A drift reduction method for pesticide spray drift

Abstract

The present application relates to the technical field of pesticide application, and particularly relates to a pesticide spraying drift prevention method, obtaining plant image information to be identified, identifying the plant to be identified according to the plant image information to be identified, obtaining a three-dimensional model of a plant to be sprayed if the identification result is a first identification result, segmenting the three-dimensional model of the plant to be sprayed into a plurality of sub-spraying areas, obtaining a preset spraying path according to characteristic parameters, spraying starting point information and the three-dimensional model of the plant to be sprayed, obtaining real-time spraying environment parameters at each spraying time node, determining a real-time spraying drift amount according to the real-time spraying environment parameters, and correcting and adjusting the preset spraying path according to the real-time spraying drift amount. The method can improve the deposition of pesticides in a target area, reduce pollution to non-target areas, protect the ecological environment, improve the use effect of pesticides and reduce resource waste.

Description

Technical Field

[0001] This invention relates to the field of pesticide application technology, and in particular to a method for preventing pesticide drift during pesticide spraying. Background Technology

[0002] In modern agricultural production, pesticide spraying is an important agricultural practice used to control pests and diseases and promote crop growth. However, pesticide drift during spraying can cause pesticides to deposit outside the target area, potentially harming non-target areas and the ecological environment. Currently, some methods exist to reduce pesticide drift, such as adjusting spraying parameters and improving spraying equipment. However, existing methods cannot completely eliminate pesticide drift to a certain extent, especially under complex weather conditions. Therefore, there is a need to propose a more efficient method for preventing pesticide drift during spraying to more effectively reduce pesticide drift during the spraying process, thereby reducing environmental pollution and the impact on non-target areas. Summary of the Invention

[0003] This invention overcomes the shortcomings of the prior art and provides a method for preventing pesticide spraying drift.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] The first aspect of this invention discloses a method for preventing pesticide spraying drift, comprising the following steps:

[0006] Obtain image information of the plant to be identified, identify the plant to be identified based on the image information, and obtain a first identification result or a second identification result;

[0007] If the recognition result is the first recognition result, then the image information of the plant to be sprayed is obtained, and the image information of the plant to be sprayed is subjected to redundancy reduction processing to obtain the redundancy-reduced image information of the plant to be sprayed; a three-dimensional model of the plant to be sprayed is obtained based on the redundancy-reduced image information of the plant to be sprayed.

[0008] The three-dimensional model of the plant to be sprayed is divided into several sub-spraying areas, and the characteristic parameters of the plant to be sprayed in each sub-spraying area are obtained, as well as the spraying starting point information of the spraying equipment. Based on the characteristic parameters, the spraying starting point information and the three-dimensional model of the plant to be sprayed, a preset spraying path is obtained.

[0009] The spraying device is controlled to spray the plants to be sprayed based on the preset spraying path; real-time spraying environment parameters are acquired at each spraying time point, the real-time spraying drift is determined based on the real-time spraying environment parameters, and the preset spraying path is corrected and adjusted based on the real-time spraying drift.

[0010] Furthermore, in a preferred embodiment of the present invention, acquiring image information of the plant to be identified, and identifying the plant to be identified based on the image information to obtain a first identification result or a second identification result, includes the following steps:

[0011] Obtain spraying task information, generate search tags based on the spraying task information, search the big data network based on the search tags, and retrieve the feature image information of the plant to be sprayed at each growth stage.

[0012] A knowledge graph is constructed, and the feature image information of the plants to be sprayed at each growth stage is imported into the knowledge graph;

[0013] Obtain the image information of the plant to be identified, import the image information of the plant to be identified into the knowledge graph, and calculate the Euclidean distance value between the image information of the plant to be identified and each feature image information using the Euclidean distance algorithm;

[0014] Based on the Euclidean distance value, the pairing rate between the plant image information to be identified and each feature image information is determined, multiple pairing rates are obtained, and the multiple pairing rates are compared one by one with the preset pairing.

[0015] If at least one pairing rate is greater than the preset pairing rate, the plant to be identified is marked as a plant to be sprayed and a first identification result is generated; if multiple pairing rates are not greater than the preset pairing rate, the plant to be identified is marked as a plant that does not need to be sprayed and a second identification result is generated.

[0016] Further, in a preferred embodiment of the present invention, if the identification result is the first identification result, then the image information of the plant to be sprayed is obtained, and the image information of the plant to be sprayed is subjected to redundancy reduction processing to obtain the redundancy-reduced image information of the plant to be sprayed, including the following steps:

[0017] If the identification result is the first identification result, then the image information of the plant to be sprayed is obtained, and the image information of the plant to be sprayed is decomposed to obtain a diagonal matrix composed of rows and an orthogonal matrix composed of columns;

[0018] Construct a two-dimensional coordinate system, where the X-axis represents a diagonal matrix and the Y-axis represents an orthogonal matrix; use the rows and columns of the diagonal and orthogonal matrices as the scales of the X and Y axes in the two-dimensional coordinate system, respectively.

[0019] The diagonal matrix formed by rows and the orthogonal matrix formed by columns are imported into the two-dimensional coordinate system to generate a point cloud data matrix of descriptors in the image of the plant to be sprayed, and a set of point cloud data coordinates of descriptors in the image of the plant to be sprayed is generated based on the point cloud data matrix.

[0020] Obtain the limit coordinate point set of the point cloud data coordinate set of the descriptor in the image of the plant to be sprayed, import the limit coordinate point set into the world coordinate system for recombination, and obtain the redundancy-reduced image information of the plant to be sprayed.

[0021] Furthermore, in a preferred embodiment of the present invention, obtaining a three-dimensional model of the plant to be sprayed based on the reduced redundancy image information includes the following steps:

[0022] The redundancy-reduced image information of the plants to be sprayed is processed by feature extraction to obtain several paired points; the local outlier factor value of each paired point is calculated by the local outlier factor algorithm, and paired points with local outlier factor values ​​greater than the preset local outlier factor value are removed to obtain sparse paired points.

[0023] Randomly select any sparse pairing point as the origin of the coordinate system, establish a spatial coordinate system based on the origin, and obtain the coordinate information of all sparse pairing points in the spatial coordinate system; calculate the Euclidean distance between each sparse pairing point based on the coordinate information.

[0024] Based on the Euclidean distance between each sparse pairing point, the nearest neighbor of each sparse pairing point is determined, and the median coordinate point between each sparse pairing point and its corresponding nearest neighbor is obtained. The median coordinate point is then marked as a new pairing point.

[0025] The sparse pairing points are combined with new pairing points to obtain dense pairing points; the point cloud data corresponding to the dense pairing points is obtained, and the point cloud data corresponding to the dense pairing points is processed into a grid to obtain a three-dimensional model of the plant to be sprayed.

[0026] Further, in a preferred embodiment of the present invention, the three-dimensional model of the plant to be sprayed is divided into several sub-spraying areas, and the characteristic parameters of the plant to be sprayed in each sub-spraying area are obtained, as well as the spraying start point information of the spraying equipment is obtained. A preset spraying path is obtained based on the characteristic parameters, the spraying start point information, and the three-dimensional model of the plant to be sprayed, including the following steps:

[0027] The three-dimensional model of the plant to be sprayed is divided into several sub-spraying areas, and the characteristic parameters of the plant to be sprayed in each sub-spraying area are obtained based on the three-dimensional model of the plant to be sprayed; wherein the characteristic parameters include leaf density, stem diameter, leaf shape, number of branches and degree of pest infestation;

[0028] Initialize the operating parameters of the spraying equipment and obtain the spraying start point information of the spraying equipment; input the spraying start point information of the spraying equipment, the characteristic parameters of the plant to be sprayed, and the three-dimensional model of the plant to be sprayed into the particle swarm optimization algorithm;

[0029] After iterative calculation using the particle swarm optimization algorithm, several spraying paths for the spraying equipment are obtained, and the distance values ​​corresponding to these spraying paths are acquired. A sequence list is constructed, and the distance values ​​corresponding to the spraying paths are imported into the sequence list and sorted by size.

[0030] After sorting, the shortest path value is extracted, and the spraying path corresponding to the shortest path value is marked as the preset spraying path.

[0031] Furthermore, in a preferred embodiment of the present invention, real-time spraying environment parameters are acquired at each spraying time point, real-time spraying drift is determined based on the real-time spraying environment parameters, and the preset spraying path is corrected and adjusted based on the real-time spraying drift, including the following steps:

[0032] The spray drift amount under various preset spraying environmental parameter combinations is obtained through big data network, a database is constructed, and the spray drift amount under various preset spraying environmental parameter combinations is imported into the database to obtain a characteristic database.

[0033] Real-time spraying environment parameters are obtained and imported into the feature database. The similarity between the real-time spraying environment parameters and various preset spraying environment parameter combinations is calculated using grey relational analysis to obtain multiple similarity scores.

[0034] Extract the maximum similarity from the multiple similarities, obtain the preset spraying environment parameter combination corresponding to the maximum similarity, and determine the real-time spraying drift of the spraying equipment under the current real-time spraying environment parameter conditions based on the preset spraying environment parameter combination corresponding to the maximum similarity.

[0035] The real-time spray drift amount is compared with the preset spray drift amount. If the real-time spray drift amount is greater than the preset spray drift amount, the position of the application axis point at the current spray time node is obtained according to the preset spray path.

[0036] The spraying correction amount is obtained based on the position of the application axis point and the real-time spraying drift amount, and the position of the application axis point is adjusted according to the spraying correction amount.

[0037] A second aspect of this invention discloses a pesticide spraying anti-drift system, the system comprising a memory and a processor, wherein the memory stores an anti-drift method program, and when the processor executes the anti-drift method program, the following steps are performed:

[0038] Obtain image information of the plant to be identified, identify the plant to be identified based on the image information, and obtain a first identification result or a second identification result;

[0039] If the recognition result is the first recognition result, then the image information of the plant to be sprayed is obtained, and the image information of the plant to be sprayed is subjected to redundancy reduction processing to obtain the redundancy-reduced image information of the plant to be sprayed; a three-dimensional model of the plant to be sprayed is obtained based on the redundancy-reduced image information of the plant to be sprayed.

[0040] The three-dimensional model of the plant to be sprayed is divided into several sub-spraying areas, and the characteristic parameters of the plant to be sprayed in each sub-spraying area are obtained, as well as the spraying starting point information of the spraying equipment. Based on the characteristic parameters, the spraying starting point information and the three-dimensional model of the plant to be sprayed, a preset spraying path is obtained.

[0041] The spraying device is controlled to spray the plants to be sprayed based on the preset spraying path; real-time spraying environment parameters are acquired at each spraying time point, the real-time spraying drift is determined based on the real-time spraying environment parameters, and the preset spraying path is corrected and adjusted based on the real-time spraying drift.

[0042] Furthermore, in a preferred embodiment of the present invention, acquiring image information of the plant to be identified, and identifying the plant to be identified based on the image information to obtain a first identification result or a second identification result, includes the following steps:

[0043] Obtain spraying task information, generate search tags based on the spraying task information, search the big data network based on the search tags, and retrieve the feature image information of the plant to be sprayed at each growth stage.

[0044] A knowledge graph is constructed, and the feature image information of the plants to be sprayed at each growth stage is imported into the knowledge graph;

[0045] Obtain the image information of the plant to be identified, import the image information of the plant to be identified into the knowledge graph, and calculate the Euclidean distance value between the image information of the plant to be identified and each feature image information using the Euclidean distance algorithm;

[0046] Based on the Euclidean distance value, the pairing rate between the plant image information to be identified and each feature image information is determined, multiple pairing rates are obtained, and the multiple pairing rates are compared one by one with the preset pairing.

[0047] If at least one pairing rate is greater than the preset pairing rate, the plant to be identified is marked as a plant to be sprayed and a first identification result is generated; if multiple pairing rates are not greater than the preset pairing rate, the plant to be identified is marked as a plant that does not need to be sprayed and a second identification result is generated.

[0048] Further, in a preferred embodiment of the present invention, the three-dimensional model of the plant to be sprayed is divided into several sub-spraying areas, and the characteristic parameters of the plant to be sprayed in each sub-spraying area are obtained, as well as the spraying start point information of the spraying equipment is obtained. A preset spraying path is obtained based on the characteristic parameters, the spraying start point information, and the three-dimensional model of the plant to be sprayed, including the following steps:

[0049] The three-dimensional model of the plant to be sprayed is divided into several sub-spraying areas, and the characteristic parameters of the plant to be sprayed in each sub-spraying area are obtained based on the three-dimensional model of the plant to be sprayed; wherein the characteristic parameters include leaf density, stem diameter, leaf shape, number of branches and degree of pest infestation;

[0050] Initialize the operating parameters of the spraying equipment and obtain the spraying start point information of the spraying equipment; input the spraying start point information of the spraying equipment, the characteristic parameters of the plant to be sprayed, and the three-dimensional model of the plant to be sprayed into the particle swarm optimization algorithm;

[0051] After iterative calculation using the particle swarm optimization algorithm, several spraying paths for the spraying equipment are obtained, and the distance values ​​corresponding to these spraying paths are acquired. A sequence list is constructed, and the distance values ​​corresponding to the spraying paths are imported into the sequence list and sorted by size.

[0052] After sorting, the shortest path value is extracted, and the spraying path corresponding to the shortest path value is marked as the preset spraying path.

[0053] Furthermore, in a preferred embodiment of the present invention, real-time spraying environment parameters are acquired at each spraying time point, real-time spraying drift is determined based on the real-time spraying environment parameters, and the preset spraying path is corrected and adjusted based on the real-time spraying drift, including the following steps:

[0054] The spray drift amount under various preset spraying environmental parameter combinations is obtained through big data network, a database is constructed, and the spray drift amount under various preset spraying environmental parameter combinations is imported into the database to obtain a characteristic database.

[0055] Real-time spraying environment parameters are obtained and imported into the feature database. The similarity between the real-time spraying environment parameters and various preset spraying environment parameter combinations is calculated using grey relational analysis to obtain multiple similarity scores.

[0056] Extract the maximum similarity from the multiple similarities, obtain the preset spraying environment parameter combination corresponding to the maximum similarity, and determine the real-time spraying drift of the spraying equipment under the current real-time spraying environment parameter conditions based on the preset spraying environment parameter combination corresponding to the maximum similarity.

[0057] The real-time spray drift amount is compared with the preset spray drift amount. If the real-time spray drift amount is greater than the preset spray drift amount, the position of the application axis point at the current spray time node is obtained according to the preset spray path.

[0058] The spraying correction amount is obtained based on the position of the application axis point and the real-time spraying drift amount, and the position of the application axis point is adjusted according to the spraying correction amount.

[0059] This invention addresses the technical deficiencies in the prior art and possesses the following beneficial effects: It acquires image information of a plant to be identified; identifies the plant based on the image information to obtain a first identification result or a second identification result; if the identification result is the first identification result, it acquires image information of a plant to be sprayed and performs redundancy reduction processing on the image information to obtain redundancy-reduced image information; it obtains a three-dimensional model of the plant to be sprayed based on the redundancy-reduced image information; it divides the three-dimensional model of the plant to be sprayed into several sub-spraying areas, acquires feature parameters of the plant to be sprayed in each sub-spraying area, and acquires the spraying start point information of the spraying equipment; it obtains a preset spraying path based on the feature parameters, the spraying start point information, and the three-dimensional model of the plant to be sprayed; it controls the spraying equipment to spray the plant based on the preset spraying path; it acquires real-time spraying environment parameters at each spraying time node, determines the real-time spraying drift based on the real-time spraying environment parameters, and corrects and adjusts the preset spraying path based on the real-time spraying drift. This method can improve pesticide deposition in the target area, reduce pollution in non-target areas, protect the ecological environment, improve pesticide effectiveness, and reduce resource waste. Attached Figure Description

[0060] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained from these drawings without creative effort.

[0061] Figure 1 A flowchart of the first method for preventing pesticide spraying drift;

[0062] Figure 2 A flowchart of the second method for preventing pesticide spraying drift;

[0063] Figure 3 A flowchart of a third method for preventing pesticide spraying drift;

[0064] Figure 4 This is a system block diagram of a pesticide spraying anti-drift system. Detailed Implementation

[0065] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0066] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0067] like Figure 1 As shown, the first aspect of this invention discloses a method for preventing pesticide spraying drift, comprising the following steps:

[0068] S102: Obtain image information of the plant to be identified, and identify the plant to be identified based on the image information to obtain a first identification result or a second identification result;

[0069] S104: If the recognition result is the first recognition result, then obtain the image information of the plant to be sprayed, and perform redundancy reduction processing on the image information of the plant to be sprayed to obtain the redundancy-reduced image information of the plant to be sprayed; obtain a three-dimensional model of the plant to be sprayed based on the redundancy-reduced image information of the plant to be sprayed.

[0070] S106: Divide the three-dimensional model of the plant to be sprayed into several sub-spraying areas, obtain the characteristic parameters of the plant to be sprayed in each sub-spraying area, and obtain the spraying starting point information of the spraying equipment. Based on the characteristic parameters, the spraying starting point information and the three-dimensional model of the plant to be sprayed, obtain the preset spraying path.

[0071] S108: Control the spraying equipment to spray the plants to be sprayed based on the preset spraying path; obtain real-time spraying environment parameters at each spraying time node, determine the real-time spraying drift amount based on the real-time spraying environment parameters, and correct and adjust the preset spraying path based on the real-time spraying drift amount.

[0072] The application equipment is an unmanned application robot.

[0073] The process of acquiring image information of the plant to be identified, and identifying the plant based on the image information to obtain a first identification result or a second identification result, includes the following steps:

[0074] Obtain spraying task information, generate search tags based on the spraying task information, search the big data network based on the search tags, and retrieve the feature image information of the plant to be sprayed at each growth stage.

[0075] A knowledge graph is constructed, and the feature image information of the plants to be sprayed at each growth stage is imported into the knowledge graph;

[0076] Obtain the image information of the plant to be identified, import the image information of the plant to be identified into the knowledge graph, and calculate the Euclidean distance value between the image information of the plant to be identified and each feature image information using the Euclidean distance algorithm;

[0077] Based on the Euclidean distance value, the pairing rate between the plant image information to be identified and each feature image information is determined, multiple pairing rates are obtained, and the multiple pairing rates are compared one by one with the preset pairing.

[0078] If at least one pairing rate is greater than the preset pairing rate, the plant to be identified is marked as a plant to be sprayed and a first identification result is generated; if multiple pairing rates are not greater than the preset pairing rate, the plant to be identified is marked as a plant that does not need to be sprayed and a second identification result is generated.

[0079] It should be noted that the spraying task information is pre-set by technicians and includes information such as the type of plant to be sprayed and the area to be sprayed. Feature image information includes leaf features, stem features, leaf texture, and leaf color. The Euclidean distance algorithm is a method for measuring the distance between two vectors, commonly used in multidimensional spaces. It can be used to measure the pairing rate of images in multidimensional space and is one of the most common distance metrics. This step can quickly identify whether the plants in the area to be sprayed are the target plants and can also identify the growth stage of the target plants.

[0080] If the identification result is the first identification result, then the image information of the plant to be sprayed is obtained, and the image information of the plant to be sprayed is subjected to redundancy reduction processing to obtain the redundancy-reduced image information of the plant to be sprayed, such as... Figure 2 As shown, it includes the following steps:

[0081] S202: If the recognition result is the first recognition result, then obtain the image information of the plant to be sprayed, and decompose the image information of the plant to be sprayed to obtain a diagonal matrix composed of rows and an orthogonal matrix composed of columns;

[0082] S204: Construct a two-dimensional coordinate system, where the X-axis represents a diagonal matrix and the Y-axis represents an orthogonal matrix; use the rows and columns of the diagonal and orthogonal matrices as the scales of the X and Y axes in the two-dimensional coordinate system, respectively.

[0083] S206: Import the diagonal matrix formed by rows and the orthogonal matrix formed by columns into the two-dimensional coordinate system to generate a point cloud data matrix of descriptors in the image of the plant to be sprayed, and generate a set of point cloud data coordinates of descriptors in the image of the plant to be sprayed based on the point cloud data matrix.

[0084] S208: Obtain the limit coordinate point set of the point cloud data coordinate set of the descriptor in the image of the plant to be sprayed, import the limit coordinate point set into the world coordinate system for recombination, and obtain the redundancy-reduced image information of the plant to be sprayed.

[0085] It should be noted that if the plant is identified as the target plant, i.e. the plant to be sprayed, the image information of the plant to be sprayed is captured again by the camera mounted on the spraying equipment. However, due to the influence of factors such as shooting angle, equipment accuracy and shooting environment, the captured image will have high redundancy, resulting in low image clarity, which will have a significant impact on the subsequent construction of the 3D model of the plant to be sprayed. Therefore, this step is to correct the captured image of the plant to be sprayed to reduce image redundancy, improve image clarity, and further improve the accuracy of subsequent model building.

[0086] The process of obtaining a three-dimensional model of the plant to be sprayed based on the reduced redundancy image information includes the following steps:

[0087] The redundancy-reduced image information of the plants to be sprayed is processed by feature extraction to obtain several paired points; the local outlier factor value of each paired point is calculated by the local outlier factor algorithm, and paired points with local outlier factor values ​​greater than the preset local outlier factor value are removed to obtain sparse paired points.

[0088] Randomly select any sparse pairing point as the origin of the coordinate system, establish a spatial coordinate system based on the origin, and obtain the coordinate information of all sparse pairing points in the spatial coordinate system; calculate the Euclidean distance between each sparse pairing point based on the coordinate information.

[0089] Based on the Euclidean distance between each sparse pairing point, the nearest neighbor of each sparse pairing point is determined, and the median coordinate point between each sparse pairing point and its corresponding nearest neighbor is obtained. The median coordinate point is then marked as a new pairing point.

[0090] The sparse pairing points are combined with new pairing points to obtain dense pairing points; the point cloud data corresponding to the dense pairing points is obtained, and the point cloud data corresponding to the dense pairing points is processed into a grid to obtain a three-dimensional model of the plant to be sprayed.

[0091] It should be noted that the redundant plant image information to be sprayed is processed by feature extraction using algorithms such as scale-invariant feature transformation and orientation rotation binarization to obtain several paired points. Outliers (noise points) are then filtered out using a local outlier factor, resulting in sparse paired points. Since the number of paired points obtained through feature extraction algorithms is limited, and some are outliers, the number of effective paired points is small, resulting in sparse paired points. If the 3D model of the plant to be sprayed is directly reconstructed using these sparse paired points, the completeness of the reconstructed 3D model is often low. Therefore, this step is necessary to obtain more new paired points, which are then combined with the sparse paired points to obtain dense paired points. Finally, the 3D model of the plant to be sprayed is reconstructed using point cloud reconstruction based on the dense paired points, thus obtaining a highly complete 3D model.

[0092] The process involves dividing the 3D model of the plant to be sprayed into several sub-spraying regions, obtaining characteristic parameters of the plant in each sub-spraying region, and acquiring the spraying start-point information of the spraying equipment. Based on the characteristic parameters, the spraying start-point information, and the 3D model of the plant to be sprayed, a preset spraying path is obtained, such as... Figure 3 As shown, it includes the following steps:

[0093] S302: Divide the three-dimensional model of the plant to be sprayed into several sub-spraying areas, and obtain the characteristic parameters of the plant to be sprayed in each sub-spraying area according to the three-dimensional model of the plant to be sprayed; wherein the characteristic parameters include leaf density, stem diameter, leaf shape, number of branches and degree of pest infestation;

[0094] S304: Initialize the operating parameters of the spraying equipment and obtain the spraying start information of the spraying equipment; input the spraying start information of the spraying equipment, the characteristic parameters of the plant to be sprayed, and the three-dimensional model of the plant to be sprayed into the particle swarm optimization algorithm;

[0095] S306: After iterative calculation using the particle swarm optimization algorithm, several spraying paths for the spraying equipment are obtained, and the distance values ​​corresponding to these spraying paths are acquired; a sequence table is constructed, and the distance values ​​corresponding to these spraying paths are imported into the sequence table and sorted by size.

[0096] S308: After sorting, extract the shortest path value and mark the spraying path corresponding to the shortest path value as the preset spraying path.

[0097] It should be noted that Particle Swarm Optimization (PSO) is a swarm intelligence-based optimization algorithm. In PSO, candidate solutions are called particles, and each particle moves in the search space, adjusting its direction and speed based on its individual optimal solution and the swarm's optimal solution to deduce the optimal path. The core idea of ​​the algorithm is to gradually search for potential optimal solutions by simulating information sharing and cooperation among individuals in the swarm. This step can automatically plan the preset spraying path of the spraying equipment based on the actual situation of the 3D model map corresponding to different plants. It can also plan the optimal spraying path in a targeted manner based on factors such as plant foliage density and pest density, so as to spray pesticides according to different plant growth conditions, more effectively reducing pesticide drift during spraying, thereby reducing environmental pollution and the impact on non-target areas.

[0098] The process involves acquiring real-time spraying environment parameters at various spraying time points, determining the real-time spraying drift amount based on these parameters, and adjusting the preset spraying path according to the real-time spraying drift amount. This includes the following steps:

[0099] The spray drift amount under various preset spraying environmental parameter combinations is obtained through big data network, a database is constructed, and the spray drift amount under various preset spraying environmental parameter combinations is imported into the database to obtain a characteristic database.

[0100] Real-time spraying environment parameters are obtained and imported into the feature database. The similarity between the real-time spraying environment parameters and various preset spraying environment parameter combinations is calculated using grey relational analysis to obtain multiple similarity scores.

[0101] Extract the maximum similarity from the multiple similarities, obtain the preset spraying environment parameter combination corresponding to the maximum similarity, and determine the real-time spraying drift of the spraying equipment under the current real-time spraying environment parameter conditions based on the preset spraying environment parameter combination corresponding to the maximum similarity.

[0102] The real-time spray drift amount is compared with the preset spray drift amount. If the real-time spray drift amount is greater than the preset spray drift amount, the position of the application axis point at the current spray time node is obtained according to the preset spray path.

[0103] The spraying correction amount is obtained based on the position of the application axis point and the real-time spraying drift amount, and the position of the application axis point is adjusted according to the spraying correction amount.

[0104] It should be noted that when pesticide solution is sprayed from the nozzle, the pesticide solution is circular, and the center point of the circle is the application axis. Spraying environmental parameters include temperature, humidity, light intensity, wind speed, and wind direction. Grey relational analysis is a multi-factor analysis method used to process multi-dimensional data. The basic idea of ​​grey relational analysis is to calculate the grey relational degree of incomplete data to find the relative degree of correlation between factors, thereby revealing the connections and trends between factors. This step allows for the calculation of pesticide drift based on real-time spraying environmental parameters, and timely adjustment of the spraying equipment parameters based on the pesticide drift. This can improve pesticide deposition in the target area, reduce pollution to non-target areas, protect the ecological environment, improve pesticide effectiveness, and reduce resource waste.

[0105] In addition, the method for preventing pesticide drift during pesticide spraying also includes the following steps:

[0106] After adjusting the position of the application axis point, the adjusted application axis point position information is obtained, and the adjusted application axis point position information is compared with the preset position information to obtain the deviation rate;

[0107] If the deviation rate is greater than the preset deviation rate, the real-time working parameter information of each sub-device in the spraying equipment is obtained, the hash value between the real-time working parameter information and the adjusted application axis position information is calculated by a hash algorithm, and the hash value is compared with the preset hash value.

[0108] Real-time operating parameter information with a hash value greater than a preset hash value is marked as abnormal operating parameters; a Bayesian network is constructed, and the abnormal operating parameters are imported into the Bayesian network for fault prediction to obtain the fault probability of the corresponding sub-device;

[0109] Sub-devices with a failure probability greater than a preset failure probability are marked as faulty devices, and the faulty devices are output.

[0110] It should be noted that if the pesticide drift remains excessive after adjusting the application axis position, it indicates a possible malfunction in a sub-equipment of the spraying equipment. For example, partial blockage of the nozzle can cause pesticide drift. This method can further deduce whether a sub-equipment of the spraying equipment has malfunctioned based on the application situation, thus avoiding the continuous application of malfunctioning equipment and preventing large-scale pesticide drift.

[0111] In addition, the method for preventing pesticide drift during pesticide spraying also includes the following steps:

[0112] If spray drift occurs during the spraying process, the current pesticide type information in the spraying equipment is obtained, and related text is generated based on the current pesticide type information;

[0113] Obtain the spray drift area corresponding to the occurrence of spray drift phenomenon, and obtain the geographical structure information of the spray drift area;

[0114] The correlation degree was obtained by performing a multi-factor regression analysis on the geographic structure information and related text.

[0115] The correlation degree is compared with a preset correlation degree; if the correlation degree is greater than the preset correlation degree, the spray drift area is marked as a contaminated area.

[0116] It obtains information on soil type and pesticide type in the polluted area, retrieves corresponding remediation plans for the polluted area through big data network based on the soil type and pesticide type information, and outputs the remediation plans for the polluted area.

[0117] It should be noted that if an area contains groundwater flow, underground springs, or underground wells, and if that area is a spray drift area, then that area will be marked as a contaminated area, and corresponding contaminated area remediation plans will be retrieved to promptly remediate the contaminated area and prevent further spread of pollution.

[0118] like Figure 4 As shown, a second aspect of the present invention discloses a pesticide spraying anti-drift system, the pesticide spraying anti-drift system including a memory 41 and a processor 42, the memory 41 storing an anti-drift method program, and when the anti-drift method program is executed by the processor 42, the following steps are implemented:

[0119] Obtain image information of the plant to be identified, identify the plant to be identified based on the image information, and obtain a first identification result or a second identification result;

[0120] If the recognition result is the first recognition result, then the image information of the plant to be sprayed is obtained, and the image information of the plant to be sprayed is subjected to redundancy reduction processing to obtain the redundancy-reduced image information of the plant to be sprayed; a three-dimensional model of the plant to be sprayed is obtained based on the redundancy-reduced image information of the plant to be sprayed.

[0121] The three-dimensional model of the plant to be sprayed is divided into several sub-spraying areas, and the characteristic parameters of the plant to be sprayed in each sub-spraying area are obtained, as well as the spraying starting point information of the spraying equipment. Based on the characteristic parameters, the spraying starting point information and the three-dimensional model of the plant to be sprayed, a preset spraying path is obtained.

[0122] The spraying device is controlled to spray the plants to be sprayed based on the preset spraying path; real-time spraying environment parameters are acquired at each spraying time point, the real-time spraying drift is determined based on the real-time spraying environment parameters, and the preset spraying path is corrected and adjusted based on the real-time spraying drift.

[0123] Furthermore, in a preferred embodiment of the present invention, acquiring image information of the plant to be identified, and identifying the plant to be identified based on the image information to obtain a first identification result or a second identification result, includes the following steps:

[0124] Obtain spraying task information, generate search tags based on the spraying task information, search the big data network based on the search tags, and retrieve the feature image information of the plant to be sprayed at each growth stage.

[0125] A knowledge graph is constructed, and the feature image information of the plants to be sprayed at each growth stage is imported into the knowledge graph;

[0126] Obtain the image information of the plant to be identified, import the image information of the plant to be identified into the knowledge graph, and calculate the Euclidean distance value between the image information of the plant to be identified and each feature image information using the Euclidean distance algorithm;

[0127] Based on the Euclidean distance value, the pairing rate between the plant image information to be identified and each feature image information is determined, multiple pairing rates are obtained, and the multiple pairing rates are compared one by one with the preset pairing.

[0128] If at least one pairing rate is greater than the preset pairing rate, the plant to be identified is marked as a plant to be sprayed and a first identification result is generated; if multiple pairing rates are not greater than the preset pairing rate, the plant to be identified is marked as a plant that does not need to be sprayed and a second identification result is generated.

[0129] Further, in a preferred embodiment of the present invention, the three-dimensional model of the plant to be sprayed is divided into several sub-spraying areas, and the characteristic parameters of the plant to be sprayed in each sub-spraying area are obtained, as well as the spraying start point information of the spraying equipment is obtained. A preset spraying path is obtained based on the characteristic parameters, the spraying start point information, and the three-dimensional model of the plant to be sprayed, including the following steps:

[0130] The three-dimensional model of the plant to be sprayed is divided into several sub-spraying areas, and the characteristic parameters of the plant to be sprayed in each sub-spraying area are obtained based on the three-dimensional model of the plant to be sprayed; wherein the characteristic parameters include leaf density, stem diameter, leaf shape, number of branches and degree of pest infestation;

[0131] Initialize the operating parameters of the spraying equipment and obtain the spraying start point information of the spraying equipment; input the spraying start point information of the spraying equipment, the characteristic parameters of the plant to be sprayed, and the three-dimensional model of the plant to be sprayed into the particle swarm optimization algorithm;

[0132] After iterative calculation using the particle swarm optimization algorithm, several spraying paths for the spraying equipment are obtained, and the distance values ​​corresponding to these spraying paths are acquired. A sequence list is constructed, and the distance values ​​corresponding to the spraying paths are imported into the sequence list and sorted by size.

[0133] After sorting, the shortest path value is extracted, and the spraying path corresponding to the shortest path value is marked as the preset spraying path.

[0134] Furthermore, in a preferred embodiment of the present invention, real-time spraying environment parameters are acquired at each spraying time point, real-time spraying drift is determined based on the real-time spraying environment parameters, and the preset spraying path is corrected and adjusted based on the real-time spraying drift, including the following steps:

[0135] The spray drift amount under various preset spraying environmental parameter combinations is obtained through big data network, a database is constructed, and the spray drift amount under various preset spraying environmental parameter combinations is imported into the database to obtain a characteristic database.

[0136] Real-time spraying environment parameters are obtained and imported into the feature database. The similarity between the real-time spraying environment parameters and various preset spraying environment parameter combinations is calculated using grey relational analysis to obtain multiple similarity scores.

[0137] Extract the maximum similarity from the multiple similarities, obtain the preset spraying environment parameter combination corresponding to the maximum similarity, and determine the real-time spraying drift of the spraying equipment under the current real-time spraying environment parameter conditions based on the preset spraying environment parameter combination corresponding to the maximum similarity.

[0138] The real-time spray drift amount is compared with the preset spray drift amount. If the real-time spray drift amount is greater than the preset spray drift amount, the position of the application axis point at the current spray time node is obtained according to the preset spray path.

[0139] The spraying correction amount is obtained based on the position of the application axis point and the real-time spraying drift amount, and the position of the application axis point is adjusted according to the spraying correction amount.

[0140] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for preventing pesticide drift during pesticide spraying, characterized in that, Includes the following steps: Obtain image information of the plant to be identified, identify the plant to be identified based on the image information, and obtain a first identification result or a second identification result; If the recognition result is the first recognition result, then the image information of the plant to be sprayed is obtained, and the image information of the plant to be sprayed is subjected to redundancy reduction processing to obtain the redundancy-reduced image information of the plant to be sprayed; a three-dimensional model of the plant to be sprayed is obtained based on the redundancy-reduced image information of the plant to be sprayed. The three-dimensional model of the plant to be sprayed is divided into several sub-spraying areas, and the characteristic parameters of the plant to be sprayed in each sub-spraying area are obtained, as well as the spraying starting point information of the spraying equipment. Based on the characteristic parameters, the spraying starting point information and the three-dimensional model of the plant to be sprayed, a preset spraying path is obtained. The spraying device is controlled to spray the plants to be sprayed based on the preset spraying path; real-time spraying environment parameters are acquired at each spraying time point, the real-time spraying drift is determined based on the real-time spraying environment parameters, and the preset spraying path is corrected and adjusted based on the real-time spraying drift. After adjusting the position of the application axis point, the adjusted application axis point position information is obtained, and the adjusted application axis point position information is compared with the preset position information to obtain the deviation rate. If the deviation rate is greater than the preset deviation rate, the real-time working parameter information of each sub-device in the spraying equipment is obtained, the hash value between the real-time working parameter information and the adjusted application axis position information is calculated by a hash algorithm, and the hash value is compared with the preset hash value. Real-time operating parameter information with a hash value greater than a preset hash value is marked as abnormal operating parameters; a Bayesian network is constructed, and the abnormal operating parameters are imported into the Bayesian network for fault prediction to obtain the fault probability of the corresponding sub-device; Sub-devices with a failure probability greater than a preset failure probability are marked as faulty devices, and the faulty devices are output. If spray drift occurs during the spraying process, the current pesticide type information in the spraying equipment is obtained, and related text is generated based on the current pesticide type information; Obtain the spray drift area corresponding to the occurrence of spray drift phenomenon, and obtain the geographical structure information of the spray drift area; The correlation degree was obtained by performing a multi-factor regression analysis on the geographic structure information and related text. Compare the correlation degree with the preset correlation degree; If the correlation degree is greater than the preset correlation degree, the spray drift area is marked as a contaminated area; It obtains information on soil type and pesticide type in the polluted area, retrieves corresponding remediation plans for the polluted area through big data network based on the soil type and pesticide type information, and outputs the remediation plans for the polluted area.

2. The method for preventing pesticide drift during pesticide spraying according to claim 1, characterized in that, Acquiring image information of the plant to be identified, and identifying the plant based on the image information to obtain a first identification result or a second identification result, includes the following steps: Obtain spraying task information, generate search tags based on the spraying task information, search the big data network based on the search tags, and retrieve the feature image information of the plant to be sprayed at each growth stage. A knowledge graph is constructed, and the feature image information of the plants to be sprayed at each growth stage is imported into the knowledge graph; Obtain the image information of the plant to be identified, import the image information of the plant to be identified into the knowledge graph, and calculate the Euclidean distance value between the image information of the plant to be identified and each feature image information using the Euclidean distance algorithm; Based on the Euclidean distance value, the pairing rate between the plant image information to be identified and each feature image information is determined, multiple pairing rates are obtained, and the multiple pairing rates are compared one by one with the preset pairing. If at least one pairing rate is greater than the preset pairing rate, the plant to be identified is marked as a plant to be sprayed and a first identification result is generated; if multiple pairing rates are not greater than the preset pairing rate, the plant to be identified is marked as a plant that does not need to be sprayed and a second identification result is generated.

3. The method for preventing pesticide drift during pesticide spraying according to claim 1, characterized in that, If the recognition result is the first recognition result, then the image information of the plant to be sprayed is obtained, and the image information of the plant to be sprayed is subjected to redundancy reduction processing to obtain the redundancy-reduced image information of the plant to be sprayed, including the following steps: If the identification result is the first identification result, then the image information of the plant to be sprayed is obtained, and the image information of the plant to be sprayed is decomposed to obtain a diagonal matrix composed of rows and an orthogonal matrix composed of columns; Construct a two-dimensional coordinate system, where the X-axis represents a diagonal matrix and the Y-axis represents an orthogonal matrix; use the rows and columns of the diagonal and orthogonal matrices as the scales of the X and Y axes in the two-dimensional coordinate system, respectively. The diagonal matrix formed by rows and the orthogonal matrix formed by columns are imported into the two-dimensional coordinate system to generate a point cloud data matrix of descriptors in the image of the plant to be sprayed, and a set of point cloud data coordinates of descriptors in the image of the plant to be sprayed is generated based on the point cloud data matrix. Obtain the limit coordinate point set of the point cloud data coordinate set of the descriptor in the image of the plant to be sprayed, import the limit coordinate point set into the world coordinate system for recombination, and obtain the redundancy-reduced image information of the plant to be sprayed.

4. The method for preventing pesticide drift during pesticide spraying according to claim 1, characterized in that, Based on the reduced redundancy image information of the plants to be sprayed, a three-dimensional model of the plants to be sprayed is obtained, including the following steps: The redundancy-reduced image information of the plants to be sprayed is processed by feature extraction to obtain several paired points; the local outlier factor value of each paired point is calculated by the local outlier factor algorithm, and paired points with local outlier factor values ​​greater than the preset local outlier factor value are removed to obtain sparse paired points. Randomly select any sparse pairing point as the origin of the coordinate system, establish a spatial coordinate system based on the origin, and obtain the coordinate information of all sparse pairing points in the spatial coordinate system; calculate the Euclidean distance between each sparse pairing point based on the coordinate information. Based on the Euclidean distance between each sparse pairing point, the nearest neighbor of each sparse pairing point is determined, and the median coordinate point between each sparse pairing point and its corresponding nearest neighbor is obtained. The median coordinate point is then marked as a new pairing point. The sparse pairing points are combined with new pairing points to obtain dense pairing points; the point cloud data corresponding to the dense pairing points is obtained, and the point cloud data corresponding to the dense pairing points is processed into a grid to obtain a three-dimensional model of the plant to be sprayed.

5. The method for preventing pesticide drift during pesticide spraying according to claim 1, characterized in that, The three-dimensional model of the plant to be sprayed is divided into several sub-spraying areas, and the characteristic parameters of the plant to be sprayed in each sub-spraying area are obtained, as well as the spraying start point information of the spraying equipment. Based on the characteristic parameters, the spraying start point information, and the three-dimensional model of the plant to be sprayed, a preset spraying path is obtained, including the following steps: The three-dimensional model of the plant to be sprayed is divided into several sub-spraying areas, and the characteristic parameters of the plant to be sprayed in each sub-spraying area are obtained based on the three-dimensional model of the plant to be sprayed; wherein the characteristic parameters include leaf density, stem diameter, leaf shape, number of branches and degree of pest infestation; Initialize the operating parameters of the spraying equipment and obtain the spraying start point information of the spraying equipment; input the spraying start point information of the spraying equipment, the characteristic parameters of the plant to be sprayed, and the three-dimensional model of the plant to be sprayed into the particle swarm optimization algorithm; After iterative calculation using the particle swarm optimization algorithm, several spraying paths for the spraying equipment are obtained, and the distance values ​​corresponding to these spraying paths are acquired. A sequence list is constructed, and the distance values ​​corresponding to the spraying paths are imported into the sequence list and sorted by size. After sorting, the shortest path value is extracted, and the spraying path corresponding to the shortest path value is marked as the preset spraying path.

6. The method for preventing pesticide drift during pesticide spraying according to claim 1, characterized in that, Real-time spraying environment parameters are acquired at each spraying time point. The real-time spraying drift is determined based on these parameters. The preset spraying path is then corrected and adjusted based on the real-time spraying drift, including the following steps: The spray drift amount under various preset spraying environmental parameter combinations is obtained through big data network, a database is constructed, and the spray drift amount under various preset spraying environmental parameter combinations is imported into the database to obtain a characteristic database. Real-time spraying environment parameters are obtained and imported into the feature database. The similarity between the real-time spraying environment parameters and various preset spraying environment parameter combinations is calculated using grey relational analysis to obtain multiple similarity scores. Extract the maximum similarity from the multiple similarities, obtain the preset spraying environment parameter combination corresponding to the maximum similarity, and determine the real-time spraying drift of the spraying equipment under the current real-time spraying environment parameter conditions based on the preset spraying environment parameter combination corresponding to the maximum similarity. The real-time spray drift amount is compared with the preset spray drift amount. If the real-time spray drift amount is greater than the preset spray drift amount, the position of the application axis point at the current spray time node is obtained according to the preset spray path. The spraying correction amount is obtained based on the position of the application axis point and the real-time spraying drift amount, and the position of the application axis point is adjusted according to the spraying correction amount.

7. A pesticide spraying drift prevention system, characterized in that, The anti-spray drift system includes a memory and a processor. The memory stores an anti-spray drift method program. When the anti-spray drift method program is executed by the processor, the anti-spray drift method steps as described in any one of claims 1 to 6 are implemented.

Images