A multi-sensor fusion precision spray control system and method

Abstract

The present invention discloses a multi-sensor fusion precision spray control system and method, which relate to the technical field of fruit tree planting, and include a data acquisition module, a fruit tree analysis module, a leaf analysis module, a drug analysis module and a drug spraying module: the data acquisition module is used to collect environmental factors, fruit tree data and spraying equipment data, the environmental factors include wind data and light data during spraying, the fruit tree data includes distribution data of fruit tree leaves and fruit tree leaf image data, and the spraying equipment data includes the equipment inclination angle and the distance between the equipment and the fruit tree; the technical key points are: during the spray fertilization process, the leaves on the fruit trees are analyzed, the healthy leaf value is calculated, whether the fruit trees are healthy is judged according to the healthy leaf value, and a fruit tree treatment plan is analyzed when the fruit trees are abnormal, and spray fertilization is combined with drug spraying, with good use effect and good use prospects.

Description

Technical Field

[0001] The present invention relates to the technical field of fruit tree planting, and in particular to a multi-sensor fusion precision spray control system and method. Background Art

[0002] Modern agriculture refers to agricultural production based on modern natural science, utilizing modern production technologies and agricultural machinery, and combining modern organizational and management methods. Characterized by mechanization, scientific development, industrialization, informatization, and sustainability, modern agriculture emphasizes the modernization of the material and technical conditions for agricultural production, as well as the modernization of agricultural organization and management. Compared to traditional agriculture, in modern agricultural production, science and technology have replaced experience as the guiding principle for production, and mechanical operations have replaced manual and animal operations, improving production efficiency. The development of modern agriculture has gradually transformed agriculture into a competitive modern economic sector, playing a vital role in transforming my country's agricultural growth model, promoting the transformation of the rural industrial structure, increasing farmers' income, and improving the quality of agricultural practitioners.

[0003] With the progress of society, people's awareness of environmental protection is getting stronger and stronger. In order to better protect the environment, effectively improve the efficiency of pesticide use, reduce pollution caused by excessive application of pesticides, and reduce pesticide residues in crops, orchards are gradually adopting precision spraying to apply liquid fertilizers and pesticides. The key to achieving this is to process the above-mentioned differentiated information and spray variables on demand during the growth of crops.

[0004] The existing patent application, CN109984115B, titled "Variable Spray System and Spray Quantity Decision Method for Fruit Trees," describes a system comprising a mobile vehicle, a spray tank mounted on the vehicle, several groups of nozzles arranged from top to bottom, and a radar. The nozzle group includes a pipeline, a solenoid valve switch, a proportional solenoid valve, and several nozzles. One end of the pipeline is connected to the spray tank; the solenoid valve switch and the proportional solenoid valve are installed on the pipeline; several nozzles are arranged in parallel from top to bottom at the other end of the pipeline; all nozzles are located in the same vertical line; and the radar is located in front of the nozzle group. This invention offers advantages such as precise and effective spraying.

[0005] However, based on the above content and the existing technology, the solution recorded in the above patent is mainly to analyze the fruit tree as a whole, calculate the amount of pesticide required for the fruit tree after a comprehensive analysis of the entire fruit tree, and then directly calculate the required spray pressure based on the amount of pesticide applied.

[0006] However, the above scheme has major defects in actual use. First, using the method described above, each fruit tree needs to stop during spraying for comprehensive analysis, scanning and calculation, which takes a long time and is therefore inefficient. Moreover, when fruit trees grow, they are affected by the external environment, and the distribution of leaves in their canopies is different in different directions. If the fruit trees are sprayed with the same scheme, there will inevitably be excessive spraying in some areas and insufficient spraying in some areas. At the same time, its application is too idealistic, and the external environment is less considered. It is difficult to achieve the expected effect in actual application and does not meet people's usage requirements. For this reason, we have developed a multi-sensor fusion precision spray control system and method. Summary of the Invention

[0007] (1) Technical problems solved

[0008] In response to the shortcomings of the existing technology, the present invention provides a multi-sensor fusion precision spray control system and method, which analyzes the leaves on the fruit trees during the spray fertilization process, calculates the healthy leaf value, determines whether the fruit trees are healthy based on the healthy leaf value, and analyzes the fruit tree treatment plan when the fruit trees are abnormal. It combines spray fertilization with spraying, has good use effect, and has good use prospects.

[0009] (2) Technical solution

[0010] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0011] A multi-sensor fusion precision spray control system, including a data acquisition module, a fruit tree analysis module, a leaf analysis module, a pesticide analysis module, and a pesticide spraying module:

[0012] A data acquisition module is used to collect environmental factors, fruit tree data, and spraying equipment data. Environmental factors include wind data and light data during spraying. Fruit tree data includes leaf distribution data and leaf image data. Spraying equipment data includes the equipment tilt angle and the distance between the equipment and the fruit tree.

[0013] The fruit tree analysis module is used to pre-process the fruit tree leaf image data, then calculate the healthy leaf value based on the processed fruit tree leaf image data, and compare the healthy leaf value with the standard range. If the healthy leaf value is within the standard range, no treatment drugs need to be added. If the healthy leaf value is outside the standard range, a leaf abnormality analysis signal is issued;

[0014] The leaf analysis module is used to start when receiving the leaf abnormality analysis signal, perform background segmentation on the abnormal fruit tree leaf image data, and then compare the acquired data with the pest and disease data recorded in the database, and obtain the comparison results and the corresponding treatment agent name and agent addition ratio;

[0015] The drug analysis module is used to calculate the drug flow rate based on the distribution data of the fruit tree leaves, and calculate the spraying speed based on the wind data, the equipment tilt angle and the distance between the equipment and the fruit tree, and generate a spraying plan based on the name and ratio of the treatment drug;

[0016] The pesticide spraying module is used to pre-spray fruit trees according to the spraying plan, obtain feedback data after the fruit trees are sprayed, calculate the spraying value based on the feedback data, and compare the spraying value with the preset standard range. When the spraying value is within the standard range, the fruit trees are sprayed according to the spraying plan. When the spraying value is outside the standard range, the pesticide flow in the spraying plan is adjusted based on the spraying value, a new spraying plan is generated, and the fruit trees are sprayed according to the new spraying plan.

[0017] Furthermore, the wind data and light data are collected by a wind speed and direction detector and a light intensity detector installed on the spraying equipment, the fruit tree leaf image data is magnified and photographed by a photographic component on the spraying equipment, the distance between the equipment and the fruit tree is detected by a laser ranging instrument, and the tilt angle of the equipment is detected by a tilt sensor built into the spraying equipment.

[0018] Furthermore, the steps of obtaining the distribution data of the leaves of the fruit trees are as follows;

[0019] Obtain the device tilt angle detected by the inclination sensor and the distance between the laser ranging instrument and the bottom of the fruit tree, and based on this, calculate the distance L between the laser ranging instrument and the fruit tree, L = L j × cos(β-α), where L j is the distance between the laser distance measuring instrument and the bottom of the fruit tree, α is the tilt angle of the device detected by the inclination sensor, and β is the rotation angle of the laser distance measuring instrument when measuring the distance between the laser distance measuring instrument and the bottom of the fruit tree;

[0020] Adjust the angle of the laser ranging instrument to monitor the highest height of the tree canopy h g and the lowest crown height h d , where h g =h y ×cosα+L×tanα1,h y is the height of the laser ranging instrument when the spraying equipment is not tilted, α1 is the highest height of the tree crown monitored by the laser ranging instrument h g The angle at which the laser ranging instrument rotates upward, h d =h y ×cosα-L×tanα2, α2 is the lowest crown height h monitored by the laser ranging instrument d The angle at which the laser ranging instrument rotates downward;

[0021] Calculate the crown height h, h = h g -h d , calculate the number n of spraying nozzles required, and divide the tree crown into spraying spaces according to the spraying heights of the spraying nozzles, randomly select K points inside a single spraying space, and detect the distance between the laser ranging instrument and the K points;

[0022] Obtain the distance data between the laser ranging instrument and K points, and calculate the average value μ of K distance data j , Among them, Ld i is the distance data between the laser ranging instrument and the i-th point, and the K distance data are respectively compared with the average value μ j Compare and count the number of distance data M that are less than the average value, and then calculate the leaf distribution data SF in the spraying space.

[0023] Furthermore, the calculation formula for the healthy leaf value is as follows:

[0024]

[0025] Where, JZ is the healthy leaf value, D is the number of fruit tree leaf images obtained, R i is the red value in the i-th processed fruit tree leaf image, G i is the green value in the i-th processed fruit tree leaf image, B i is the blue value in the i-th processed fruit tree leaf image, a, b, c are constant coefficients, and 0<b<c<a<1, Lx is the external light intensity during shooting, and Lb is the optimal shooting light intensity;

[0026] The standard interval is (JZmin, JZmax). If the calculated JZ is less than JZmin, a blade abnormality analysis signal is issued;

[0027] If JZmin≤JZ≤JZmax, no additional therapeutic drugs are needed.

[0028] Furthermore, when performing background segmentation on abnormal fruit tree leaf image data, the normalized super-green and super-red operators are used to segment the abnormal fruit tree leaf image data, and then the fruit tree leaf disease image recognition model based on deep learning is used to compare the segmented images with the pest and disease graphic data recorded in the database to identify the cause of the disease and the corresponding treatment agent name and agent addition ratio.

[0029] Furthermore, the calculation formula for the flow rate of the drug required for the spraying space is as follows:

[0030]

[0031] In the formula, Dose is the flow rate of the agent required for the spraying space, To calculate the average rotation angle of the laser ranging instrument when K distance data are obtained, kjxs is the influence ratio of the crown width of the spraying space position on the dosage of the pesticide, f is the constant coefficient, and 0<f<1, gsbz is the basic irrigation flow corresponding to the fruit tree type.

[0032] Furthermore, the formula for calculating the spraying speed based on wind data, the tilt angle of the equipment, and the distance between the equipment and the fruit tree is as follows:

[0033]

[0034] Where Vps is the spraying speed of the spraying nozzle corresponding to the spraying space, BZJL is the set standard spraying distance, jixs is the influence coefficient of the spraying distance on the spraying speed adjustment ratio, fl is the wind speed, z is the constant coefficient, 6<z<10, δ is the angle between the wind direction and the spraying direction, fsxs is the influence coefficient of the wind force on the spraying speed adjustment ratio, V bz is the set standard spraying speed;

[0035] When δ is not zero, adjust the delay time between the completion of data acquisition by the laser ranging instrument and the spraying of the sprinkler nozzle. The specific calculation formula is as follows:

[0036] T=T0-fl×sinδ×yxxs

[0037] Among them, T is the delay time between the completion of laser ranging instrument data acquisition and the spraying of the sprinkler nozzle after adjustment, T0 is the delay time between the completion of laser ranging instrument data acquisition and the spraying of the sprinkler nozzle in the absence of wind, yxxs is the correction coefficient to reduce the influence of wind, 0<yxxs<1.

[0038] Furthermore, the feedback data is the droplet coverage rate at x locations reported by the staff through the water-sensitive paper set on the fruit trees. The formula for calculating the spraying value based on the droplet coverage rate is as follows:

[0039]

[0040] Where PSz is the spraying value, fgl i is the droplet coverage at the i-th position, Ls i is the distance between the droplet coverage and the tree trunk, ρ is the influence coefficient of distance on the droplet coverage, ρ>1.

[0041] Furthermore, the standard range is 7-10%. If the calculated spraying value is 7%≤PSz≤10%, the fruit trees are sprayed according to the spraying plan. When the spraying value is less than 4%<PSz<7% or 10%<PSz<13%, the drug flow rate Dose in the spraying plan is adjusted. The drug flow rate in the adjusted spraying plan is Where λ is the adjustment coefficient. If PSz≤4% or PSz≥13%, calibrate and maintain the spraying equipment, and then re-perform pre-spraying.

[0042] Furthermore, a multi-sensor fusion precision spray control method includes the following steps:

[0043] Collect environmental factors, fruit tree data, and spraying equipment data. Environmental factors include wind data and light data during spraying. Fruit tree data includes leaf distribution data and leaf image data. Spraying equipment data includes equipment tilt angle and distance between the equipment and the fruit tree.

[0044] Preprocess the fruit tree leaf image data, then calculate the healthy leaf value based on the processed fruit tree leaf image data, and compare the healthy leaf value with the standard range. If the healthy leaf value is within the standard range, no treatment drug needs to be added. If the healthy leaf value is outside the standard range, a leaf abnormality analysis signal is issued;

[0045] When receiving the leaf abnormality analysis signal, the background of the abnormal fruit tree leaf image data is segmented, and then the acquired data is compared with the pest and disease data recorded in the database, and the comparison results and the corresponding treatment agent name and agent addition ratio are obtained;

[0046] The flow rate of the drug is calculated based on the distribution data of the fruit tree leaves, and the spraying speed is calculated based on the wind data, the tilt angle of the equipment and the distance between the equipment and the fruit tree. The spraying plan is generated by combining the name and ratio of the treatment drug;

[0047] The fruit trees are pre-sprayed according to the spraying plan, and feedback data after the fruit trees are sprayed is obtained. The spraying value is calculated based on the feedback data, and the spraying value is compared with a preset standard range. When the spraying value is within the standard range, the fruit trees are sprayed according to the spraying plan. When the spraying value is outside the standard range, the flow rate of the pesticide in the spraying plan is adjusted based on the spraying value, a new spraying plan is generated, and the fruit trees are sprayed according to the new spraying plan.

[0048] (3) Beneficial effects

[0049] The present invention provides a multi-sensor fusion precision spray control system and method, which has the following beneficial effects:

[0050] 1. The present invention provides a multi-sensor fusion precision spray control system and method, which analyzes the leaves on the fruit trees during the spray fertilization process, calculates the healthy leaf value, determines whether the fruit trees are healthy based on the healthy leaf value, and analyzes the fruit tree treatment plan when the fruit trees are abnormal. It combines spray fertilization with spraying, and can also be used separately. It has high applicability, good use effect, and good use prospects.

[0051] 2. The present invention records a multi-sensor fusion precision spray control system and method. During the spray fertilization process, it adopts the method of collecting and scanning the canopy of fruit trees, and calculates the corresponding pesticide flow rate based on the real-time collected data. This method can effectively reduce the precision spraying of liquid medicine and fertilizer in the case of irregular fruit trees and uneven distribution of branches and leaves of fruit trees, effectively improve the efficiency of pesticide use, reduce pollution caused by excessive application of pesticides, and reduce pesticide residues in crops.

[0052] 3. The present invention records a multi-sensor fusion precision spray control system and method, which comprehensively analyzes the situation in the orchard, fully considers the influence of equipment tilt caused by uneven ground in the orchard and wind speed and direction on spraying, so that the calculated results are more accurate, and further realizes the precise spraying of medicines and liquid fertilizers. It has a wide range of applicable scenarios, good overall use effect, and good prospects for use. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 This is a flow chart of a multi-sensor fusion precision spray control system of the present invention;

[0054] Figure 2 This is a schematic structural diagram of a spraying device in a multi-sensor fusion precision spray control system of the present invention;

[0055] Figure 3 This is a working schematic diagram of the spraying equipment in a multi-sensor fusion precision spray control system of the present invention. DETAILED DESCRIPTION

[0056] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0057] Research reasons

[0058] In order to spray accurately, the existing spraying scheme will scan the tree canopy over a large area to calculate the volume of the canopy and the number of leaves in the canopy. It requires an overall analysis of the fruit tree. After a comprehensive analysis of the entire fruit tree, the required amount of pesticide to be applied to the fruit tree is calculated, and then the required spray pressure is directly calculated based on the amount of pesticide applied.

[0059] In the above-described method, each fruit tree needs to stop for a period of time during spraying to collect data, conduct comprehensive analysis, scanning and calculation, especially calculate the leaf area index of the fruit tree, which takes a long time and is therefore inefficient.

[0060] In addition, when fruit trees grow, they are affected by the external environment (light, pests and diseases, etc.), and the distribution of leaves in their canopies is different in different directions (for example, there is a big difference in leaf distribution between the sunny side and the dorsum side). If the fruit trees are sprayed with the same scheme, there will inevitably be excessive spraying in some areas and insufficient spraying in some areas. Moreover, the spraying is based on the entire canopy, which is difficult to work for fruit trees with large differences in leaf distribution.

[0061] At the same time, its application is too idealistic, with little consideration for the external environment. It is difficult to achieve the expected results in actual application and does not meet people's usage requirements.

[0062] Design ideas

[0063] Most existing systems are based on the entire fruit tree, and comprehensive data collection of the fruit trees is required in the early stages. However, this method only analyzes the distribution of leaves and the thickness of the fruit tree canopy, which cannot fully reflect the condition of the fruit trees.

[0064] Therefore, during the research and development process, the initial idea was how to analyze fruit trees more comprehensively. Later, it was discovered that based on the existing fruit tree disease and pest identification solution, relevant data of fruit tree leaves can be comprehensively collected during use, and the identification of fruit tree diseases and pests can be realized based on the existing deep learning-based fruit tree leaf disease image recognition model.

[0065] However, in actual use, it was found that the time required for analysis using this method and the existing technology is the same. Both methods require scanning data in front of the fruit trees and analyzing and processing the data, resulting in relatively low efficiency in spraying fruit trees and failing to meet efficiency requirements.

[0066] Therefore, in the middle of the research and development, we changed our thinking, improved the existing leaf recognition technology, and added a step to calculate the leaf health value. This process is collected during the operation of the spraying equipment, and the analysis is fast, without waiting, to analyze whether there are abnormalities in the fruit trees. If there are abnormalities, we will use the existing plan for further detailed analysis, and further change the fruit tree spraying plan to achieve real-time spraying of fruit trees, thereby greatly improving the efficiency of spraying.

[0067] However, in actual use, it was found that during the spraying process, the wind would cause great interference to the spraying. At the same time, the ground under the fruit trees was uneven and the equipment was tilted. These situations would cause great changes in the spraying effect.

[0068] Therefore, in the later stage of research and development, environmental factors were fully considered, and combined with previous solutions, a multi-sensor fusion precision spray control system and method was developed.

[0069] Study plan

[0070] A multi-sensor fusion precision spray control system, which includes two parts: hardware and software, such as Figure 2 As shown, the hardware is a spraying device and corresponding auxiliary devices installed on the spraying device.

[0071] The corresponding auxiliary devices include a wind speed and direction detector (model EC-A2) and a light intensity detector (model GD51-KGZ) installed on the top of the spraying equipment, a laser rangefinder (model XKC-KL200) installed on both sides of the spraying equipment, and a photographic component installed on the front end of the spraying equipment. Electronic instruments with the same functions can also be used.

[0072] The spraying equipment adopts the existing fruit tree spray vehicle with a spray boom composed of multiple groups of nozzles. The auxiliary devices and the fruit tree spray vehicle are connected through the Internet of Things or wired connections, and the fruit tree spray vehicle and the database are connected through the network.

[0073] The software includes data acquisition module, fruit tree analysis module, leaf analysis module, pesticide analysis module and pesticide spraying module:

[0074] Data Acquisition

[0075] In order to spray liquid fertilizers and drugs in a targeted manner, you must first understand the fruit tree data. In order to accurately collect fruit tree data, you must understand environmental factors, based on which you can obtain accurate drug flow. For more comprehensive spraying, you must also understand the spraying equipment data. The collection of environmental factors, fruit tree data, and spraying equipment data is based on the data acquisition module.

[0076] The data acquisition module is used to collect environmental factors, fruit tree data, and spraying equipment data. Environmental factors include wind data and light data during spraying. Fruit tree data includes leaf distribution data and leaf image data. Spraying equipment data includes the equipment tilt angle and the distance between the equipment and the fruit tree.

[0077] Wind data and light data are collected through the wind speed and direction detector and light intensity detector installed on the spraying equipment. The image data of fruit tree leaves are magnified and photographed by the photographic component on the spraying equipment. The distance between the equipment and the fruit tree is detected by a laser rangefinder, and the tilt angle of the equipment is detected by the tilt sensor built into the spraying equipment.

[0078] The steps to obtain the distribution data of fruit tree leaves are as follows;

[0079] Obtain the device tilt angle detected by the inclination sensor and the distance between the laser ranging instrument and the bottom of the fruit tree, and based on this, calculate the distance L between the laser ranging instrument and the fruit tree, L = L j × cos(β-α), where L j The distance between the laser rangefinder and the bottom of the fruit tree is measured by the laser rangefinder. This method is mainly used to avoid interference from branches and ensure the accuracy of detection. α is the tilt angle of the device detected by the inclination sensor (the angle between the device and the horizontal plane. Under normal circumstances, the angle between the device and the horizontal plane is less than 20°, that is, α is less than 20°). β is the rotation angle of the laser rangefinder when measuring the distance between the laser rangefinder and the bottom of the fruit tree.

[0080] like Figure 3 As shown, by the formula L=L j × cos(β-α) calculated as L is the distance from the bottom of the equipment to the fruit trees on the surface. The distance from the equipment to the fruit trees above is L = L j ×cos(β′+α), where β is greater than β′.

[0081] like Figure 3 As shown, in actual use, the road surface in the orchard is often uneven. Therefore, although the laser rangefinder is used to measure and keep the spraying equipment near the middle between the fruit trees, due to the tilt of the vehicle body, spraying in the previous way may cause some areas to be unable to be sprayed.

[0082] Therefore, the influence of the equipment tilt angle must be considered, and the most direct impact of the equipment tilt is that the distance between the spray nozzle and the fruit tree trunk is not equal. After calculating the distance using the above method, subsequent calculations can be facilitated.

[0083] The following calculations are based on the distance from the bottom of the equipment to the fruit trees on the surface ( Figure 3 The calculation principle for the other direction is the same, only the angle is different.

[0084] Adjust the angle of the laser ranging instrument to monitor the highest height of the tree canopy h g and the lowest crown height h d , where h g =h y ×cosα+L×tanα1,h y is the height of the laser rangefinder when the spraying equipment is not tilted. The height changes when the equipment is tilted. In order to improve the accuracy, further calculation is required. α1 is the highest height of the tree crown monitored by the laser rangefinder h g The angle at which the laser ranging instrument rotates upward, h d =h y ×cosα-L×tanα2, α2 is the lowest crown height h monitored by the laser ranging instrument d The angle at which the laser ranging instrument rotates downward;

[0085] The laser distance measuring instrument is installed on the side of the spraying instrument, and its general height is 1.5-2m. Therefore, it is located at the crown of the fruit tree. The main purpose of setting it at this height is to be able to fully measure the highest crown height h g

[0086] Calculate the crown height h, h = h g -h d , calculate the number n of spraying nozzles required, and divide the tree crown into spraying spaces according to the spraying heights of the spraying nozzles, randomly select K points inside a single spraying space, and detect the distance between the laser ranging instrument and the K points;

[0087] For example Figure 3 As shown, h is 4 meters, and the spraying height of the spray nozzle is 0.5 meters, so there are 8 spraying spaces that need to be divided. 4 points are selected inside each spraying space, that is, K=4, so that 4 distances are detected in each spraying space, and 32 distances are detected in 8 spraying spaces.

[0088] In this plan, calculations are performed based on the spraying space, and each subsequent spraying space will form a separate spraying plan. This approach will make spraying more accurate.

[0089] The spraying distance is different and the spraying space height is different, which is set according to the orchard road and the spraying range of the spray nozzle.

[0090] The spray nozzle is equipped with a corresponding solenoid valve, which can be controlled independently, can spray liquid at any flow rate, and can be opened and closed independently.

[0091] Obtain the distance data between the laser ranging instrument and K points, and calculate the average value μ of K distance data j , Among them, Ld i is the distance data between the laser ranging instrument and the i-th point, and the K distance data are respectively compared with the average value μ j Compare and count the number of distance data M that are less than the average value, and then calculate the leaf distribution data SF in the spraying space.

[0092] Less than the average μ j That is, when measuring, the laser passes through the blade layer, which will result in a value smaller than the average value.

[0093] The larger the value of K, the more accurate it is.

[0094] You can also make a further judgment. When it is less than the average value, calculate the difference between the distance data and the average, compare the difference data with the set value. The set value is above 0.2m, and count the number of points where the difference is greater than the set value. This method is more accurate.

[0095] The larger the SF, the sparser the leaves, and the smaller the SF, the more leaves.

[0096] Each of the above-mentioned spraying spaces (the spraying space is monitored in real time, K points are located in the same plane, and the constructed plane is perpendicular to the horizontal plane) calculates the leaf distribution data SF, and then calculates the corresponding required agent flow rate based on the leaf distribution data SF. It belongs to real-time monitoring, and is not the existing technology that directly realizes comprehensive monitoring and analysis of the entire fruit tree canopy. Therefore, the analyzed data is very fast.

[0097] During the spray fertilization process, the method of collecting and scanning the canopy of fruit trees is adopted, and the corresponding pesticide flow rate is calculated based on the real-time collected data. This method can effectively reduce the precise spraying of liquid medicine and fertilizer in the case of irregular fruit trees and uneven distribution of branches and leaves of fruit trees, effectively improve the efficiency of pesticide use, reduce pollution caused by excessive application of pesticides, and reduce pesticide residues in crops.

[0098] Data Analysis

[0099] After the relevant data is collected, it must be further analyzed and calculated, including fruit tree analysis, leaf analysis and drug analysis. Fruit tree analysis is to analyze whether the fruit tree is abnormal and determine whether drug treatment is needed on the basis of existing liquid fertilizer. Leaf analysis is to further analyze the leaves when the fruit tree is abnormal, determine the specific abnormality of the fruit tree, and analyze the required treatment drugs and the proportion of treatment drugs. Drug analysis determines the real-time drug flow and spraying speed required for each canopy of the fruit tree to form a spraying plan.

[0100] Fruit tree analysis is based on the fruit tree analysis module, which pre-processes the fruit tree leaf image data, then calculates the healthy leaf value based on the processed fruit tree leaf image data and compares the healthy leaf value with the standard range. If the healthy leaf value is within the standard range, no treatment drugs are needed. If the healthy leaf value is outside the standard range, a leaf abnormality analysis signal is issued.

[0101] The calculation formula for healthy leaf value is as follows:

[0102]

[0103] Where JZ is the healthy leaf value, D is the number of fruit tree leaf images obtained, and R i is the red value in the i-th processed fruit tree leaf image, G i is the green value in the i-th processed fruit tree leaf image, B i is the blue value in the i-th processed fruit tree leaf image, a, b, c are constant coefficients, and 0<b<c<a<1, Lx is the external light intensity during shooting, and Lb is the optimal shooting light intensity;

[0104] The standard interval is (JZmin, JZmax). If the calculated JZ is less than JZmin, a blade abnormality analysis signal is issued;

[0105] If JZmin≤JZ≤JZmax, no additional therapeutic drugs are needed.

[0106] There is an upper limit to the color of leaves, so there is a maximum value for healthy leaves, namely JZmax.

[0107] There are obvious differences between normal fruit tree leaves and abnormal fruit tree leaves. Almost all of them have the problem that the leaves are not green enough and are yellowing. Therefore, it is possible to judge whether the fruit tree is abnormal by color. However, in reality, due to the influence of external ambient light, there will be large differences. In order to use the same standard for judgment, it is necessary to eliminate the influence of light. After analysis and processing, a method for calculating the healthy leaf value is generated.

[0108] The leaf analysis module is activated upon receiving the leaf abnormality analysis signal, and performs background segmentation on the abnormal fruit tree leaf image data. It then compares the acquired data with the pest and disease data recorded in the database, and obtains the comparison results and the corresponding treatment agent name and agent addition ratio;

[0109] When performing background segmentation on abnormal fruit tree leaf image data, the normalized super-green and super-red operators are used to segment the abnormal fruit tree leaf image data. Then, a fruit tree leaf disease image recognition model based on deep learning is used to compare the segmented images with the pest and disease image data recorded in the database to identify the cause of the disease and the corresponding treatment agent name and agent addition ratio. This method is mostly used in existing pest and disease identification software.

[0110] The use of normalized super-green and super-red operators to segment abnormal fruit tree leaf image data and compare with the fruit tree leaf disease image recognition model based on deep learning is a commonly used pest and disease identification technology in the prior art. Since it is a common technology in this field, the present invention only applies it, so no excessive description is given.

[0111] During the spray fertilization process, the leaves on the fruit trees are analyzed, and the healthy leaf value is calculated. The health of the fruit trees is judged based on the healthy leaf value, and the fruit tree treatment plan is analyzed when the fruit trees are abnormal. Spray fertilization is combined with spraying, and can also be used separately. It has high applicability, good use effect, and good use prospects.

[0112] The drug analysis module calculates the drug flow rate based on the distribution data of the fruit tree leaves, and calculates the spraying speed based on the wind data, the tilt angle of the equipment and the distance between the equipment and the fruit tree, and generates a spraying plan based on the name and ratio of the treatment drug;

[0113] The calculation formula for the flow rate of the spraying space required is as follows:

[0114]

[0115] In the formula, Dose is the flow rate of the agent required for the spraying space, To calculate the average rotation angle of the laser ranging instrument when K distance data are obtained, kjxs is the influence ratio of the crown width of the spraying space position on the dosage of the pesticide, f is the constant coefficient, and 0<f<1, gsbz is the basic irrigation flow corresponding to the fruit tree type.

[0116] Different fruit trees have different leaf sizes and shapes, and their leaf areas are completely different. Therefore, the basic irrigation flow rate gsbz they require is also different.

[0117] The present invention mainly collects data through a laser ranging instrument, which does not require multiple equipment collection instruments and has a relatively low overall cost.

[0118] Because wind can change the direction and speed of the agent's movement, causing it to deviate, the impact of wind must be considered. Wind is an uncontrollable factor, so real-time data collection is also required.

[0119] The formula for calculating the spraying speed based on wind data, the tilt angle of the equipment and the distance between the equipment and the fruit tree is as follows:

[0120]

[0121] Where Vps is the spraying speed of the spraying nozzle corresponding to the spraying space, BZJL is the set standard spraying distance, jixs is the influence coefficient of the spraying distance on the spraying speed adjustment ratio, fl is the wind speed, z is the constant coefficient, 6<z<10, δ is the angle between the wind direction and the spraying direction, fsxs is the influence coefficient of the wind force on the spraying speed adjustment ratio, V bz is the set standard spraying speed;

[0122] Standard spraying speed is for standard distance and in no wind conditions.

[0123] Usually the standard spraying distance is 1-2 meters from the outermost edge of the tree crown.

[0124] When δ is not zero, adjust the delay time between the completion of data acquisition by the laser ranging instrument and the spraying of the sprinkler nozzle. The specific calculation formula is as follows:

[0125] T=T0-fl×sinδ×yxxs

[0126] Among them, T is the delay time between the completion of laser ranging instrument data acquisition and the spraying of the sprinkler nozzle after adjustment, T0 is the delay time between the completion of laser ranging instrument data acquisition and the spraying of the sprinkler nozzle in the absence of wind, yxxs is the correction coefficient to reduce the influence of wind, 0<yxxs<1.

[0127] During the above adjustment process, if δ≠0° and δ≠180°, there is no need to adjust the delay time for 0° and 180°.

[0128] Affected by air resistance, the speed at which the liquid medicine moves due to wind force will be lower than the speed of the wind.

[0129] T0 is usually the time required for the equipment to move from the position of the laser ranging instrument facing the fruit tree to the position of the spray nozzle facing the fruit tree. This time is usually 10-15 seconds, and the time required for calculation is 5-8 seconds.

[0130] Spraying will only be carried out when the wind speed is less than 5m / s, so the impact caused by the wind is less than 2 seconds, which is completely sufficient for regulation.

[0131] After adjustment, the sprayed liquid can accurately fall on the leaves of fruit trees, reducing the impact of wind and avoiding pesticide pollution and waste.

[0132] Comprehensively analyzing the situation in the orchard, fully considering the impact of equipment tilt caused by uneven ground in the orchard and wind speed and direction on spraying, makes the calculated results more accurate, and further realizes the precise spraying of medicines and liquid fertilizers. It has a wide range of applicable scenarios, good overall use effect, and good prospects for use.

[0133] Spraying of chemicals

[0134] After developing a corresponding plan, it is necessary to verify whether the plan is standard. If it is standard, large-scale spraying can be carried out. If it is not standard, the spraying plan needs to be further adjusted. This step is based on the pesticide spraying module.

[0135] The pesticide spraying module is used to pre-spray fruit trees according to the spraying plan, obtain feedback data after the fruit trees are sprayed, calculate the spraying value based on the feedback data, and compare the spraying value with the preset standard range. When the spraying value is within the standard range, the fruit trees are sprayed according to the spraying plan. When the spraying value is outside the standard range, the pesticide flow in the spraying plan is adjusted based on the spraying value, a new spraying plan is generated, and the fruit trees are sprayed according to the new spraying plan.

[0136] The feedback data is the droplet coverage rate at x locations reported by the staff through the water-sensitive paper set on the fruit trees. The formula for calculating the spraying value based on the droplet coverage rate is as follows:

[0137]

[0138] Where PSz is the spraying value, fgl i is the droplet coverage at the i-th position, Ls i is the distance between the droplet coverage and the tree trunk, ρ is the influence coefficient of distance on the droplet coverage, ρ>1.

[0139] The standard range is 7-10%. If the calculated spraying value is 7%≤PSz≤10%, spray the fruit trees according to the spraying plan. When the spraying value is less than 4%<PSz<7% or 10%<PSz<13%, adjust the drug flow rate Dose in the spraying plan. After adjustment, the drug flow rate in the spraying plan Where λ is the adjustment coefficient. If PSz≤4% or PSz≥13%, calibrate and maintain the spraying equipment, and then re-perform pre-spraying.

[0140] The following table shows the droplet coverage and total amount of medicine used when fertilizing fruit trees at a distance of 100 meters using a conventional spraying scheme with spray equipment and this scheme.

[0141]

[0142] The solution described in the present invention can achieve uniform and stable fertilization of fruit trees, effectively avoid over-fertilization, greatly reduce the amount of medicine used, and has good use effect.

[0143] The weight coefficient is determined using the coefficient of variation method, which is a method of assigning weights to each indicator based on the degree of variation between the current value of each evaluation indicator and the target value. If the numerical difference of an indicator is large, it can clearly distinguish the evaluated objects, indicating that the indicator has rich discrimination information, and thus the indicator should be given a larger weight. On the contrary, if the numerical difference of each evaluated object on a certain indicator is small, then the ability of this indicator to distinguish the evaluation objects is weak, and thus the indicator should be given a smaller weight. This method directly uses the information contained in each indicator to obtain the weight of the indicator through calculation, and therefore is objective.

[0144] A multi-sensor fusion precision spray control method includes the following steps:

[0145] Collect environmental factors, fruit tree data, and spraying equipment data. Environmental factors include wind data and light data during spraying. Fruit tree data includes leaf distribution data and leaf image data. Spraying equipment data includes equipment tilt angle and distance between the equipment and the fruit tree.

[0146] Preprocess the fruit tree leaf image data, then calculate the healthy leaf value based on the processed fruit tree leaf image data, and compare the healthy leaf value with the standard range. If the healthy leaf value is within the standard range, no treatment drug needs to be added. If the healthy leaf value is outside the standard range, a leaf abnormality analysis signal is issued;

[0147] When receiving the leaf abnormality analysis signal, the background of the abnormal fruit tree leaf image data is segmented, and then the acquired data is compared with the pest and disease data recorded in the database, and the comparison results and the corresponding treatment agent name and agent addition ratio are obtained;

[0148] The flow rate of the drug is calculated based on the distribution data of the fruit tree leaves, and the spraying speed is calculated based on the wind data, the tilt angle of the equipment and the distance between the equipment and the fruit tree. The spraying plan is generated by combining the name and ratio of the treatment drug;

[0149] Pre-spray the fruit trees according to the spraying plan, obtain feedback data after the fruit trees are sprayed, calculate the spraying value based on the feedback data, and compare the spraying value with the preset standard range. If the calculated spraying value is 7% ≤ PSz ≤ 10%, spray the fruit trees according to the spraying plan. When the spraying value is less than 4% < PSz < 7% or 10% < PSz < 13%, adjust the drug flow rate Dose in the spraying plan, and adjust the drug flow rate in the spraying plan. Where λ is the adjustment coefficient. Spray the fruit trees according to the adjusted plan. If PSz≤4% or PSz≥13%, calibrate and maintain the spraying equipment, and then repeat the pre-spraying step.

[0150] The above formula is a formula obtained by collecting a large amount of data and performing software simulation to obtain the latest real situation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0151] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will appreciate that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution.

[0152] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment as needed.

[0153] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A multi-sensor fusion precision spray control system, characterized in that: include: A data acquisition module is used to collect environmental factors, fruit tree data, and spraying equipment data. Environmental factors include wind data and light data during spraying. Fruit tree data includes leaf distribution data and leaf image data. Spraying equipment data includes the equipment tilt angle and the distance between the equipment and the fruit tree. The fruit tree analysis module is used to pre-process the fruit tree leaf image data, then calculate the healthy leaf value based on the processed fruit tree leaf image data, and compare the healthy leaf value with the standard range. If the healthy leaf value is within the standard range, no treatment drugs need to be added. If the healthy leaf value is outside the standard range, a leaf abnormality analysis signal is issued; The leaf analysis module is used to start when receiving the leaf abnormality analysis signal, perform background segmentation on the abnormal fruit tree leaf image data, and then compare the acquired data with the pest and disease data recorded in the database, and obtain the comparison results and the corresponding treatment agent name and agent addition ratio; The drug analysis module is used to calculate the drug flow rate based on the distribution data of the fruit tree leaves, and calculate the spraying speed based on the wind data, the equipment tilt angle and the distance between the equipment and the fruit tree, and generate a spraying plan based on the name and ratio of the treatment drug; The pesticide spraying module is used to pre-spray fruit trees according to the spraying plan, obtain feedback data after the fruit trees are sprayed, calculate the spraying value based on the feedback data, and compare the spraying value with the preset standard range. When the spraying value is within the standard range, the fruit trees are sprayed according to the spraying plan. When the spraying value is outside the standard range, the pesticide flow in the spraying plan is adjusted based on the spraying value, a new spraying plan is generated, and the fruit trees are sprayed according to the new spraying plan.

2. The multi-sensor fusion precision spray control system according to claim 1, characterized in that: The wind data and light data are collected by a wind speed and direction detector and a light intensity detector installed on the spraying equipment. The fruit tree leaf image data is magnified and photographed by a photographic component on the spraying equipment. The distance between the equipment and the fruit tree is detected by a laser rangefinder, and the tilt angle of the equipment is detected by a tilt sensor built into the spraying equipment.

3. The multi-sensor fusion precision spray control system according to claim 2, characterized in that: The steps to obtain the distribution data of fruit tree leaves are as follows; Obtain the device tilt angle detected by the inclination sensor and the distance between the laser ranging instrument and the bottom of the fruit tree, and based on this, calculate the distance L between the laser ranging instrument and the fruit tree, L = L j × cos(β-α), where L j is the distance between the laser distance measuring instrument and the bottom of the fruit tree, α is the tilt angle of the device detected by the inclination sensor, and β is the rotation angle of the laser distance measuring instrument when measuring the distance between the laser distance measuring instrument and the bottom of the fruit tree; Adjust the angle of the laser ranging instrument to monitor the highest height of the tree canopy h g and the lowest crown height h d , where h g =h y ×cosα+L×tanα1,h y is the height of the laser ranging instrument when the spraying equipment is not tilted, α1 is the highest height of the tree crown monitored by the laser ranging instrument h g The angle at which the laser rangefinder rotates upward, h d =h y ×cosα-L×tanα2, α2 is the lowest crown height h monitored by the laser ranging instrument d The angle at which the laser ranging instrument rotates downward; Calculate the crown height h, h = h g -h d , calculate the number n of spraying nozzles required, and divide the tree crown into spraying spaces according to the spraying heights of the spraying nozzles, randomly select K points inside a single spraying space, and detect the distance between the laser ranging instrument and the K points; Obtain the distance data between the laser ranging instrument and K points, and calculate the average value μ of K distance data j , Among them, Ld i is the distance data between the laser ranging instrument and the i-th point, and the K distance data are respectively compared with the average value μ j Compare and count the number of distance data M that are less than the average value, and then calculate the leaf distribution data SF in the spraying space.

4. The multi-sensor fusion precision spray control system according to claim 3, characterized in that: The calculation formula for healthy leaf value is as follows: Where, JZ is the healthy leaf value, D is the number of fruit tree leaf images obtained, R i is the red value in the i-th processed fruit tree leaf image, G i is the green value in the i-th processed fruit tree leaf image, B i is the blue value in the i-th processed fruit tree leaf image, a, b, c are constant coefficients, and 0<b<c<a<1, Lx is the external light intensity during shooting, and Lb is the optimal shooting light intensity; The standard interval is (JZmin, JZmax). If the calculated JZ is less than JZmin, a blade abnormality analysis signal is issued; If JZmin≤JZ≤JZmax, no additional therapeutic drugs are needed.

5. The multi-sensor fusion precision spray control system according to claim 4, characterized in that: When performing background segmentation on abnormal fruit tree leaf image data, the normalized super-green and super-red operators are used to segment the abnormal fruit tree leaf image data. Then, a fruit tree leaf disease image recognition model based on deep learning is used to compare the segmented images with the pest and disease image data recorded in the database to identify the cause of the disease and the corresponding treatment agent name and agent addition ratio.

6. The multi-sensor fusion precision spray control system according to claim 5, characterized in that: The calculation formula for the flow rate of the spraying space required is as follows: In the formula, Dose is the flow rate of the agent required for the spraying space, To calculate the average rotation angle of the laser ranging instrument when K distance data are obtained, kjxs is the influence ratio of the crown width of the spraying space position on the dosage of the pesticide, f is the constant coefficient, and 0<f<1, gsbz is the basic irrigation flow corresponding to the fruit tree type.

7. The multi-sensor fusion precision spray control system according to claim 6, characterized in that: The formula for calculating the spraying speed based on wind data, the tilt angle of the equipment and the distance between the equipment and the fruit tree is as follows: Where Vps is the spraying speed of the spraying nozzle corresponding to the spraying space, BZJL is the set standard spraying distance, jixs is the influence coefficient of the spraying distance on the spraying speed adjustment ratio, fl is the wind speed, z is the constant coefficient, 6<z<10, δ is the angle between the wind direction and the spraying direction, fsxs is the influence coefficient of the wind force on the spraying speed adjustment ratio, V bz is the set standard spraying speed; When δ is not zero, adjust the delay time between the completion of data acquisition by the laser ranging instrument and the spraying of the sprinkler nozzle. The specific calculation formula is as follows: T=T0-fl×sinδ×yxxs Among them, T is the delay time between the completion of laser ranging instrument data acquisition and the spraying of the sprinkler nozzle after adjustment, T0 is the delay time between the completion of laser ranging instrument data acquisition and the spraying of the sprinkler nozzle in the absence of wind, yxxs is the correction coefficient to reduce the influence of wind, 0<yxxs<1.

8. The multi-sensor fusion precision spray control system according to claim 7, characterized in that: The feedback data is the droplet coverage rate at x locations reported by the staff through the water-sensitive paper set on the fruit trees. The formula for calculating the spraying value based on the droplet coverage rate is as follows: Where PSz is the spraying value, fgl i is the droplet coverage at the i-th position, Ls i is the distance between the droplet coverage and the tree trunk, ρ is the influence coefficient of distance on the droplet coverage, ρ>1.

9. The multi-sensor fusion precision spray control system according to claim 8, characterized in that: The standard range is 7-10%. If the calculated spraying value is 7%≤PSz≤10%, spray the fruit trees according to the spraying plan. When the spraying value is less than 4%<PSz<7% or 10%<PSz<13%, adjust the drug flow rate Dose in the spraying plan. After adjustment, the drug flow rate in the spraying plan Where λ is the adjustment coefficient. If PSz≤4% or PSz≥13%, calibrate and maintain the spraying equipment, and then re-perform pre-spraying.

10. A multi-sensor fusion precision spray control method, using the system according to any one of claims 1 to 9, characterized in that: The steps include: Collect environmental factors, fruit tree data, and spraying equipment data. Environmental factors include wind data and light data during spraying. Fruit tree data includes leaf distribution data and leaf image data. Spraying equipment data includes equipment tilt angle and distance between the equipment and the fruit tree. Preprocess the fruit tree leaf image data, then calculate the healthy leaf value based on the processed fruit tree leaf image data, and compare the healthy leaf value with the standard range. If the healthy leaf value is within the standard range, no treatment drug needs to be added. If the healthy leaf value is outside the standard range, a leaf abnormality analysis signal is issued; When receiving the leaf abnormality analysis signal, the background of the abnormal fruit tree leaf image data is segmented, and then the acquired data is compared with the pest and disease data recorded in the database, and the comparison results and the corresponding treatment agent name and agent addition ratio are obtained; The flow rate of the drug is calculated based on the distribution data of the fruit tree leaves, and the spraying speed is calculated based on the wind data, the tilt angle of the equipment and the distance between the equipment and the fruit tree. The spraying plan is generated by combining the name and ratio of the treatment drug; The fruit trees are pre-sprayed according to the spraying plan, and feedback data after the fruit trees are sprayed is obtained. The spraying value is calculated based on the feedback data, and the spraying value is compared with a preset standard range. When the spraying value is within the standard range, the fruit trees are sprayed according to the spraying plan. When the spraying value is outside the standard range, the flow rate of the pesticide in the spraying plan is adjusted based on the spraying value, a new spraying plan is generated, and the fruit trees are sprayed according to the new spraying plan.

Images