Agricultural cultivation and integrated seeding and pesticide spraying device and intelligent pesticide spraying method

Abstract

The application provides a device for ploughing, sowing and pesticide spraying and an intelligent pesticide spraying method. The device is provided with a power module, a rotary ploughing module, a pesticide spraying module and a sowing module. The rotary ploughing module, the pesticide spraying module and the sowing module are installed at the rear end of the power module. The power module, the rotary ploughing module, the pesticide spraying module and the sowing module are independently arranged and can be combined for use according to actual ploughing requirements. The device can realize unmanned autonomous ploughing or manual operation control ploughing through a visual detection module arranged at the front end of the power module. The intelligent pesticide spraying method of the device is that the power module controls the pesticide spraying module to realize intelligent pesticide spraying. The device is modularly designed and combined for use. The soil loosening, sowing, pesticide spraying and fertilization are integrated, and the ploughing is prevented from being rolled multiple times by mechanical equipment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of agricultural machinery equipment, in particular to a device for plowing, sowing and pesticide spraying and an intelligent pesticide spraying method. BACKGROUND

[0002] In agricultural plowing, modern agriculture has deviated from the primitive state of cutting and burning, and instead uses modern mechanical equipment for modern plowing. Using mechanical equipment for plowing can greatly improve plowing efficiency, save time and labor, effectively save labor, and enable one person to operate a machine to plow a large piece of land. Existing plowing machinery is diverse according to the needs of plowing, including rotary tillers, combine harvesters, seeders, pesticide sprayers, and picking machines, among others. The diversification of forms can meet the plowing needs of most land. However, most of the current production machinery is independently designed according to a process or requirement in the plowing process, which results in the single function of many existing mechanical equipment, especially in the aspect of soil loosening and sowing. Traditional mechanical equipment plowing requires the use of a rotary tiller to loosen the soil, a seeder to sow, and a fertilizer sprayer to apply fertilizer for the next step of plowing. When plowing the same piece of land, different mechanical equipment needs to be purchased, and multiple plowing processes need to be implemented, which reduces efficiency and compacts the loosened soil, which is not conducive to the growth of crops. The purchase of different equipment increases the cost of plowing for farmers. Therefore, the applicant integrates some processes in the plowing process according to the actual needs of plowing, designs and invents a device for plowing, sowing and pesticide spraying and an intelligent pesticide spraying method, which can achieve intelligent autonomous plowing or manual operation control plowing, effectively improve plowing efficiency, and use modularly designed equipment in combination, which can achieve soil loosening, sowing, pesticide spraying and fertilization in an integrated manner, effectively avoiding the multiple compaction of the plowed land by mechanical equipment and effectively saving energy. SUMMARY

[0003] To solve the above technical problems, the present application provides a device for plowing, sowing and pesticide spraying and an intelligent pesticide spraying method. The device includes a modular rotary tiller module, a power module, a pesticide spraying module and a sowing module. The rotary tiller module, the pesticide spraying module and the sowing module are installed at the rear end of the power module. The power module, the rotary tiller module, the pesticide spraying module and the sowing module are independently arranged and can be used in combination according to the actual needs of plowing. The device for plowing, sowing and pesticide spraying can achieve unmanned autonomous plowing or manual operation control plowing through the visual detection module arranged at the front end of the power module. The intelligent pesticide spraying method of the device for plowing, sowing and pesticide spraying controls the pesticide spraying module through the power module to achieve intelligent pesticide spraying. The modularly designed equipment is used in combination, soil loosening, sowing, pesticide spraying and fertilization are integrated, multiple compaction of the plowed land by mechanical equipment is avoided, and energy is effectively saved.

[0004] To achieve the above object, the technical scheme adopted by the present application is:

[0005] The application discloses a device integrating ploughing, seeding and pesticide spraying, which comprises a rotary ploughing module, a power module, a pesticide spraying module and a seeding module.

[0006] Further improvement of the structure of the application, the device integrating ploughing, seeding and pesticide spraying is provided with a rotary ploughing module, the rotary ploughing module is provided with a mounting shaft, a fixing frame, a rotary ploughing motor, a transmission shaft, a transmission chain, a rotating shaft and rotary ploughing blades, the main structure of the rotary ploughing module is the fixing frame, side plates are arranged on the two sides of the fixing frame, the fixing frame is provided with the mounting shaft, the rotary ploughing motor is arranged in the fixing frame, the transmission shaft is arranged on the rotary ploughing motor, the transmission chain on the side plates of the fixing frame is connected with the transmission shaft, the rotating shaft is arranged in the side plates of the fixing frame, a plurality of groups of rotary ploughing blades are arranged on the rotating shaft, and the transmission chain is connected with the rotating shaft.

[0007] Further improvement of the structure of the application, the device integrating ploughing, seeding and pesticide spraying is provided with a power module, the power module is provided with a solar panel, a cockpit and a visual detection module, the main body of the power module adopts a four-wheel tractor as a power source, the cockpit is arranged in the middle of the power module, the solar panel is arranged on the top of the cockpit, the visual detection module is arranged at the front end of the power module, and the rotary ploughing module and the pesticide spraying module are arranged at the tail of the power module.

[0008] Further improvement of the structure of the application, the device integrating ploughing, seeding and pesticide spraying is provided with a pesticide spraying module, the pesticide spraying module is provided with a pesticide liquid tank, a liquid injection port, an electric stirrer, a liquid outlet pipe and a pesticide liquid spraying head, the main structure of the pesticide spraying module is the pesticide liquid tank, two liquid injection ports are arranged on the pesticide liquid tank, the liquid injection ports are a water injection port and a pesticide injection port, the electric stirrer is arranged in the pesticide liquid tank, the liquid outlet pipe is arranged at the bottom of the pesticide liquid tank, the pesticide liquid spraying head is arranged at the front end of the liquid outlet pipe, and the electric pump is arranged on the liquid outlet pipe.

[0009] Further improvement of the structure of the application, the intelligent cultivation, sowing and pesticide spraying integrated device is provided with a sowing module, the sowing module is provided with a connecting shaft, an adjusting rod, a buffer rod, a storage bin, a front guide wheel, a rear guide wheel, a falling pipe, a falling head, a turning blade, a mixing roller motor, a mixing roller, a wheel and a rack, the main body structure of the sowing module is the rack, the adjusting rod is arranged at the front end of the rack, the buffer rod is arranged at the upper part of the adjusting rod, the connecting shaft is arranged at the front end of the adjusting rod, the storage bin is arranged at the upper part of the rack, two groups of mixing roller motors and mixing rollers are arranged in the storage bin, the falling pipe is connected to the lower part of the storage bin, the falling head is arranged on the falling pipe, the turning blade is arranged behind the falling head, the rear guide wheel is arranged at the lower part of the falling pipe, the front guide wheel is arranged at the front side of the rear guide wheel, and the wheel is arranged at the middle part of the lower side of the rack.

[0010] The application provides an intelligent pesticide spraying method for the cultivation, sowing and pesticide spraying integrated device, and the specific steps are as follows.

[0011] 1) Video acquisition device positioning;

[0012] Before pesticide spraying, two cameras need to be positioned to ensure the consistency and correctness of the image angle.

[0013] 2) Double-view angle algorithm classification pre-operation;

[0014] The YOLO algorithm is used to realize the classification task, wherein a smooth L1 loss coordinate function is used to measure the accuracy of the position and size of the predicted boundary box.

[0015] 3) Double model classification task verification;

[0016] The results of the last SoftMax activation of the two sets of classification models are merged according to the merging formula as the overall loss function, and the model relationship is updated through back propagation.

[0017] 4) Plant pesticide proportioning;

[0018] The classified objects predicted by the algorithm are compared with the pesticide formula in the database for proportioning, the required pesticide type is injected into the liquid injection port, and an electric mixer is used for instant stirring, and finally the stirred pesticide is put into the pesticide tank.

[0019] 5) Double-view angle algorithm volume representation;

[0020] The classification photos in the front and above are merged, data is extracted, and relative spatial volume calculation is performed according to the relative spatial volume expression.

[0021] 6) Plant pesticide dosage calculation;

[0022] According to different relative space volumes, calculate the dosage of pesticides required for different plants and spray them;

[0023] 7) Adjust the position of the liquid spray nozzle;

[0024] The optimal spraying height for each plant species is calculated based on different relative spatial volumes, wherein the height values ​​set for the liquid medicine nozzles 3-5 are calculated according to the height conversion formula.

[0025] A further improvement of the method of the present invention is that the smoothed L1 loss coordinate function in step 2) is expressed as:

[0026]

[0027] Where x represents the difference between the predicted value and the true value. The smooth loss is calculated for each of the four coordinates of the predicted bounding box and weighted summed to obtain the final smooth L1 loss value for reverse propagation.

[0028] The method of the present invention is further improved, and the combined formula in step 3) is expressed as:

[0029] θ i =X i +Y i , i=1,2,3,...

[0030] Where i represents a specific label. For example, if the classification task requires classification A and B, i = 1 represents the model's confidence in A, and i = 2 represents the model's confidence in B. i Represents the confidence of the first set of model photos in front of the i object, Y i Represents the confidence of the second set of models for the i-th object in the photo directly above. i With Y i Add them together to get the comprehensive confidence θ of the two models i , and then give the predicted label according to the comprehensive confidence, perform back propagation, and iteratively update the weight parameters.

[0031] The method of the present invention is further improved, and the relative space volume expression formula in step 5) is expressed as:

[0032] V=whh′, where w and h are the width and height of the prediction box captured by the first set of models, and h′ is the height captured by the second set of models. According to this algorithm, the relative spatial volume of the target object can be extracted.

[0033] The method of the present invention is further improved, and the relative space volume expression formula in step 7) is expressed as:

[0034]

[0035] wherein, delta is the height of the nozzle setting, H is the maximum relative spatial volume V of the plant max V is the current relative spatial volume of the plant.

[0036] The present application provides a device for ploughing, sowing and pesticide spraying and an intelligent pesticide spraying method. The device is modularized and can be combined for use according to actual ploughing needs. The device can realize loosening of soil, sowing, pesticide spraying and fertilization in an integrated manner, effectively reduces the cost of the device and the cost of ploughing, and can realize autonomous ploughing or manual operation control ploughing.

[0037] 1. The device for ploughing, sowing and pesticide spraying is modularized and can be combined for use according to actual ploughing needs, can realize loosening of soil, sowing, pesticide spraying and fertilization in an integrated manner, effectively reduces the cost of the device and the cost of ploughing.

[0038] 2. The device for ploughing, sowing and pesticide spraying can realize autonomous ploughing or manual operation control ploughing. Autonomous ploughing is controlled by Beidou positioning and mechanical vision technology, effectively improves the ploughing efficiency, and diversifies the ploughing operation.

[0039] 3. The device for ploughing, sowing and pesticide spraying optimizes and integrates the ploughing process, can realize loosening of soil, sowing, pesticide spraying and fertilization in an integrated manner, effectively avoids multiple compaction of the farmland by mechanical equipment, and effectively saves energy.

[0040] 4. The device for ploughing, sowing and pesticide spraying can be adjusted for use according to different ploughing objects, is suitable for ploughing of various crops, and has a wide range of applications.

[0041] 5. The device for ploughing, sowing and pesticide spraying and the intelligent pesticide spraying method can identify the type of the plant, then match the type of the pesticide, can quantitatively analyze the relative spatial volume of the plant, determine the dose of the pesticide, and then realize overall high-precision pesticide spraying.

[0042] 6. The device and method for intelligent pesticide spraying are characterized in that the single-view YOLO algorithm is used for classifying plants, the accuracy and timeliness of classification are improved, and different pesticides can be automatically matched for different plants.

[0043] 7. The device and method for intelligent pesticide spraying are characterized in that the relative spatial volume of plants is predicted, and the pesticide dosage is determined, so that different plants can be sprayed with different dosages of pesticides. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 It is a schematic diagram of the overall structure of the application.

[0045] Figure 2 It is a schematic diagram of the rotary tillage module structure of the application.

[0046] Figure 3 It is a schematic diagram of the power module and pesticide spraying module structure of the application.

[0047] Figure 4 It is a schematic diagram of the seeding module structure of the application.

[0048] Figure 5 It is a schematic diagram of the local structure of the seeding module of the application. Figure 1

[0049] Figure 6 It is a schematic diagram of the local structure of the seeding module of the application. Figure 2

[0050] Figure 7 The intelligent pesticide spraying method provided by the embodiment of the application is shown in the flowchart.

[0051] Figure 8 The dual-view camera positioning diagram of the intelligent pesticide spraying method based on the YOLO algorithm provided by the embodiment of the application is shown in the figure, and the labels are as follows: 1, rotary tillage module; 1-1, mounting shaft; 1-2, fixing frame; 1-3, rotary tillage motor; 1-4, transmission shaft; 1-5, transmission chain; 1-6, rotating shaft; 1-7, rotary tillage blade; 2, power module; 2-1, solar panel; 2-2, cockpit; 2-3, visual detection module; 3, pesticide spraying module; 3-1, pesticide liquid tank; 3-2, liquid filling port; 3-3, electric stirrer; 3-4, liquid outlet pipe; 3-5, pesticide liquid nozzle; 4, seeding module; 4-1, connecting shaft; 4-2, adjusting rod; 4-3, buffer rod; 4-4, storage bin; 4-5, front guide wheel; 4-6, rear guide wheel; 4-7, material falling pipe; 4-7-1, material falling head; 4-7-2, soil turning blade; 4-8, material mixing roller motor; 4-8-1, material mixing roller; 4-9, wheel; 4-10, rack. DETAILED DESCRIPTION ​​

[0052] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0053] like Figure 1 As shown: It is a schematic diagram of the structure of the equipment for farming, sowing and pesticide spraying, which is provided with a rotary tillage module 1, a power module 2, a pesticide spraying module 3 and a sowing module 4. Figure 1 As shown, the main body of the equipment for farming, sowing and pesticide spraying is a power module 2, the power module 2 provides power for the equipment, and drags other module equipment to work. The rear end of the power module 2 is installed with a rotary tillage module 1, a pesticide spraying module 3 and a sowing module 4. The rotary tillage module 1 is installed on the lower side of the tail of the power module 2, and the pesticide spraying module 3 is set on the upper side of the tail of the power module 2. Part of the structure of the pesticide spraying module 3 is located at the rear end of the rotary tillage module 1, and the sowing module 4 is dragged and set at the rear end of the power module 2. The rotary tillage module 1, power module 2, pesticide spraying module 3 and sowing module 4 are independently modularized and can be used in combination with the power module 2 according to actual farming needs. The equipment can perform intelligent autonomous farming or manually controlled farming. Beidou positioning and machine vision technology are used for control during autonomous farming. The algorithm adopted when using machine vision technology is the Yolo algorithm, which identifies the spatial range of plants in the xy and xz planes of the land that needs to be cultivated. Data of different heights of the same plants and different heights of different plants can be obtained. The optimal height position for plant spraying or fertilizing can be obtained based on the data, and the motion control of the power module 2 and the pesticide spraying module 3 can be realized, effectively improving the farming efficiency. The modular design of the equipment can be used in combination to realize the integrated loosening of soil, sowing, spraying of pesticides and fertilization, effectively avoiding multiple crushing of cultivated land by mechanical equipment, achieving the purpose of farming in one go, and effectively saving energy.

[0054] like Figure 2As shown, a rotary tillage module 1 is provided for an integrated equipment for tillage, sowing and pesticide spraying. The rotary tillage module 1 structure is provided with a mounting shaft 1-1, a fixing frame 1-2, a rotary tillage motor 1-3, a transmission shaft 1-4, a transmission chain 1-5, a rotating shaft 1-6 and a rotary tillage blade 1-7. The main structure of the rotary tillage module 1 is a fixing frame 1-2. The fixing frame 1-2 is a structural frame that provides mounting support for various equipment. Side panels are provided on both sides of the fixing frame 1-2, and the side panels are used as protective panels and support panels. The fixing frame 1-2 is provided with a mounting shaft 1-1, and the rotary tillage module is connected and fixed to the power module 2 through the mounting shaft 1-1. A rotary tillage motor 1-3 is installed inside the fixing frame 1-2, and the rotary tillage motor 1-3 is The rotary tillage module 1 provides kinetic energy, and the electric energy required by the rotary tillage motor 1-3 is provided by the power module 2. The rotary tillage motor 1-3 is provided with a transmission shaft 1-4, and the transmission shaft 1-4 is connected to the transmission chain 1-5 on the side plate of the fixed frame 1-2. The side plates on both sides of the fixed frame 1-4 are installed with a rotating shaft 1-6, and multiple groups of rotary tillage blades 1-7 are installed on the rotating shaft 1-6. The transmission chain 1-5 is connected to the rotating shaft 1-6. When the rotary tillage motor 1-3 moves, the power is transmitted to the transmission chain 1-5 through the transmission shaft 1-4, and the transmission chain 1-5 then transmits the power to the rotating shaft (1-6). When the rotating shaft (1-6) moves, it drives the rotary tillage blades 1-7 to rotate, and the rotation of the rotary tillage blades 1-7 realizes the loosening of the land.

[0055] like Figure 3 As shown, a power module 2 is provided for an integrated tillage, sowing and pesticide spraying device. The power module 2 is structurally provided with a solar panel 2-1, a cockpit 2-2 and a visual detection module 2-3. The main body of the power module 2 uses a four-wheel tractor as a power source. The power module 2 can perform intelligent autonomous tillage control or manual operation control of tillage. A cockpit 2-2 is provided in the middle of the power module 2. The cockpit 2-2 is a manual control operating table. A solar panel 2-1 is provided on the top of the cockpit 2-2. The solar panel 2-1 can provide solar charging for the power module 2. A visual detection module 2-3 is installed at the front end of the power module 2. The visual detection module 2-3 realizes the intelligent autonomous tillage control of the power module 2. The rear of the power module 2 is provided with a rotary tillage module 1 and a pesticide spraying module 3.

[0056] like Figure 3As shown, the pesticide spraying module 3 is shown as an integrated device setting of tillable and sowing and pesticide spraying, the pesticide spraying module 3 is shown as a pesticide tank 3-1, a liquid injection port 3-2, an electric stirrer 3-3, a liquid outlet pipe 3-4 and a pesticide spraying head 3-5, the main structure of the pesticide spraying module 3 is the pesticide tank 3-1, the pesticide tank 3-1 is a pesticide storage and mixing tank, two liquid injection ports 3-2 are arranged on the pesticide tank 3-1, which are water injection port and pesticide injection port respectively, water and pesticide are injected into the pesticide tank 3-1 through the liquid injection port 3-2, a cover is arranged on the liquid injection port 3-2, an electric stirrer 3-3 is arranged in the pesticide tank 3-1, the electric stirrer 3-3 stirs and mixes the water and pesticide newly added into the pesticide tank 3-1, a liquid outlet pipe 3-4 is arranged at the bottom of the pesticide tank 3-1, a pesticide spraying head 3-5 is arranged at the front end of the liquid outlet pipe 3-4, an electric pump is arranged on the liquid outlet pipe 3-4, when the pesticide spraying module 3 works, the electric pump arranged on the liquid outlet pipe 3-4 works, the pesticide is pressurized and sprayed from the pesticide spraying head 3-5, and the spraying of the pesticide is realized.

[0057] As Figures 4-6The illustrated is a sowing module 4 of the integrated device for plowing, sowing and pesticide spraying, which is provided with a connecting shaft 4-1, an adjusting rod 4-2, a buffer rod 4-3, a storage bin 4-4, a front guide wheel 4-5, a rear guide wheel 4-6, a dropping pipe 4-7, a dropping head 4-7-1, a turning blade 4-7-2, a mixing roller motor 4-8, a mixing roller 4-8-1, a wheel 4-9 and a rack 4-10. The main structure of the illustrated sowing module 4 is the rack 4-10, which is designed according to the installation requirements of various structures of the sowing module 4. The front end of the illustrated rack 4-10 is provided with the adjusting rod 4-2, which can adjust the distance and height of the sowing module 4 and the power module 2 to meet the plowing requirements of different lands and crops. The upper part of the illustrated adjusting rod 4-2 is provided with the buffer rod 4-3, which can effectively buffer the stress of the power module 2 in the process of driving the plowing module 4 to move forward, thereby protecting the plowing module 4 from uneven stress damage. The front end of the illustrated adjusting rod 4-2 is provided with the connecting shaft 4-1, which realizes the connection and use of the plowing module 4 and the power module 2. The upper part of the illustrated rack 2-10 is provided with the storage bin 4-4, which is used for storing seeds or fertilizers. Two groups of mixing roller motors 4-8 and mixing rollers 4-8-1 are arranged in the illustrated storage bin 4-4. The rotation of the mixing rollers 4-8-1 can also realize the mixing of the seeds and fertilizers in the storage bin. The mixing roller motor 4-8 drives the mixing roller 4-8-1 to rotate to realize the distribution of the materials in the storage bin 4-4. The lower part of the illustrated storage bin 4-4 is connected with the dropping pipe 4-7. The materials seeds distributed by the mixing roller 4-8-1 enter the dropping pipe 4-7. The dropping pipe 4-7 is provided with the dropping head 4-7-1. The dropping head 4-7-1 sows the fertilizers or seeds in the dropping pipe 4-7 into the land. The rear of the illustrated dropping head 4-7-1 is provided with the turning blade 4-7-2. The dropping head 4-7-1 covers the sowed seeds or fertilizers with soil. The lower part of the illustrated dropping pipe 4-7 is provided with the rear guide wheel 4-6. The front side of the illustrated rear guide wheel 4-6 is provided with the front guide wheel 4-5. The rotation of the front guide wheel 4-5 and the rear guide wheel 4-6 realizes the interval feeding of the fertilizers or seeds in the dropping pipe 4-7 by the dropping head 4-7-1. The lower side of the illustrated rack 4-10 is provided with the wheel 4-9, which is a supporting element for the forward movement of the plowing module 4.

[0058] As shown in the figure, Figure 7 The figure shows the intelligent pesticide spraying method based on the YOLO algorithm provided by the present application.

[0059] Step S1: Positioning of the video acquisition device.

[0060] As shown in the figure, Figure 8 The figure shows the double-view camera positioning diagram of the intelligent pesticide spraying method based on the YOLO algorithm provided by the present application.

[0061] Before pesticide spraying, the dual camera needs to be positioned first, one is placed in front of the plant, and the other is placed above the plant.

[0062] The former image collected is used as the data set for the front photo classification of the plant, and also as the source of the plant width and height information. The latter is used to extract the depth information of the plant and the data set for the overhead photo classification.

[0063] The pictures collected by the two video acquisition devices are first uniformly converted, including converting all images to the same format, and then converting the images to a fixed size. For the front photo and overhead photo of the plant used for classification tasks, in order to enhance the diversity and robustness of the data, data enhancement processing is performed on the plant photos, including random cropping, rotation, scaling, and flipping. Finally, the photos are given corresponding label information, including the coordinates of the bounding box and the class.

[0064] Step S2: Pre-operation of dual-view algorithm classification.

[0065] This time, YOLO algorithm is used to realize classification. In step S1, the data set collection and certain preprocessing operation are realized.

[0066] This time, YOLO algorithm is divided into two sets of models. One set classifies the front photo, and the other set classifies the overhead photo.

[0067] In the first set of models, the processed front photo is input into the network model of YOLO algorithm for training. The training includes forward propagation, boundary box loss transmission, and class loss transmission. The class loss function is described in detail in step S3.

[0068] Among them, the smooth L1 loss coordinate loss function is used to measure the accuracy of the position and size of the predicted bounding box:

[0069]

[0070] Where x represents the difference between the predicted value and the true value. The smooth loss of the four coordinates of the predicted bounding box is calculated respectively, and the weighted sum is obtained, and the final smooth L1 loss value is obtained. Back propagation is performed.

[0071] The second set of models classifies the overhead photo, and the training process is different from the first set of models.

[0072] Step S3: Verification of dual-model classification task

[0073] The present application adopts a dual-view camera mode, which can realize a dual-model simultaneous verification method, mainly using a dual-channel set loss function. In this classification task, the first and second models use the same network structure, and the last layer is a SoftMax network structure. The results of the two models after SoftMax activation are combined as the overall loss function, and the model relationship is updated by back propagation. The combination formula is:

[0074] θ i =X i +Y i , i = 1, 2, 3...⑵

[0075] Where i represents a specific label, for example, the classification task needs to classify A and B, i = 1 represents the model's confidence in A, and i = 2 represents the model's confidence in B. X i represents the confidence of the first model in the front photo of i objects, and Y i represents the confidence of the second model in the photo above i objects. X i and Y i are added to obtain the comprehensive confidence θ i of the two models, and then the predicted label is given according to the comprehensive confidence, and the weight parameters are updated by back propagation and iteration.

[0076] Finally, when the loss function of the training set and the test set is reduced to a certain extent, the classification model training is completed.

[0077] Step S4: Plant pesticide proportioning.

[0078] Steps S2 and S3 complete the YOLO algorithm classification model, which can predict the data collected in real time and view the type of plant that needs to be sprayed with pesticide at that time.

[0079] Then retrieve the type of pesticide needed for the type of plant in the database, inject the type of pesticide needed into the liquid injection port, and use an electric mixer to mix it immediately. Finally, put the mixed pesticide into the pesticide tank to achieve immediate proportioning of specific pesticides for different types of plants. The proportioned pesticide is proportioned according to the maximum amount per plant.

[0080] Step S5: Dual-view algorithm volume representation.

[0081] Since there is no depth information in the YOLO algorithm, it is not possible to calculate the relative spatial volume of the plant, so a video capture device needs to be added above to capture the depth information of the plant.

[0082] The relative spatial volume expression formula is:

[0083] V = whh' (3)

[0084] where w, h are the width and height of the bounding box of the first model, and h' is the height of the second model. According to this algorithm, the relative spatial volume of the target object can be extracted.

[0085] Step S6: Calculate the dose of plant pesticide.

[0086] For the plant, too much pesticide spraying will cause plant damage, too much pesticide spraying, soil and water pollution, and less pesticide will cause poor control effect, drug resistance risk and disease spread. Therefore, it is necessary to give individual according to the volume of each plant. In step S5, the relative spatial volume of the plant is calculated, so the required pesticide dose needs to be specifically measured according to the relative spatial volume of each plant. In step S4, the pesticide ratio work and qualitative work of the plant are completed, and the maximum dose of the pesticide after the ratio is directly input into the 3-1 liquid tank, according to the relative spatial volume of the plant calculated in step S5, the specific dose of pesticide required in the database is compared, and the excess pesticide is input into the storage room of the 3-1 liquid tank for next time spraying.

[0087] Step S7: Adjust the position of the pesticide liquid nozzle.

[0088] When spraying pesticide on plants, the height of the nozzle relative to the plant is an important consideration factor involving coverage, spray uniformity, avoiding dripping and loss. Therefore, different nozzle spray areas need to be set according to the specific height of each plant.

[0089] In step S-5, the relative spatial volume of the plant is obtained, and then the height value of the pesticide liquid nozzle can be set by the following formula:

[0090]

[0091] where δ is the height of the nozzle set, H is the maximum relative spatial volume V of the plant, and V is the relative spatial volume of the current plant. max

[0092] Step S8: Intelligent and fine spraying.

[0093] When the above steps are completed, the intelligent and fine pesticide spraying can be repeated and iterated.

[0094] ​The above merely describes the preferred embodiments of the present application, but does not constitute any other form of limitation to the present application, and any modification or equivalent change made according to the technical essence of the present application still falls within the scope of the present application.

Claims

1. An intelligent pesticide spraying method of a device integrating plowing, seeding and pesticide spraying, a pesticide spraying module (3) of the device integrating plowing, seeding and pesticide spraying is provided with a pesticide liquid tank (3-1), a liquid injection port (3-2), an electric stirrer (3-3), a liquid outlet pipe (3-4) and a pesticide liquid spraying head (3-5), two liquid injection ports (3-2) are arranged on the pesticide liquid tank (3-1), the liquid injection ports (3-2) are water injection ports and pesticide injection ports, the electric stirrer (3-3) is arranged in the pesticide liquid tank (3-1), the liquid outlet pipe (3-4) is arranged at the bottom of the pesticide liquid tank (3-1), the pesticide liquid spraying head (3-5) is arranged at the front end of the liquid outlet pipe (3-4), and an electric pump is arranged on the liquid outlet pipe (3-4), and the specific steps are as follows, characterized in that: 1) video acquisition device positioning; Before pesticide spraying, two cameras need to be positioned to ensure the consistency and correctness of the image angle; 2) double-view angle algorithm classification pre-operation; YOLO algorithm is used to realize the classification task, wherein a smooth L1 loss coordinate function is used to measure the accuracy of the position and size of the predicted boundary box; 3) double-model classification task verification; The results of the last SoftMax activation of the two sets of classification models are combined according to a combination formula as a total loss function, and the model relationship is updated through back propagation; 4) plant pesticide proportioning; The classified objects predicted by the algorithm are compared with the pesticide formulations in the database for proportioning, the required pesticide types are injected into the liquid injection ports (3-2), and the electric stirrer (3-3) is used for instant stirring, and finally the stirred pesticide is put into the pesticide liquid tank (3-1); 5) double-view angle algorithm volume representation; The classification photos in the front and above are combined, data is extracted, and the relative spatial volume is calculated according to the relative spatial volume expression; 6) plant pesticide dosage calculation; According to different relative spatial volumes, the required pesticide dosage for different plants is calculated for spraying; 7) pesticide liquid spraying head position adjustment; According to different relative spatial volumes, the best spraying height of each plant is calculated, and the height value of the pesticide liquid spraying head 3-5 is calculated according to a height conversion formula.

2. The intelligent pesticide spraying method of the plowing, sowing and pesticide spraying integrated apparatus according to claim 1, characterized in that: The device integrating plowing, seeding and pesticide spraying comprises a rotary plowing module (1), a power module (2), a pesticide spraying module (3) and a seeding module (4), the rotary plowing module (1), the pesticide spraying module (3) and the seeding module (4) are installed at the rear end of the power module (2), the rotary plowing module (1) is installed at the lower end of the tail of the power module (2), the pesticide spraying module (3) is arranged on the upper side of the tail of the power module (2), and the seeding module (4) is arranged at the rear end of the power module (2) in a trailing manner, the rotary plowing module (1), the power module (2), the pesticide spraying module (3) and the seeding module (4) are independently modularized, and are used in combination with the power module (2) according to actual plowing requirements. The rotary tillage module (1) is provided with a mounting shaft (1-1), a fixed frame (1-2), a rotary tillage motor (1-3), a transmission shaft (1-4), a transmission chain (1-5), a rotating shaft (1-6) and rotary tillage blades (1-7), the main structure of the rotary tillage module (1) is the fixed frame (1-2), the fixed frame (1-2) is provided with side plates on both sides, the fixed frame (1-2) is provided with a mounting shaft (1-1), the fixed frame (1-2) is internally provided with a rotary tillage motor (1-3), the rotary tillage motor (1-3) is provided with a transmission shaft (1-4), the transmission shaft (1-4) is connected with the transmission chain (1-5) on the side plate of the fixed frame (1-2), the rotating shaft (1-6) is installed in the side plates of the fixed frame (1-2), a plurality of rotary tillage blades (1-7) are installed on the rotating shaft (1-6), the transmission chain (1-5) is connected with the rotating shaft (1-6); The sowing module (4) is provided with a connecting shaft (4-1), an adjusting rod (4-2), a buffer rod (4-3), a storage bin (4-4), a front guide wheel (4-5), a rear guide wheel (4-6), a material falling pipe (4-7), a material falling head (4-7-1), a soil turning blade (4-7-2), a mixing roller motor (4-8), a mixing roller (4-8-1), a wheel (4-9) and a rack (4-10), the main structure of the sowing module (4) is the rack (4-10), the rack (4-10) is provided with an adjusting rod (4-2) at the front end, the adjusting rod (4-2) is provided with a buffer rod (4-3) at the upper part, the adjusting rod (4-2) is provided with a connecting shaft (4-1) at the front end, the rack (2-10) is provided with a storage bin (4-4) at the upper part, the storage bin (4-4) is provided with two groups of mixing roller motors (4-8) and mixing rollers (4-8-1) inside, the storage bin (4-4) is connected with a material falling pipe (4-7) at the lower part, the material falling pipe (4-7) is provided with a material falling head (4-7-1), the material falling head (4-7-1) is provided with a soil turning blade (4-7-2) at the rear, the material falling pipe (4-7) is provided with a rear guide wheel (4-6) at the lower part, the rear guide wheel (4-6) is provided with a front guide wheel (4-5) at the front side, and the rack (2-10) is provided with a wheel (4-9) at the lower side.

3. The intelligent pesticide spraying method of the cultivating and sowing and pesticide spraying integrated apparatus according to claim 2, characterized by: The power module (2) is provided with a solar panel (2-1), a cockpit (2-2) and a visual detection module (2-3), the power module (2) adopts a four-wheel tractor as a power source, the power module (2) is provided with a cockpit (2-2) at the middle, the cockpit (2-2) is provided with a solar panel (2-1) at the top, the power module (2) is provided with a visual detection module (2-3) at the front end, and the power module (2) is provided with a rotary tillage module (1) and a pesticide spraying module (3) at the tail.

4. The intelligent pesticide spraying method of the equipment for cultivating, sowing and spraying pesticides in one according to claim 1, characterized in that: The smooth L1 loss coordinate function in step 2) is represented as: ； wherein represents the difference between the predicted value and the true value, the smooth loss is calculated by respectively calculating the four coordinates of the predicted bounding box, and the final smooth L1 loss value is obtained by weighted sum, and the reverse transmission is carried out.

5. The intelligent pesticide spraying method of the integrated plowing, seeding and pesticide spraying device according to claim 1, characterized in that: The formula in step 3) is combined and expressed as: ； wherein, represents a specific label, for example, a classification task needs to classify A and B, =1 represents the confidence of the model to A, =2 represents the confidence of the model to B, wherein represents the confidence of the first set of models to the object in the photo pair in front, represents the confidence of the second set of models to the object in the photo pair directly above, and are added to obtain the comprehensive confidence of the two models , and then the predicted label is given according to the comprehensive confidence, the back propagation is carried out, and the weight parameters are iteratively updated.

6. The intelligent pesticide spraying method of the integrated plowing, seeding and pesticide spraying device according to claim 1, characterized in that: The formula in step 5) is expressed as a relative spatial volume: ； wherein, , are the width and height of the bounding box taken by the first set of models respectively, is the height taken by the second set of models, according to which algorithm the relative spatial volume of the target object can be distilled.

7. The intelligent pesticide spraying method of the integrated plowing, seeding and pesticide spraying device according to claim 2, characterized in that: The formula in step 7) is expressed as a relative spatial volume: ； wherein, the height of the spray head setting, is the maximum relative spatial volume for a plant is the optimal height set by a human, is the relative spatial volume of the current plant.

Images