9) and a gear (10). The slider (8) is slidably connected to the slide rail (7). The slider (8) is provided with a hollow channel. The gear (10) is arranged in the channel and meshes with the rack. The stepper motor (9) is arranged on the upper side of the slider (8). The top surface of the slider (8) is provided with a mounting hole. The output end of the stepper motor (9) is connected to the input end of the gear (10) through the mounting hole. The stepper motor (9) drives the independent spray bar unit to slide along the transverse main beam (4) by meshing with the rack through the drive gear (10).

3. The intelligent zoned and variable-rate spraying machine for corn-soybean strip intercropping according to claim 2, characterized in that, The independent spray bar unit also includes an electric push rod (11), which is installed on the front side of the slider (8). The independent atomizing component is installed at the bottom of the electric push rod (11). A connecting rod (12) is installed on the lower part of the electric push rod (11). The flexible isolation cover (5) is installed on the outside of the independent spray bar unit through the connecting rod (12).

4. The intelligent zoned and variable-rate spraying machine for corn-soybean strip intercropping according to claim 1, characterized in that, The intelligent recognition module includes a high-definition visual camera and a lidar; the lidar is used to collect three-dimensional point cloud data, and the high-definition visual camera is used to collect field images. The two work together to identify the crop planting zone type, boundary and weed distribution through a multi-source sensor fusion algorithm.

5. The intelligent zoned and variable-rate spraying machine for 5-meter soybean strip intercropping according to claim 1, characterized in that, The medicine box (3) includes a corn-specific medicine box and a soybean-specific medicine box, which are set independently; the strip-type application mechanism is provided with two independent medicine supply pipelines, which are respectively connected to the corn-specific medicine box and the soybean-specific medicine box; the independent atomizing component includes an atomizing nozzle and a controllable solenoid valve group; the controllable solenoid valve group is provided with two independent valves, which are respectively connected to the two independent medicine supply pipelines. The partition variable control module controls the opening and closing of two independent valves, opening only one valve at a time to allow the atomizing nozzle to spray the special agent for the corresponding crop planting zone.

6. The intelligent zoned and variable-rate spraying machine for corn-soybean strip intercropping according to claim 1, characterized in that, The variable application control command output by the partition variable control module to the independent atomizing component is a PWM pulse width modulation signal; the independent atomizing component adjusts the spray flow rate of the atomizing nozzle according to the duty cycle of the PWM pulse width modulation signal.

7. The intelligent zoned and variable-rate spraying machine for corn-soybean strip intercropping according to claim 6, characterized in that, The intelligent identification module calculates the weed coverage rate ρ and determines the weed density level based on the ρ value: When ρ < 5%, it is level 0, and the duty cycle of the PWM pulse width modulation signal is 0%. When 5%≤ρ<20%, ​​it is level 1, and the duty cycle of the PWM pulse width modulation signal is 30%-50%; When 20%≤ρ<50%, it is level 2, and the duty cycle of the PWM pulse width modulation signal is 50%-80%; When ρ≥50%, it is level 3, and the duty cycle of the PWM pulse width modulation signal is 80%-100%.

8. A control method, said control method being based on an intelligent zoned and variable-rate sprayer for corn-soybean strip intercropping as described in any one of claims 1 to 7, characterized in that, Includes the following steps: S1. Pre-operation calibration and path generation: The planting parameters of the field to be operated are entered through the central control system, the sprayer is controlled to travel to the starting position of the field, the boundary of the first row of crops is calibrated, and an automatic operation path is generated. S2. Real-time perception and multi-source fusion identification: During the process of the sprayer moving along the automatic operation path, the intelligent identification module collects field images and three-dimensional point cloud data in real time, and identifies the crop planting zone type, crop planting zone boundary coordinates and weed distribution based on the multi-source sensing fusion algorithm. S3. Automatic row alignment and vertical adaptive adjustment: The central control system issues row alignment instructions to the partition variable control module based on the crop planting zone boundary coordinates identified in S2; the partition variable control module drives each group of independent spray bar units to slide along the transverse main beam (4) to the top of the corresponding crop planting zone, and drives each vertical lifting and vertical extension according to the crop canopy height, so that the bottom of the flexible isolation cover (5) maintains a preset distance from the crop canopy or the ground. S4, magnetic sealing safety interlock: When each group of independent spray bar units descends to the working position, the flexible magnetic strips (6) on the edge of the adjacent flexible isolation cover (5) adhere to each other and form a continuous closed physical isolation barrier spanning the distance between adjacent crop planting belts. At the same time, the control system detects the sealing status through the Hall sensor located on the edge of the flexible isolation cover (5). S5. Dynamic calculation of weed coverage and variable application of pesticides: If a sealed barrier is detected to have been formed, the central control system dynamically calculates the weed coverage ρ in each crop planting zone based on the weed distribution identified in S2. The corresponding PWM duty cycle is determined based on the ρ value, and the variable drug application control command containing the PWM duty cycle is issued to the partition variable control module. The partition variable control module drives the independent atomizing component to administer medication according to the instruction; wherein: When ρ < 5%, it is level 0, the duty cycle of the PWM pulse width modulation signal is 0%, and no pesticide is applied; When 5%≤ρ<20%, ​​it is level 1, the duty cycle of the PWM pulse width modulation signal is 30%-50%, and the dosage is low. When 20%≤ρ<50%, it is level 2, the duty cycle of the PWM pulse width modulation signal is 50%-80%, and the dosage is medium. When ρ≥50%, it is level 3, the duty cycle of the PWM pulse width modulation signal is 80%-100%, and high-volume application is allowed.

9. The control method according to claim 8, characterized in that, The intelligent recognition module includes a high-definition visual camera and a LiDAR, which work together to identify targets using a multi-source fusion strategy. The recognition process of the high-definition visual camera includes: Real-time field images are acquired, and an instance segmentation model is used for pixel-level image recognition. The model extracts multi-scale features from the images through a backbone network, embedding a coordinate attention mechanism during the feature extraction process to enhance the model's spatial sensitivity to the extension and width directions of crop strips and suppress background noise interference from soil, light, and shadow. The model outputs semantic segmentation, target detection, and instance segmentation branches in parallel through the instance segmentation head to distinguish between corn strips, soybean strips, and background areas, and outputs the location and outline of weed targets. The classification results of this model serve as the final basis for determining the crop strip type. The identification process of the lidar includes: The system collects 3D point cloud data of the working scene in real time, transforms the point cloud from the radar coordinate system to the vehicle world coordinate system, and performs downsampling processing. It calculates the ground clearance Hi of each point and dynamically sets a height threshold Hth based on the current crop type fed back by the high-definition visual camera: when the visual recognition identifies the current area as a corn belt, Hth is taken as half the average height of the corn canopy in that area; when the visual recognition identifies the current area as a soybean belt, Hth is taken as half the average height of the soybean canopy in that area. Points with Hi≥Hth are marked as candidate points for high-level crops, and points with 0.05m≤Hi<Hth are marked as candidate points for low-level background, achieving preliminary segmentation between the crop layer and the ground background. A clustering algorithm is used to extract single-row crop point cloud clusters from the segmented high-level point cloud, and the least squares method is used to fit a straight line equation for each point cloud cluster, outputting the centerline and boundary line equations of each crop belt.

10. The control method according to claim 8, characterized in that, The specific steps for the intelligent recognition module to calculate the weed coverage rate ρ include: mapping the two-dimensional image recognition results of high-definition vision and the three-dimensional point cloud data of lidar to a unified coordinate system through joint calibration of sensors, and projecting the visual semantic labels onto the three-dimensional point cloud to realize point cloud coloring and three-dimensional semantic reconstruction. Based on the 3D semantic reconstruction results, the weed point cloud within each crop belt was statistically analyzed, and the weed coverage rate ρ was calculated using the following formula: Where, N total N represents the total point cloud count within a certain crop zone. weed This represents the number of point clouds marked as weeds within this range. The system determines the weed density level based on the ρ value.

Images