A fruit tree canopy intelligent detection and precise pesticide application control method and system

Abstract

This application provides a method and system for intelligent detection and precise spraying control of fruit tree canopies, which is used to solve the problem of inaccurate spraying of liquid medicine in the existing technology. The specific steps of the method are: S1 collects image data of the side and top of the fruit tree canopy; S2 calculates the area and density of the side and top of the fruit tree canopy; S3 calculates the ground spraying decision coefficient K based on the area and density of the side and top of the fruit tree canopy g and aerial application decision coefficient K s , and calculates the ground and aerial application rates separately based on the application decision coefficient; S4 uses an ESO-based fuzzy adaptive control algorithm to accurately control the pipeline pressure and flow of ground and aerial application equipment. This application constructs an application rate calculation model by calculating the area and density of the sides and tops of fruit tree canopies, reducing the amount of pesticide applied to each fruit tree. At the same time, it uses an ESO-based fuzzy adaptive control algorithm to accurately control the pipeline dosage, improving the utilization rate of variable spray solution.

Description

Technical Field

[0001] The present invention relates to the field of agricultural pesticide application, and in particular to a method and system for intelligent detection and precise pesticide application control of fruit tree canopies. Background Art

[0002] Existing unmanned plant protection operations in hilly orchards face two difficulties: 1. The orchard's uneven terrain interferes with the stability of the spraying equipment; 2. The fruit trees are distributed irregularly, making conventional continuous spraying uneconomical and environmentally friendly. Variable-volume spraying requires ensuring adequate coverage of each tree. Existing spraying control methods and systems are unable to meet the precise spraying requirements of orchards in harsh terrain and complex planting patterns. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for intelligent detection and precise pesticide application control of fruit tree canopies. By combining plant protection control technology with machine vision technology to perform variable-rate pesticide application, the method solves the technical problem of inaccurate pesticide spraying in the existing technology.

[0004] A method for intelligent detection and precise pesticide application control of fruit tree canopies, comprising the following specific steps:

[0005] S1: Collect image data of the side and top of the fruit tree canopy;

[0006] S2: Calculate the area and density of the sides and top of the fruit tree canopy;

[0007] S3: Calculate the decision coefficient K for ground application based on the area and density of the sides and top of the fruit tree canopy g and aerial application decision coefficient K s , and calculate the ground and aerial application rates respectively according to the application decision coefficient;

[0008] S4: An ESO-based fuzzy adaptive control algorithm is used to accurately control the pipeline pressure and flow of ground and aerial spraying equipment respectively.

[0009] Optionally, the specific method for calculating the side area and top area of ​​the fruit tree canopy is:

[0010] The collected images of the side and top of the fruit tree canopy are gray-scaled and segmented to obtain the side and top contours of the fruit tree canopy, which only contain the leaf area.

[0011] Identify and count the pixel points in the leaf area of ​​the collected images of the side and top of the fruit tree canopy, find the pixel points of all leaves, and calculate the area of ​​the side and top of the fruit tree canopy:

[0012] S=A×s

[0013] Where S is the side or top area of ​​the fruit tree canopy, s is the actual area represented by a unit pixel in the side or top image of the fruit tree canopy, and A is the number of canopy leaf pixels in the side or top image of the fruit tree canopy.

[0014] Optionally, the specific steps for calculating the density of the side and top of the fruit tree canopy are as follows:

[0015] Morphological processing operations are used to fill the holes in the leaf areas on the sides and tops of the fruit tree canopies displayed after canopy segmentation, to obtain the contour area of ​​the fruit tree canopy after the holes are filled, and to calculate the density of the fruit tree canopy sides and tops:

[0016]

[0017] Where δ is the density of the side or top of the fruit tree canopy, S1 is the contour area of ​​the side or top of the fruit tree canopy after the holes are filled, and S is the area of ​​the side or top of the fruit tree canopy.

[0018] Optionally, construct the decision coefficient K for ground application g , aerial spraying decision coefficient K s for:

[0019] K g =0.8×(m Kg *S g +n Kg *δ g )+0.2×(m Ks *S s +n Ks *δ s )

[0020] K s =0.2×(m Kg *S g +n Kg *δ g )+0.8×(m Ks *S s +n Ks *δ s )

[0021] Where S g is the lateral area of ​​the fruit tree canopy, S s The top area of ​​the fruit tree canopy, δ g is the lateral density of the fruit tree canopy, δ s Density of fruit tree canopy top, m Kg 、m Ks 、n Kg 、n Ks are weight coefficients, m KgIt is the ratio of the actual value of the side area of ​​the canopy of the fruit tree to be measured to the average value of the side area of ​​the canopy of several fruit trees in the orchard, m Ks is the ratio of the actual value of the top area of ​​the canopy of the fruit tree to be measured to the average value of the total area of ​​the top areas of the canopies of several fruit trees in the orchard, n Kg is the ratio of the actual density of the canopy side of the fruit tree to be tested to the average density of the canopy side of several fruit trees in the orchard, n Ks It is the ratio of the actual density of the top of the canopy of the fruit tree to be tested to the average density of the top of the canopy of several fruit trees in the orchard.

[0022] Optionally, based on the ground application decision coefficient K g , aerial spraying decision coefficient K s , calculate the ground and aerial application rates as:

[0023] V=K·Q1·t

[0024] Where V is the amount of pesticide applied to the side or top of the fruit tree canopy, K is the decision coefficient for aerial or ground application, Q1 is the actual flow rate of the ground or aerial spray nozzle, and t is the duration of spraying.

[0025] Optionally, the specific steps for accurately controlling the pipeline pressure and flow of ground and aerial spraying equipment using an ESO-based fuzzy adaptive control algorithm are as follows:

[0026] S4.1: Determine the input variables and output variables of the control system and establish an inference system of the input variables, output variables, and fuzzy rules. The input variables of the control system are the deviation e and the rate of change of the deviation ec of the output variable. The output variables of the control system are the pipeline pressure and flow rate of the ground or aerial spraying equipment.

[0027] S4.2: Use the extended state observer (ESO) to estimate the system error signal and use the error signal as the input of the fuzzy inference system;

[0028] S4.3: Perform fuzzy inference on the fuzzy set of input variables and the error signal to obtain the output signal of the controller;

[0029] S4.4: Use the output signal of the controller to control the pipeline pressure and flow of the pesticide application equipment to obtain the actual output pressure and flow;

[0030] S4.5: Use the difference between the actual output and the desired output of the system as the input of the ESO, use the ESO to estimate the error signal and update the state of the error estimator;

[0031] S4.6: Use the feedback mechanism to continuously adjust the controller parameters to gradually optimize the spraying rate of the precision spraying control system.

[0032] A fruit tree canopy intelligent detection and precision pesticide application control system includes a ground-based automatic pesticide application device for collecting side image data of the fruit tree canopy and performing precision pesticide application, an aerial pesticide application drone for collecting top image data of the fruit tree canopy and performing precision pesticide application, and a remote monitoring and control module for data exchange with the ground-based automatic pesticide application device and the aerial pesticide application drone;

[0033] The ground automatic pesticide application equipment and the aerial pesticide application drone communicate data with each other. The ground automatic pesticide application equipment includes a mobile body, and the aerial pesticide application drone includes a flying drone. Both the mobile body and the flying drone are provided with machine vision components for collecting image data, as well as pesticide application components for precise pesticide application.

[0034] Optionally, the machine vision component includes a metal box, a camera installed in the metal box for collecting image data, and an image processing module installed in the metal box for processing and identifying the image data;

[0035] The camera and the image processing module are both in data communication with the first processor module.

[0036] Optionally, the pesticide application assembly includes a pesticide storage box and a spray pipe for spraying pesticide mist, and the spray pipe is provided with a plurality of pesticide spray nozzles;

[0037] The medicine storage box is connected to the liquid inlet end of the spraying pipe through a pumping mechanism, and the pumping mechanism is controlled by a control mechanism to accurately apply the medicine. The pumping control mechanism includes a high-pressure diaphragm pump, and the liquid inlet end of the high-pressure diaphragm pump is connected to the medicine storage box through a pipeline;

[0038] The liquid outlet end of the high-pressure diaphragm pump is connected to the first proportional solenoid valve and the second proportional solenoid valve through pipelines respectively. The first proportional solenoid valve is connected to the medicine storage box. A pressure transmitter is also provided on the pipeline connecting the high-pressure diaphragm pump and the second proportional solenoid valve. The liquid outlet end of the second proportional solenoid valve is connected to the liquid inlet end of the spray pipe through a pipeline. A turbine flowmeter is also provided on the pipeline connecting the second proportional solenoid valve and the spray pipe. The liquid inlet end of the spray pipe is provided with an electric switch valve for controlling its opening and closing.

[0039] Optionally, the control mechanism includes a second processor module, a current-to-voltage module for signal conversion, and a PWM drive module for controlling the first proportional solenoid valve and the second proportional solenoid valve, and a relay for controlling the electric switching valve;

[0040] The current-to-voltage module, PWM drive module and relay all communicate data with the second processor module.

[0041] Due to the adoption of the above technical solution, the present invention has the following advantages:

[0042] 1. This application establishes a three-way collaborative fruit tree canopy intelligent detection and precision pesticide application control system through the three-way collaboration of ground automatic pesticide application equipment, aerial pesticide application drones, and remote monitoring interfaces. Equipped with a dedicated image processing chip, a real-time adjustable pesticide application calculation model based on the external characteristics of the fruit tree canopy, and a precise pesticide application control algorithm, the system performs intelligent identification and precise pesticide application. The method of sharing fruit tree canopy characteristic parameters on the ground and in the air and applying pesticides simultaneously ensures the uniformity and effectiveness of pesticide application. The remote monitoring interface makes orchard pesticide application more intelligent by remotely controlling and monitoring the application process, saving farmers time for on-site surveys and reducing manual management costs. The adjustable pesticide application calculation model and precise pesticide application control algorithm reduce pesticide usage, further reducing farmers' economic costs and reducing the ecological pressure of residual pesticide decomposition.

[0043] 2. The present invention establishes a precise pesticide application model for fruit trees through visual processing means to achieve the purpose of precise pesticide application. First, the present invention adopts a machine vision device and is equipped with a dedicated Jetson nano image processing chip to identify the external features of fruit trees and carry out targeted pesticide application from the ground and the air. Compared with the indiscriminate spraying of one side of the canopy of fruit trees in the orchard, other areas of fruit trees, and non-fruit tree areas, it can greatly improve the accuracy of pesticide application. Secondly, a pesticide application calculation model is constructed by calculating the side area of ​​the canopy of fruit trees, the top area of ​​the canopy, the side density of the canopy, and the top density of the canopy based on machine vision, which realizes differentiated spraying between fruit trees and reduces the pesticide dosage of a single fruit tree.

[0044] 3. The present invention uses a fuzzy adaptive control algorithm based on ESO to accurately control the amount of medicine in the pipeline, thereby improving the utilization rate of variable spray liquid. The present invention simultaneously monitors the pressure and flow of the application pipeline, and uses an adaptive control algorithm to control the pressure and flow to make the pipeline liquid output more stable, thereby avoiding waste of liquid medicine. The liquid medicine nozzle of the single-sided spray output pipeline of the ground automatic pesticide application equipment has three-way liquid output, corresponding to spraying the upper, middle and lower layers of the fruit tree canopy; the liquid medicine nozzle of the aerial pesticide drone also has three-way liquid output, corresponding to spraying the outer, middle and inner layers of the fruit tree canopy. The liquid medicine nozzle ensures that the sides and top areas of the fruit tree canopy are fully covered while using a smaller amount of medicine, and the application of medicine is more uniform.

[0045] 4. This invention uses Internet of Things technology to achieve remote monitoring of precise pesticide application, further meeting the needs of plantation personnel for remote control via mobile phones. The designed remote monitoring interface is rich in functions. It not only meets the function of dynamic real-time data monitoring during the pesticide application process, providing strong data support for optimizing the real-time online pesticide application rate model calculation based on the external characteristic parameters of the fruit tree canopy, but also meets the function of remote pesticide application control, reducing the cost of on-site manual monitoring of the pesticide application device, and improving the efficiency of precise pesticide application in the orchard.

[0046] 5. The present invention uses a machine vision-based droplet deposition detection method to establish a spraying effect evaluation scheme after precise pesticide application, which can effectively measure the effect of orchard pesticide application. The three indicators of liquid coverage, droplet deposition point density, and liquid deposition rate are used to comprehensively measure the spraying effect of the fruit tree canopy intelligent detection and precise pesticide application control method and system described in this application, while meeting the requirements of liquid coverage greater than 30%, droplet deposition point density greater than 25 drops / cm 2 When the liquid deposition rate is greater than 85%, it is evaluated as an effective precision spraying system. Implementing the precision spraying effect evaluation in orchards can further reduce the amount of pesticide used in subsequent spraying processes while ensuring the spraying effect, thereby reducing costs and increasing efficiency.

[0047] Other advantages, objects, and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art upon examination of the following description or may be learned from practice of the present invention. The objects and other advantages of the present invention may be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The accompanying drawings of the present invention are described below.

[0049] Figure 1 This is a flow chart of the fruit tree canopy intelligent detection and precise pesticide application control method of the present invention.

[0050] Figure 2 This is a structural diagram of the fruit tree canopy intelligent detection and precise pesticide application control system of the present invention.

[0051] Figure 3 This is a schematic structural diagram of the automatic ground pesticide application equipment of the present invention.

[0052] Figure 4 The figure is a schematic structural diagram of the aerial pesticide spraying UAV of the present invention.

[0053] Figure 5 This is a schematic diagram of the internal structure of the automatic ground pesticide application equipment.

[0054] Figure 6 It is a structural schematic diagram of the pesticide application component in the automatic ground pesticide application equipment of the present invention.

[0055] Figure 7 This is a schematic diagram of the connection of the liquid medicine pipeline of the medicine dispensing assembly of the present invention.

[0056] Figure 8 This is a signal transmission diagram of the fruit tree canopy intelligent detection and precise pesticide application control system of the present invention.

[0057] Figure 9This is a structural diagram of the precise pesticide application process controlled by the ESO-based fuzzy adaptive controller of the present invention.

[0058] Figure 10 The figure is a flow chart of the method for evaluating the effect of precise pesticide application.

[0059] In the figure: 1- ground automatic pesticide application equipment; 2- aerial pesticide application drone; 101- mobile body; 102- chassis control cabinet; 103- tire; 104- first bracket; 105- first base plate; 106- second bracket; 107- second base plate; 108- shell; 109- liquid filling avoidance hole; 2- aerial pesticide application drone; 201- flying drone; 3- machine vision component; 301- metal box; 302- camera; 303- first metal rod; 4- pesticide application component; 401- drug storage box; 402- spray pipe; 403- liquid spray head; 404- high-pressure diaphragm pump; 405- first proportional solenoid valve; 406- second proportional solenoid valve; 407- pressure transmitter; 408- turbine flowmeter; 409- electric switch valve; 410- liquid filling port; 411- control box; 412- second metal rod; 5- remote terminal. DETAILED DESCRIPTION

[0060] The present invention will be further described below with reference to the accompanying drawings and examples.

[0061] Example 1:

[0062] like Figure 2 、 Figure 3 and Figure 4 The system shows an intelligent detection and precise pesticide application control system for a fruit tree canopy, comprising a ground-based automatic pesticide application device 1 for collecting image data of the side of the fruit tree canopy and performing precise pesticide application, an aerial pesticide application drone 2 for collecting image data of the top of the fruit tree canopy and performing precise pesticide application, and a remote monitoring and control module for data exchange with the ground-based automatic pesticide application device 1 and the aerial pesticide application drone 2.

[0063] The ground automatic pesticide application equipment 1 and the aerial pesticide application drone 2 communicate data with each other. The ground automatic pesticide application equipment 1 includes a mobile body 101, and the aerial pesticide application drone 2 includes a flying drone 201. Both the mobile body 101 and the flying drone 201 are provided with a machine vision component 3 for collecting image data, and a pesticide application component 4 for precise pesticide application.

[0064] In this embodiment, the ground automatic pesticide application equipment 1 includes a chassis control cabinet 102, and a tire 103 installed at the bottom of the chassis control cabinet 102. A first bracket 104 is provided on the chassis control cabinet 102, a first base plate 105 is installed on the first bracket 104, a second bracket 106 is installed on the first base plate 105, and a second base plate 107 is installed on the second bracket 106. The first base plate 105 is also provided with a shell 108, and the medicine storage box 401 and the pumping control mechanism are located in the shell 108. A liquid filling avoidance hole 109 for the medicine storage box 401 is provided on the top of the shell 108.

[0065] In this embodiment, during operation, the machine vision component 3 on the mobile vehicle 101 collects and processes image data of the side of the fruit tree canopy, and transmits the data to the pesticide application component 4 on the mobile vehicle 101. The pesticide application component 4 calculates the pesticide application amount based on the image data of the side of the fruit tree canopy and accurately controls the pesticide application on the side of the fruit tree canopy. The machine vision component 3 on the flying drone 201 collects and processes image data of the top of the fruit tree canopy, and transmits the data to the pesticide application component 4 on the flying drone 201. The pesticide application component 4 calculates the pesticide application amount based on the image data of the top of the fruit tree canopy and accurately controls the pesticide application on the top of the fruit tree canopy.

[0066] like Figure 2 、 Figure 3 、 Figure 4 、 Figure 5 and Figure 6 The machine vision component 3 shown includes a metal box 301, a camera 302 installed in the metal box 301 for collecting image data, and an image processing module installed in the metal box 301 for processing and identifying the image data;

[0067] The camera 301 and the image processing module are both in data communication with the first processor module.

[0068] In this embodiment, the image processing module uses a Jetson nano image processing module, the first processor module uses a single-chip microcomputer, and the machine vision component 3 also includes a power module. Two hollow first metal rods 303 are mounted on either side of one end of the mobile body 101. Machine vision components 3 are mounted on top of each of these first metal rods 303. As the mobile body 101 moves, the two machine vision components 3 capture side images of the fruit trees on either side of the mobile body 101. On the flying drone 201, the machine vision component 3 is mounted on the front side of the bottom of the drone 201.

[0069] In this embodiment, the first processor module transmits the image data processed by the image processing module to the second processor module of the dispensing assembly 4 via serial communication. In this embodiment, the housing 108 is provided with a threading hole, and the data transmission line of the first processor module passes through the hollow first metal rod 303 and the threading hole of the housing 108 in sequence to communicate with the second processor module.

[0070] like Figure 2 、 Figure 3 、 Figure 4 、 Figure 5 and Figure 6 As shown, the medicine application assembly 4 includes a medicine storage box 401 and a spray pipe 402 for spraying medicine mist, and the spray pipe 402 is provided with a plurality of medicine liquid spray heads 403;

[0071] The medicine storage box 401 is connected to the liquid inlet end of the spraying pipe 402 via a pumping mechanism, and the pumping mechanism is controlled by a control mechanism to accurately apply the medicine.

[0072] In this embodiment, two hollow second metal rods 412 are mounted on either side of the other end of the mobile body 101. Two spraying pipes 402 are mounted on the top of each of the second metal rods 412. The two spraying pipes 402 are used to spray the sides of the fruit trees on both sides of the mobile body 101. The medicine storage box 401 is mounted on the first base plate 104. On the flying drone 201, the medicine storage box 401 is mounted on the bottom of the flying drone 201. The flying drone 201 is also equipped with two spraying pipes 402 for spraying the top layers of the fruit trees below the flying drone 201.

[0073] In this embodiment, a liquid filling port is provided on each medicine storage box 401. The liquid filling port of the medicine storage box 401 on the mobile body 101 is provided on its top, and the liquid filling port 410 of the medicine storage box 401 on the flying drone 201 is provided on its bottom. By providing the liquid filling port 410, it is convenient to place medicine liquids of different varieties and concentrations according to the actual situation of the orchard.

[0074] like Figure 2 、 Figure 3 、 Figure 4 、 Figure 5 and Figure 6 As shown, the pumping control mechanism includes a high-pressure diaphragm pump 404, and the liquid inlet end of the high-pressure diaphragm pump 404 is connected to the medicine storage box 401 through a pipeline;

[0075] The liquid outlet of the high-pressure diaphragm pump 404 is connected to a first proportional solenoid valve 405 and a second proportional solenoid valve 406 via pipelines, respectively. The first proportional solenoid valve 405 is in communication with the medicine storage tank 401. A pressure transmitter 407 is further provided on the pipeline connecting the high-pressure diaphragm pump 404 and the second proportional solenoid valve 406. The liquid outlet of the second proportional solenoid valve 406 is in communication with the liquid inlet of the spray pipe 402 via a pipeline. A turbine flowmeter 408 is further provided on the pipeline connecting the second proportional solenoid valve 406 and the spray pipe 402.

[0076] The liquid inlet end of the spray pipe 402 is provided with an electric switch valve 409 for controlling the opening and closing thereof.

[0077] In this embodiment, if Figure 5 、 Figure 7 and Figure 8 As shown, the high-pressure diaphragm pump 404 extracts the medicine liquid in the medicine storage box 401, and most of the medicine liquid flows to the pressure transmitter 407 through the pipeline, and the excess medicine liquid returns to the medicine storage box 401 through the first proportional solenoid valve 405. The pressure transmitter 407 measures the pipeline pressure signal, and the medicine liquid flows out through the pressure transmitter 407 to the second proportional solenoid valve 406, and then flows through the pipeline to the turbine flowmeter 408. The turbine flowmeter 408 measures the pipeline flow signal; the medicine liquid flows out through the turbine flowmeter 408 to the spray pipe 402.

[0078] In this embodiment, the electric switch valve 409 avoids waste of liquid medicine caused by continuous application of pesticides by adjusting the spray on and off time of the liquid medicine nozzle 403; the liquid medicine nozzle 403 on the spray pipe 402 is a three-layer structure, and the three liquid medicine nozzles 403 are arranged in sequence along the vertical direction to achieve simultaneous spraying of the upper, middle and lower layers of the side of the fruit tree canopy, or to achieve simultaneous spraying of the inner, middle and outer layers of the top of the fruit tree canopy.

[0079] like Figure 2 、 Figure 3 、 Figure 4 、 Figure 5 and Figure 6 As shown, the control mechanism includes a second processor module, a current-to-voltage module for signal conversion, a PWM drive module for controlling the first proportional solenoid valve 405 and the second proportional solenoid valve 406, and a relay for controlling the electric switch valve 409;

[0080] The current-to-voltage module, PWM drive module and relay all communicate data with the second processor module.

[0081] In this embodiment, if Figure 5 、 Figure 7 and Figure 8As shown, the control mechanism is installed in a control box 411. The second processor module communicates with the remote monitoring control module via WIFI communication technology. The second processor module on the mobile frame 101 also communicates with the second processor module on the flying drone via WIFI communication technology. On the mobile frame 1, the control box 411 is mounted on the second base plate 106. On the flying drone 201, the control box 411 is mounted on the top of the flying drone 201.

[0082] In this embodiment, the second processor module receives the fruit tree image data transmitted by the first processor module, segments the fruit tree canopy portion of the image, and removes the remaining complex background. The second processor module calculates the lateral area and lateral density of the fruit tree canopy within the image and fits the required ground application rate for the fruit tree based on the pesticide rate calculation model based on the fruit tree canopy characteristic parameters.

[0083] In this embodiment, there are two current-to-voltage modules. The second processor module converts the required dosage into pressure and flow signals, and controls the first proportional solenoid valve 405 and the second proportional solenoid valve 406 via the PWM drive module.

[0084] In this embodiment, the pipeline pressure signal of the pressure transmitter 407 is transmitted to the second processor module via a current-to-voltage module, and the pipeline flow signal of the turbine flowmeter 408 is transmitted to the second processor module via another current-to-voltage module. The second processor module compares the pressure and flow signals of the required amount of pesticide application for the fruit trees with the errors between the feedback pipeline pressure signal and pipeline flow signal; and controls the first proportional solenoid valve 405 and the second proportional solenoid valve 406 again through the PWM drive module with the error signal, so that the pipeline pressure and flow are infinitely close to the calculated values ​​of the second processor module.

[0085] As an embodiment of the present invention, the remote monitoring and control module is loaded on the remote terminal 5, and the remote monitoring and control module includes a pressure setting submodule, a flow setting submodule, a control mode switching submodule and a spraying mode switching submodule; the initial pressure, flow and other parameters of the system are set through the control interface of the remote terminal. At the same time, the remote terminal control interface also displays 7 real-time data including the driving path, crown area, crown density, model spraying amount, flow change, pressure change and video interface of the ground automatic spraying equipment 1 and the aerial spraying drone 2, thereby realizing dynamic monitoring of the spraying process.

[0086] Example 2:

[0087] like Figure 1 The method for intelligent detection and precise pesticide application control of a fruit tree canopy is shown, which adopts the intelligent detection and precise pesticide application control system of a fruit tree canopy described in Example 1. The specific steps are as follows:

[0088] S1: Collect image data of the side and top of the fruit tree canopy;

[0089] In this embodiment, the machine vision component 3 of the ground automatic pesticide application equipment 1 and the aerial pesticide application drone 2 in Example 1 respectively collects image data of the side and top of the fruit tree canopy.

[0090] S2: Calculate the area and density of the sides and top of the fruit tree canopy;

[0091] S2.1: The image data is processed using the Jetson nano image processing module. The collected images of the side and top of the fruit tree canopy are grayscaled and segmented to obtain the outline of the side and top of the fruit tree canopy. The outline only contains the leaf area.

[0092] S2.2: Identify and count the pixel points in the leaf area of ​​the collected images of the side and top of the fruit tree canopy, find all the pixel points of the leaves, and calculate the area of ​​the side and top of the fruit tree canopy:

[0093] S=A×s

[0094] Where S is the side or top area of ​​the fruit tree canopy, s is the actual area represented by a unit pixel in the side or top image of the fruit tree canopy, and A is the number of canopy leaf pixels in the side or top image of the fruit tree canopy.

[0095] S2.3: Use morphological processing operations to fill the holes in the leaf areas on the sides and top of the fruit tree canopy displayed after canopy segmentation, obtain the contour area of ​​the fruit tree canopy after the holes are filled, and calculate the density of the fruit tree canopy on the sides and top:

[0096]

[0097] Where δ is the density of the side or top of the fruit tree canopy, S1 is the contour area of ​​the side or top of the fruit tree canopy after the holes are filled, and S is the area of ​​the side or top of the fruit tree canopy.

[0098] In this embodiment, before the spraying operation, the ground automatic spraying equipment 1 and the aerial spraying drone 2 measure and calculate the relationship between the area represented by their unit pixels and the actual spraying distance of the liquid spray head 403 as follows:

[0099] s=0.49L 1.33

[0100] Where s is the actual area represented by a unit pixel in the image of the top or side of the fruit tree canopy, and L is the actual spraying distance of the ground automatic spraying equipment 1 or the aerial spraying drone 2.

[0101] S3: Calculate the decision coefficient K for ground application based on the area and density of the sides and top of the fruit tree canopy g and aerial application decision coefficient K s , and calculate the ground and aerial application rates respectively according to the application decision coefficient;

[0102] S3.1: Calculate the decision coefficient K for ground application g and aerial application decision coefficient K s for:

[0103] K g =0.8×(m Kg *S g +n Kg *δ g )+0.2×(m Ks *S s +n Ks *δ s )

[0104] K s =0.2×(m Kg *S g +n Kg *δ g )+0.8×(m Ks *S s +n Ks *δ s )

[0105] Where S g is the lateral area of ​​the fruit tree canopy, S s The top area of ​​the fruit tree canopy, δ g is the lateral density of the fruit tree canopy, δ s Density of fruit tree canopy top, m Kg 、m Ks 、n Kg 、n Ks are weight coefficients, m Kg It is the ratio of the actual value of the side area of ​​the canopy of the fruit tree to be measured to the average value of the side area of ​​the canopy of several fruit trees in the orchard, m Ks is the ratio of the actual value of the top area of ​​the canopy of the fruit tree to be measured to the average value of the total area of ​​the top areas of the canopies of several fruit trees in the orchard, n Kg is the ratio of the actual density of the canopy side of the fruit tree to be tested to the average density of the canopy side of several fruit trees in the orchard, n Ks It is the ratio of the actual density of the top of the canopy of the fruit tree to be tested to the average density of the top of the canopy of several fruit trees in the orchard.

[0106] S3.2: According to the decision coefficient K of ground application g , aerial spraying decision coefficient K s , calculate the ground and aerial application rates as:

[0107] V=K·Q1·t

[0108] Where V is the amount of pesticide applied to the side or top of the fruit tree canopy, K is the decision coefficient for aerial or ground application, Q1 is the actual flow rate of the ground or aerial spray nozzle, and t is the duration of spraying.

[0109] S4: The fuzzy adaptive control algorithm based on ESO is used to accurately control the pipeline pressure and flow of ground and aerial spraying equipment respectively, such as Figure 9 As shown, the specific steps are:

[0110] S4.1: Determine the input variables and output variables of the control system and establish an inference system of the input variables, output variables, and fuzzy rules. The input variables of the control system are the deviation e and the rate of change of the deviation ec of the output variable. The output variables of the control system are the pipeline pressure and flow rate of the ground or aerial spraying equipment.

[0111] S4.2: Use the extended state observer (ESO) to estimate the system error signal and use the error signal as the input of the fuzzy inference system;

[0112] S4.3: Perform fuzzy inference on the fuzzy set of input variables and the error signal to obtain the output signal of the controller;

[0113] S4.4: Use the output signal of the controller to control the pipeline pressure and flow of the pesticide application equipment to obtain the actual output pressure and flow;

[0114] S4.5: Use the difference between the actual output and the desired output of the system as the input of the ESO, use the ESO to estimate the error signal and update the state of the error estimator;

[0115] S4.6: Use the feedback mechanism to continuously adjust the controller parameters to gradually optimize the spraying rate of the precision spraying control system.

[0116] Example 3:

[0117] like Figure 10 As shown, the present application also provides a fruit tree canopy intelligent detection and precise pesticide application control method and a precise pesticide application effect evaluation method of the system, the specific steps are:

[0118] S1: Before precise application of pesticides, place several water-sensitive papers for testing on the leaves of fruit trees;

[0119] S2: After precise application of pesticides, image data of several water-sensitive papers used in the experiment are collected;

[0120] S3: The collected water-sensitive paper image data is imported into the computer Image J software for processing to obtain the number of droplet deposition points and droplet radius;

[0121] S4: Calculate the liquid coverage, droplet deposition point density and liquid deposition rate respectively;

[0122] S4.1 Calculate the coverage of the liquid medicine:

[0123]

[0124] Where C is the coverage rate of the liquid, A S is the pixel value of the total deposition area of ​​droplets on the water-sensitive paper, A P is the pixel value of the water-sensitive paper area.

[0125] S4.2 Calculate the droplet deposition point density as:

[0126]

[0127] Where K is the droplet deposition point density, drops / cm 2 ; N is the number of droplet deposition points; M is the total area of ​​water-sensitive paper, cm 2 .

[0128] S4.3 Calculate the deposition rate of the liquid medicine:

[0129]

[0130] Where, σ is the deposition rate of the liquid, V S is the actual amount of pesticide applied to the fruit trees, including the sum of the spraying amounts of the ground automatic pesticide application equipment 1 and the aerial pesticide application drone 2, and L is the actual amount of pesticide liquid deposited on the fruit trees.

[0131] In this embodiment, the actual amount of pesticide applied to the fruit trees is V S The actual amount of liquid deposited on the fruit tree, L, is calculated by multiplying the flow rate value of each time period detected by the flow sensor and the total application time:

[0132]

[0133] Where, N is the total number of droplet deposition points on all water-sensitive test papers tested; r is the median radius of the droplets on the water-sensitive test paper; W is the number of water-sensitive test papers tested on fruit trees; S g The side area of ​​the fruit tree canopy calculated for the ground automatic pesticide application device 1; S s The top area of ​​the fruit tree canopy calculated by the aerial pesticide application drone 2; S b It is the area of ​​a single piece of water-sensitive test paper.

[0134] S5: Set the thresholds for the liquid medicine coverage, droplet deposition point density, and liquid medicine deposition rate to evaluate the effect of precise pesticide application.

[0135] In this embodiment, the evaluation index threshold of the liquid coverage rate is set to be greater than 30%, and the evaluation index threshold of the droplet deposition point density is set to be greater than 25 drops / cm 2 The threshold value of the evaluation index of liquid deposition rate is greater than 85%. When the above three indicators are met at the same time, the system is evaluated as an effective and precise pesticide application system.

[0136] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A method for intelligent detection and precise pesticide application control of fruit tree canopies, characterized in that: The specific steps are: S1: Collect image data of the side and top of the fruit tree canopy; S2: Calculate the area and density of the sides and top of the fruit tree canopy; S3: Calculate the decision coefficient for ground application based on the area and density of the sides and top of the fruit tree canopy and decision coefficient for aerial application , and calculate the ground and aerial application rates respectively according to the application decision coefficient; S4: Use ESO-based fuzzy adaptive control algorithm to accurately control the pipeline pressure and flow of ground and aerial spraying equipment respectively; The specific method for calculating the side area and top area of ​​the fruit tree canopy is: The collected images of the side and top of the fruit tree canopy are gray-scaled and segmented to obtain the outline of the side and top of the fruit tree canopy, which only contains the leaf area. Identify and count the pixel points in the leaf area of ​​the collected images of the side and top of the fruit tree canopy, find the pixel points of all leaves, and calculate the area of ​​the side and top of the fruit tree canopy: ； Where, is the side or top area of ​​the fruit tree canopy, is the actual area represented by a unit pixel in the side or top image of the fruit tree canopy. is the number of canopy leaf pixels in the side or top image of the fruit tree canopy; The specific steps for calculating the density of the side and top of the fruit tree canopy are as follows: Morphological processing operations are used to fill the holes in the leaf areas on the sides and tops of the fruit tree canopies displayed after canopy segmentation, to obtain the contour area of ​​the fruit tree canopy after the holes are filled, and to calculate the density of the fruit tree canopy sides and tops: ； Where, It is the density of the side or top of the fruit tree canopy. is the contour area of ​​the fruit tree canopy after the cavity on the side or top is filled, and S is the area of ​​the fruit tree canopy on the side or top; Constructing the decision coefficient for ground application , decision coefficient for aerial spraying for: ； Where, is the lateral area of ​​the fruit tree canopy, The top area of ​​the fruit tree canopy, is the lateral density of the fruit tree canopy, The density of the top of the fruit tree canopy, 、 、 、 are weight coefficients, It is the ratio of the actual value of the canopy lateral area of ​​the fruit tree to be measured to the average value of the overall canopy lateral area of ​​several fruit trees in the orchard. It is the ratio of the actual value of the top area of ​​the canopy of the fruit tree to be measured to the average value of the total area of ​​the top areas of the canopies of several fruit trees in the orchard. It is the ratio of the actual density of the canopy side of the fruit tree to be tested and the average density of the canopy side of several fruit trees in the orchard. It is the ratio of the actual density of the top of the canopy of the fruit tree to be tested to the average density of the top of the canopy of several fruit trees in the orchard.

2. The method for intelligent detection and precise pesticide application control of fruit tree canopies according to claim 1, characterized in that: According to the decision coefficient of ground application , decision coefficient for aerial spraying , calculate the ground and aerial application rates as: ； Where, The amount of pesticide applied to the side or top of the fruit tree canopy. is the decision coefficient for aerial or ground application, The actual flow rate of the ground spray nozzle or aerial spray nozzle. The duration of application.

3. The method for intelligent detection and precise pesticide application control of fruit tree canopy according to claim 1, characterized in that: The specific steps for accurately controlling the pipeline pressure and flow of ground and aerial spraying equipment using the ESO-based fuzzy adaptive control algorithm are as follows: S4.1: Determine the input variables and output variables of the control system and establish an inference system of input variables, output variables and fuzzy rules. The input variable of the control system is the deviation of the output variable. and deviation change rate , the output variables of the control system are the pipeline pressure and flow rate of the ground or aerial spraying equipment; S4.2: Use the extended state observer (ESO) to estimate the system error signal and use the error signal as the input of the fuzzy inference system; S4.3: Perform fuzzy inference on the fuzzy set of input variables and the error signal to obtain the output signal of the controller; S4.4: Use the output signal of the controller to control the pipeline pressure and flow of the pesticide application equipment to obtain the actual output pressure and flow; S4.5: Use the difference between the actual output and the desired output of the system as the input of the ESO, use the ESO to estimate the error signal and update the state of the error estimator; S4.6: Use the feedback mechanism to continuously adjust the controller parameters to gradually optimize the spraying rate of the precision spraying control system.

4. A fruit tree canopy intelligent detection and precise pesticide application control system, characterized in that: A method for intelligent detection and precise pesticide application control of a fruit tree canopy for realizing any one of claims 1 to 3, comprising a ground automatic pesticide application device (1) for collecting image data of the side of a fruit tree canopy and performing precise pesticide application, an aerial pesticide application drone (2) for collecting image data of the top of a fruit tree canopy and performing precise pesticide application, and a remote monitoring control module for data intercommunication with the ground automatic pesticide application device (1) and the aerial pesticide application drone (2); The ground automatic pesticide application equipment (1) and the aerial pesticide application drone (2) are in data communication with each other. The ground automatic pesticide application equipment (1) includes a mobile body (101), and the aerial pesticide application drone (2) includes a flying drone (201). Both the mobile body (101) and the flying drone (201) are provided with a machine vision component (3) for collecting image data, and a pesticide application component (4) for precise pesticide application.

5. The fruit tree canopy intelligent detection and precise pesticide application control system according to claim 4, characterized in that: The machine vision component (3) comprises a metal box (301), a camera (302) installed in the metal box (301) for collecting image data, and an image processing module installed in the metal box (301) for processing and identifying the image data; The camera (302) and the image processing module are both in data communication with the first processor module.

6. The fruit tree canopy intelligent detection and precise pesticide application control system according to claim 4, characterized in that: The drug application assembly (4) includes a drug storage box (401) and a spray pipe (402) for spraying drug mist, wherein the spray pipe (402) is provided with a plurality of drug liquid spray heads (403); The medicine storage box (401) is connected to the liquid inlet end of the spraying pipe (402) via a pumping mechanism, and the pumping mechanism controls the precise application of medicine via a control mechanism. The control mechanism includes a high-pressure diaphragm pump (404), and the liquid inlet end of the high-pressure diaphragm pump (404) is connected to the medicine storage box (401) via a pipeline. The liquid outlet of the high-pressure diaphragm pump (404) is connected to a first proportional solenoid valve (405) and a second proportional solenoid valve (406) through pipelines, respectively. The first proportional solenoid valve (405) is in communication with the medicine storage box (401). A pressure transmitter (407) is also provided on the pipeline connecting the high-pressure diaphragm pump (404) and the second proportional solenoid valve (406). The liquid outlet of the second proportional solenoid valve (406) is in communication with the liquid inlet of the spray pipe (402) through a pipeline. A turbine flowmeter (408) is also provided on the pipeline connecting the second proportional solenoid valve (406) and the spray pipe (402). The liquid inlet of the spray pipe (402) is provided with an electric switch valve (409) for controlling its opening and closing.

7. The fruit tree canopy intelligent detection and precise pesticide application control system according to claim 6, characterized in that: The control mechanism includes a second processor module, a current-to-voltage module for signal conversion, a PWM drive module for controlling the first proportional solenoid valve (405) and the second proportional solenoid valve (406), and a relay for controlling the electric switch valve (409); The current-to-voltage module, PWM drive module and relay all communicate data with the second processor module.

Images