Agricultural precise spraying method and system based on RFID positioning model

Abstract

The invention relates to the technical field of modern agriculture, in particular to an agricultural precise spraying method and system based on an RFID positioning model, and the method comprises the steps: S1, generating a prescription map with the prescription value of each layer of a pre-sprayed tree plant; s2, RFID tag cards are arranged on the pre-spraying trees and plants in the pre-spraying area; s3, an RSSI positioning model based on the RFID is constructed; s4, identifying the pre-sprayed tree plant by the spraying equipment; and S5, spraying the pre-sprayed tree plant layer by layer by spraying equipment. According to the tree plant spraying device, spraying can be accurately carried out on all layers of tree plants, excessive spraying and waste caused by a traditional spraying mode are effectively avoided, the spraying amount is reduced, meanwhile, the spraying efficiency is improved, the labor cost and the time cost are reduced, and therefore the cost in agricultural production is reduced to a certain degree on the whole.

Description

Technical Field

[0001] The present application relates to the field of modern agricultural technology, and in particular to a method and system for agricultural precision spraying based on an RFID positioning model. Background Art

[0002] Variable spraying combines spraying technology, sensor technology and automatic control technology to achieve "prescription" according to crop characteristic parameters during the plant protection process, which is the core technology to improve the utilization rate of pesticides. The key issue in achieving precise variable spraying is how to interpret the established spray prescription map in real time and accurately based on the existing positioning conditions. The absolute positioning technology such as the US Global Positioning System (GPS) and China's Beidou, which are widely used in agriculture, is easily affected by the canopy of trees in the orchard. The signal is unstable and it is difficult to meet the needs of precise spraying.

[0003] Radio Frequency Identification (RFID) is a positioning technology. Wireless signals are affected by the multipath effect during propagation, resulting in large fluctuations in signal strength RSSI. Direct use cannot meet the accuracy requirements of orchard positioning. In the actual spraying process, due to the movement of the sprayer and the complexity of the environment, there are multiple tag reflection problems, which can easily lead to misreading of other non-target canopy tag information.

[0004] The current RFID spraying method mainly uses handheld spraying, which cannot realize the autonomous operation of the sprayer. Given the different growth forms of trees and the differences in canopy width at different heights, the existing technology has problems such as insufficient positioning accuracy and delayed response of the control system, which may lead to excessive or insufficient pesticides, or even cause the spray device to be out of the canopy of the fruit tree, thus affecting the spraying effect and pesticide utilization rate. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] The main purpose of the present invention is to propose a method and system for agricultural precision spraying based on RFID positioning model to solve the above problems.

[0007] (II) Technical solution

[0008] To achieve the above object, the present invention provides a method for agricultural precision spraying based on an RFID positioning model, comprising the steps of:

[0009] S1, generate a prescription map with the prescription values ​​of each layer of pre-spray tree planting, including:

[0010] S11, obtaining a three-dimensional point cloud map of the pre-spraying area based on a multi-sensor scanning device;

[0011] S12, processing by a point cloud processing algorithm to obtain a single pre-spray tree value point cloud;

[0012] S13, stratifying the pre-sprayed tree planting according to the set height, obtaining information on the volume and canopy width of each layer, and obtaining the prescription value of each layer of a single tree planting through a spray unit model;

[0013] S14, number all pre-sprayed trees to form an ordered prescription map code;

[0014] S2, arranging RFID tag cards for the pre-spray trees in the pre-spray area, including: writing the number of the pre-spray trees, the prescription information of each layer and the volume of each layer into the RFID tag card; wherein each pre-spray tree is equipped with an RFID tag card, and the RFID tag cards are at the same level as the RFID reader;

[0015] S3, build an RSSI positioning model based on RFID, including:

[0016] S31, constructing a RSSI-distance relationship model;

[0017] S32, collection and analysis of RSSI data;

[0018] S33, RSSI data filtering and tag model fitting: Gaussian filtering and mean filtering are performed on the RSSI data obtained in step S32, and the RSSI logarithmic ranging model of n tags is fitted using the least squares method;

[0019] S34, the difference between the RSSI value predicted by the model and the actually measured RSSI value is measured through the loss function, and the gradient of the loss function is calculated to optimize the model parameters and obtain the final positioning model:

[0020]

[0021] Where, d is the distance between the reader and the tag; η C A is the attenuation factor of the RF signal based on the pre-spray area environment; C The strength information obtained by the reader when the distance between the reader and the tag is 1m based on the pre-spray area environment; RSSI (dBm) It is the strength RSSI information;

[0022] S4, the spraying device identifies the pre-spraying tree; during the spraying operation, the spraying device uses the RFID reader to obtain the tree corresponding to the RFID tag card with the smallest distance value as the target tree according to the RFID-based RSSI positioning model, and obtains the information in the RFID tag card;

[0023] S5, the spray equipment sprays the pre-spray tree planting layer according to the information in the RFID tag card of the target tree planting, based on the duty cycle sequence and crown width sequence information of each layer of the target tree planting, when the sensors of each layer corresponding to the spray equipment sense the tree crown, spray according to the corresponding duty cycle information, and when the sensors of each layer sense that they have left the tree crown, they enter delayed spraying until they stop and move to the next working position; wherein, the duration of the delayed spraying is calculated by the crown width sequence information.

[0024] Preferably, the step S13 specifically includes:

[0025] S131, single pre-spray tree value point cloud stratification:

[0026] The single pre-sprayed tree planting point cloud is divided into u layers according to height, and the volume of each layer is V u =(V1, V2, ..., V n ), the maximum crown width of each layer is S u =(S1, S2, ..., S n );

[0027] S132, spray volume calculation:

[0028]

[0029] Among them, Q i is the target spray volume of the i-th layer; q is the spray volume per unit time; S i is the target canopy width of the i-th layer; v is the forward speed of the spray system; V i is the canopy volume of the i-th layer; m a The spray volume required per unit canopy volume;

[0030] S133, based on the nozzle flow control relationship, calculate the PWM duty cycle under the prescription information

[0031] q=k flow ×D+b

[0032] Where q is the spray volume per unit time; D is the PWM duty cycle; k flow is the proportional factor of the nozzle flow expression; b is the intercept constant;

[0033]

[0034] D i is the PWM duty cycle of the i-th layer; V i is the volume of the i-th layer, S i is the maximum width of the i-th layer; k flow is the proportional factor of the nozzle flow expression; b is the intercept constant.

[0035] Preferably, the RSSI-distance relationship model in step S31 includes:

[0036]

[0037] Among them, RSSI is the received signal strength value; η is the attenuation factor of the radio frequency signal; d is the distance between the RFID reader and the RFID tag; A is the strength information obtained by the reader when the distance between the RFID reader and the RFID tag is 1m.

[0038] Preferably, the step S32 specifically includes: keeping the RFID reader and the RFID tag at the same horizontal height in the pre-spray area, setting multiple sampling points, collecting a sufficient number of RSSI values ​​within a GPIO trigger window time at the sampling points, and obtaining a set of original RSSI sequences:

[0039]

[0040] Among them, R is a frame of RSSI strength information set; R n is the RSSI value sequence of tag n; RSSI nt is the RSSI value of tag n received by the reader at time t within the GPIO trigger window.

[0041] Preferably, the step S33 includes:

[0042] Perform Gaussian filtering on the RSSI value data:

[0043]

[0044] Where f(x) is the Gaussian probability density function; F(x) is the Gaussian cumulative distribution function; μ and σ represent the original strength information of a set of RFID, R1, R2, ... R n The respective means and variances of ; x represents the intensity information;

[0045] The RSSI values ​​with confidence values ​​within the set interval are retained, and the rest are discarded;

[0046]

[0047] Among them, R filtered is the filtered RSSI strength information set; is the RSSI value of the tag after Gaussian filtering;

[0048] The intensity values ​​of each group after Gaussian filtering are processed by mean filtering to obtain:

[0049]

[0050] Where R processed For the collected data set, is the RSSI value sequence of tag n after processing, is the value of tag 1 at 1 meter after being processed by the original data processing algorithm. is the value of label n at 10 meters after being processed by the original data processing algorithm.

[0051] The n RSSI logarithmic ranging models corresponding to the n tags fitted by the least squares method are shown in the following formula:

[0052]

[0053] Where A1 represents the strength information obtained by the reader when the distance between the reader and tag 1 is 1m; η1 represents the attenuation factor of the RF signal fitted by tag 1; A n It represents the strength information obtained by the reader when the distance between the reader and tag N is 1m; η n Indicates the attenuation factor of the RF signal fitted by tag N.

[0054] Preferably, the loss function and the gradient of the loss function in step S34 are as shown in the following formula:

[0055]

[0056] Where J represents the loss function based on the orchard logarithmic ranging model; ▽J represents the gradient of the loss function; N represents the number of RFID tags; Indicates the RSSI strength value actually measured at the i-th sampling point; Represents the RSSI strength value of the i-th sampling point predicted by the model is the partial derivative of the loss function with respect to A; represents the partial derivative of the loss function with respect to η;

[0057] When ▽J reaches the most appropriate value, the final positioning model is determined.

[0058] Preferably, the delayed spraying time in step S5 is:

[0059]

[0060] Where: t i represents the spray delay time of the i-th layer, S i represents the spray area of ​​the i-th layer, v represents the moving speed of the spray device, and b represents the overlap rate.

[0061] The present invention also provides an agricultural precision spraying system based on an RFID positioning model, comprising a multi-sensor scanning device, a variable spraying device and a computer system; wherein:

[0062] The multi-sensor scanning device includes an IMU sensor, a 3D laser radar sensor, and a handheld bracket; the IMU sensor is fixed above the 3D laser radar sensor and installed together on the handheld bracket to form a handheld multi-sensor scanning device;

[0063] The variable spray device includes a variable spray execution unit, an information collection unit, an energy supply unit and a spray architecture;

[0064] The computer system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the control method for agricultural precision spraying based on the RFID positioning model as described in any one of the above items are implemented.

[0065] Preferably, the information collection unit includes: an RFID UHF reader / writer, an RFID tag, an infrared sensor array trigger group and an infrared sensor array control group;

[0066] The infrared sensor array trigger group includes a plurality of infrared sensors arranged vertically and spaced apart to determine the canopy layer to trigger and activate the RFID reader to determine the target tree planting and obtain prescription information;

[0067] The infrared sensor array control group includes n infrared sensors, wherein n is the number of set layers, which are used to detect the canopy layer triggering judgment to achieve stratified spraying.

[0068] Preferably, the variable spray execution unit comprises a nozzle group, which is provided with n nozzles, and the n nozzles are arranged vertically and spaced apart to spray each set layer, wherein n is the number of set layers;

[0069] The distance between the nozzles is d;

[0070]

[0071] Among them, H is the tree height, that is, the height of the target spraying area; L is the spray distance of the spray system, d is the nozzle spacing; β is the nozzle spray angle; h is the spray width of a single nozzle; r is the overlap rate of the spray areas of two adjacent nozzles, r∈[0.25, 0.3]; n is the number of set layers.

[0072] (III) Beneficial effects

[0073] The present invention configures RFID tags for pre-sprayed trees, and uses an RFID-based RSSI positioning model to accurately identify target trees and non-target trees. Target trees are layered according to height, and prescription values ​​for each layer are calculated based on the crown width through a spray unit model, and accurate layered spraying is performed based on the prescription values ​​for each layer. When the sensor senses that it has left the crown, delayed spraying begins, and the time for the delayed spray is calculated based on the crown width information; the edge areas of the fruit tree canopy are usually the most difficult places to spray because these areas may not be within the direct detection range of the sensor. The delayed stop stage ensures that these edge areas are also adequately sprayed by additionally extending the spray time.

[0074] The invention of the present application obtains a three-dimensional point cloud map of the pre-spraying area through a multi-sensor scanning device, thereby realizing rapid acquisition of relevant information such as the volume and crown width of the pre-sprayed trees.

[0075] The invention of the present application adopts an agricultural precision spraying system based on an RFID positioning model, and the variable spraying device is easy to deploy, easy to maintain, and has a fast response speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 A schematic diagram of a method for agricultural precision spraying based on an RFID positioning model provided in this embodiment;

[0077] Figure 2 Schematic diagram of the deployment of RFID tags for pre-spraying trees in the pre-spraying area in a method for agricultural precision spraying based on an RFID positioning model provided in this embodiment

[0078] Figure 3 A schematic diagram of the system operation of agricultural precision spraying based on the RFID positioning model provided in this embodiment

[0079] Figure 4 A schematic diagram of the interaction process of an RFID reader / writer of an agricultural precision spraying system based on an RFID positioning model provided in this embodiment;

[0080] Figure 5 A schematic diagram of the spray equipment identification and pre-spraying tree planting process of an agricultural precision spraying system based on an RFID positioning model provided in this embodiment;

[0081] Figure 6 A schematic diagram of the spraying process of the pre-spraying tree planting layer by the spraying equipment of the agricultural precision spraying system based on the RFID positioning model provided in this embodiment;

[0082] Figure 7 A schematic diagram of the system layered spray spacing of agricultural precision spraying based on the RFID positioning model provided in this embodiment;

[0083] Figure 8 A schematic diagram of a multi-sensor scanning device and a variable spray device of an agricultural precision spraying system based on an RFID positioning model provided in this embodiment.

[0084] [Description of Reference Numerals]

[0085] Figure 8 With marking instructions: 1: IMU sensor; 2: 3D laser radar sensor; 3: microcomputer; 4: handheld bracket; 5: PC; 6: six-way relay; 7: medicine box; 8: one-way valve; 9: diaphragm pump; 10: three-way overflow valve; 11: digital pressure transmitter; 12: vortex flowmeter; 13: nozzle group; 14: two-way solenoid valve group; 15: explosion-proof water pipe; 16: RFID UHF reader; 17: RFID tag; 18: infrared sensor array trigger group; 19: infrared sensor array control group; 20: STM32 controller; 21: aluminum profile frame; 22: crawler chassis; 23: dual-channel DC brushless motor driver. DETAILED DESCRIPTION

[0086] In order to better explain the present invention and facilitate understanding, the present invention is described in detail below through specific implementation modes in conjunction with the accompanying drawings.

[0087] It should be noted that all directional indications in the embodiments of the present invention (such as up, down, left, right, front, back, etc.) are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0088] In addition, in the present invention, descriptions such as "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0089] In the present invention, unless otherwise clearly specified and limited, the terms "connection", "fixation", etc. should be understood in a broad sense. For example, "fixation" can be a fixed connection, a detachable connection, or an integral connection; "connection" can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, and it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0090] like Figure 1 As shown, this embodiment provides a method for agricultural precision spraying based on an RFID positioning model, including steps S1-S5.

[0091] S1, generating a prescription map with prescription values ​​for each layer of pre-spray tree planting. Including, the prescription map refers to a chart for guiding the spraying of water, pesticides, fertilizers or other agricultural chemicals. This chart accurately divides the spray area and sets different spray amounts and spray methods for different farmland areas according to relevant factors such as the pre-spray tree planting volume to achieve precision agricultural management.

[0092] The step S1 specifically includes S11-S14:

[0093] S11, obtaining a three-dimensional point cloud map of the pre-spray area based on a multi-sensor scanning device.

[0094] S12, processing is performed through a point cloud processing algorithm to obtain a single pre-sprayed tree planting point cloud.

[0095] S13, stratify the pre-spray trees according to the set height. Obtain the information of the volume of each layer and the crown width of each layer, and obtain the prescription value of each layer of a single tree through the spray unit model. The location of the stratification can be selected according to the actual situation, such as stratifying the entire tree body exposed on the surface or stratifying the crown of the tree body. Specifically, in this embodiment, the stratified part is the stratification of the crown, and the number of layers can be selected according to the actual situation, such as two layers, three layers or even more than three layers. Specifically, in this embodiment, the number of layers is four layers. In reality, there are certain differences in the crown width of trees at different heights. Stratifying the pre-spray trees can more accurately obtain the spray volume required for the pre-spray trees; the spray unit model comprehensively considers the information of each layer of the pre-spray trees (including volume and crown width), the forward speed and the nozzle flow control relationship, and calculates the target spray volume and the corresponding PWM (pulse width modulation) duty cycle value;

[0096] S14, numbering all pre-sprayed trees to form an ordered prescription map coding. The prescription map coding method includes: Figure 2 As shown, the code num of the pre-spray tree prescription unit in the lower left corner of the prescription map is set to 1; the code gradually increases along the direction of the spray equipment operation, and the number increases with each additional pre-spray tree prescription unit; during the coding process, the spray equipment of the odd-numbered rows and the even-numbered rows move in opposite directions, that is, the odd-numbered rows are numbered from left to right, and the even-numbered rows are numbered from right to left. In other embodiments, the spray equipment of the odd-numbered rows and the even-numbered rows move in the same direction, and there is no difference in the coding direction of the odd-numbered rows and the even-numbered rows.

[0097] S2, deploying RFID tag cards for pre-spray trees in the pre-spray area, specifically including: writing the number of the pre-spray trees, the prescription information of each layer and the volume of each layer into the RFID tag card; wherein, each pre-spray tree is configured with an RFID tag card, and the horizontal heights of the RFID tag cards are the same. The RFID tag should have the characteristics of being waterproof, UV-resistant, and heat-resistant to adapt to the outdoor environment. The format of writing data into the RFID tag card is: writing the pre-spray tree planting code into the EPC (electronic product code) area of ​​the RFID tag memory, and changing the EPC number of the tag to the pre-spray tree planting code to achieve a one-to-one mapping between the RFID tag and the pre-spray tree planting; the prescription information and width information of each layer are written into the user area of ​​the tag memory according to a predetermined format. The specific example format is: "D 4H D 4L D 3H D 3L D 2H D 2L D 1H D 1L S 4H S 4L S 3H S 3L S 2H S 2L S 1H S 1L ”;

[0098] S3, build an RSSI positioning model based on RFID, including:

[0099] S31, constructing an RSSI-distance relationship model; the RSSI-distance relationship model refers to the relationship equation between the RSSI strength information obtained by the RFID reader and the distance d in a lightweight application scenario, as shown in the following formula:

[0100]

[0101] Among them, RSSI is the received signal strength value, dBm; η is the attenuation factor of the RF signal, which reflects the attenuation speed of the signal during the propagation process; d is the distance between the reader and the tag, m; A is the strength information obtained by the reader when the distance between the reader and the tag is 1m, dBm.

[0102] S32, collecting and analyzing RSSI data: collecting RSSI values ​​of a sufficient number of RFID tags at different sampling points.

[0103] S33, RSSI data filtering and tag model fitting: Gaussian filtering and mean filtering are performed on the RSSI data obtained in step S32, and the least squares method is used to fit the RSSI logarithmic ranging model of a sufficient number of tag RFIDs, thereby reducing noise interference and further improving positioning accuracy.

[0104] S34, the difference between the RSSI value predicted by the model and the actually measured RSSI value is measured through the loss function, and the gradient of the loss function is calculated to optimize the model parameters and obtain the final positioning model:

[0105]

[0106] Where d is the distance between the reader and the tag, m; η C is the attenuation factor of the RF signal based on the orchard environment, which is a fitting constant; A C The RSSI is the strength information obtained by the reader when the distance between the reader and the tag is 1m in the orchard environment. It is a fitting constant. (dBm) The RSSI information is in dBm.

[0107] The RFID-based RSSI positioning model can solve the problem of multi-tag reflection by comparing the tags with the closest distance to identify the target tree.

[0108] S4, the spraying device identifies the pre-spraying tree; during the spraying operation, the spraying device uses the RFID reader to obtain the tree corresponding to the RFID tag card with the smallest distance value as the target tree according to the RFID-based RSSI positioning model, and obtains the information in the RFID tag card. In this embodiment, the spraying device identifying the pre-spraying tree also includes obtaining a trigger reading signal, detecting the tree crown through the trigger sensor, and starting RFID for identification.

[0109] S5, the spray equipment sprays the pre-spray tree planting layer, according to the information in the RFID tag card of the target tree planting, based on the duty cycle sequence and crown width sequence information of each layer of the target tree planting, when the corresponding sensors of each layer set by the spray equipment sense the tree crown, spray according to the corresponding duty cycle information, and when the sensors of each layer sense that they have left the tree crown, they enter delayed spraying until they stop and move to the next working position; wherein, the duration of the delayed spraying is calculated by the crown width sequence information. The edge areas of the fruit tree canopy are usually the most difficult places to spray, because these areas may not be within the direct detection range of the sensor. The delayed stop stage ensures that these edge areas can also be fully sprayed by additionally extending the spray time. In other embodiments, when the sensors of each layer sense that they have left the tree crown, the spraying stops immediately.

[0110] Specifically, in this embodiment, step S13 specifically includes:

[0111] S131, single pre-spray tree value point cloud stratification:

[0112] The single pre-sprayed tree planting point cloud is divided into u layers according to height, and the volume of each layer is V u =(V1, V2, ..., V n ), the maximum crown width of each layer is S u =(S1, S2, ..., S n );

[0113] S132, spray volume calculation:

[0114]

[0115] Among them, Q i is the target spray volume of the i-th layer; q is the spray volume per unit time; S i is the target canopy width of the i-th layer; v is the forward speed of the spray system; V i is the canopy volume of the i-th layer; m a The spray volume required per unit canopy volume;

[0116] S133, when the nozzle structure, model and spray pressure are determined, the PWM duty cycle under the prescription information is calculated based on the nozzle flow control relationship

[0117] q=k flow ×D+b

[0118] Where, q is the spray volume per unit time, L / min; D is the PWM duty cycle, %; k flow is the proportional factor of the nozzle flow expression, L / min / %; b is the intercept constant;

[0119] By combining the spray volume calculation formula with the nozzle flow control relationship, when the pre-spray tree information (V and S) is obtained, the corresponding PWM duty cycle can be calculated:

[0120]

[0121] Among them, D i is the PWM duty cycle of the i-th layer; V i is the volume of the i-th layer, S i is the maximum width of the i-th layer; k flow is the proportional factor of the nozzle flow expression; b is the intercept constant;

[0122] In this embodiment, the calculation formula for the spraying amount of a single pre-spray tree is as follows:

[0123]

[0124] Among them, Q1, Q2, Q3 and Q4 are the target spray volumes of the first, second, third and fourth layers of a single fruit tree, respectively, L; S1, S2, S3 and S4 are the canopy widths of the first, second, third and fourth layers of a single fruit tree, respectively, m; V1, V2, V3 and V4 are the canopy volumes of the first, second, third and fourth layers of a single fruit tree, respectively, m 3 ;

[0125] The corresponding PWM duty cycle values ​​under the prescription information are as follows:

[0126]

[0127] Among them, D1, D2, D3 and D4 correspond to the PWM duty ratios of the first, second, third and fourth layers in a single fruit tree, %; m a is the spray volume required for unit canopy volume, L / m3; k is the proportional factor of the nozzle flow expression; b is the intercept constant of the nozzle flow expression.

[0128] Preferably, the RSSI-distance relationship model in step S31 includes:

[0129]

[0130] Among them, RSSI is the received signal strength value, dBm; η is the attenuation factor of the RF signal, which reflects the attenuation speed of the signal during the propagation process; d is the distance between the RFID reader and the RFID tag, m; A is the strength information dBm obtained by the reader when the distance between the RFID reader and the RFID tag is 1m.

[0131] Specifically, the step S32 specifically includes: keeping the RFID reader and the RFID tag at the same level in the pre-spray area to reduce signal transmission loss and interference caused by height differences, so that the collected RSSI values ​​are more comparable. Figure 2 As described above, setting multiple sampling points at different positions can achieve a more comprehensive understanding of the distribution of RFID signals in the pre-spray area, which is helpful for subsequent data analysis and processing. Specifically, the positions of the sampling points can be, for example, Figure 2 The position of the reference point of the positioning model; the distance interval between the sampling points can be set according to the actual situation, for example, 0.5m, 1m or 2m, etc. Specifically, in this embodiment, the distance interval between the sampling points is 0.5m; a sufficient number of RSSI values ​​are collected at the sampling points within a GPIO trigger window time, and the sufficient number of RSSI values ​​can be set to a certain number according to the actual situation, and a group of original RSSI sequences are obtained:

[0132]

[0133] Among them, R is a frame of RSSI strength information set; R n is the RSSI value sequence of tag n; RSSI nt is the RSSI value of tag n received by the reader at time t during the GPIO trigger window period. Specifically, in this embodiment, a GPIO trigger window time is 1 minute, the transmission power of the RFID reader is -33dBm, and the sampling distance range is 0.1 to 10m. Sampling within the range of 0.1m to 10m can ensure that the system can accurately measure and process RSSI values ​​at different distances. By properly selecting the transmission power and filtering method, RSSI fluctuations can be reduced and the measurement accuracy can be improved, thereby improving positioning accuracy.

[0134] As a preferred embodiment of the present invention, step S33 includes:

[0135] Perform Gaussian filtering on the RSSI value data:

[0136]

[0137] Where f(x) is the Gaussian probability density function; F(x) is the Gaussian cumulative distribution function; μ and σ represent the original strength information of a set of RFID, R1, R2, ... R n The respective means and variances of ; x represents the intensity information;

[0138] The RSSI values ​​with confidence values ​​within the set interval are retained, and the rest are discarded;

[0139]

[0140] Among them, R filtered is the filtered RSSI strength information set, dBm; is the RSSI value of the tag after Gaussian filtering, dBm; in this embodiment, the interval of the retained confidence value can be set according to the actual situation. Specifically, in this embodiment, the interval of the confidence value is [0.15, 0.83], and unreliable data is filtered to enhance positioning accuracy.

[0141] The intensity values ​​of each group after Gaussian filtering are processed by mean filtering to obtain:

[0142]

[0143] Where R processed For the collected data set, is the RSSI value sequence of tag n after processing, is the value of tag 1 at 1 meter after being processed by the original data processing algorithm. is the value of label n at 10 meters after being processed by the original data processing algorithm.

[0144] The n RSSI logarithmic ranging models corresponding to the n tags fitted by the least squares method are shown in the following formula:

[0145]

[0146] Where A1 represents the strength information obtained by the reader when the distance between the reader and tag 1 is 1m; η1 represents the attenuation factor of the RF signal fitted by tag 1; A n It represents the strength information obtained by the reader when the distance between the reader and tag N is 1m; η n Indicates the attenuation factor of the RF signal fitted by tag N.

[0147] Furthermore, in this embodiment, the loss function and the gradient of the loss function in step S34 are as shown in the following formula:

[0148]

[0149] Where J represents the loss function based on the orchard logarithmic ranging model; ▽J represents the gradient of the loss function; N represents the number of RFID tags; Indicates the RSSI strength value actually measured at the i-th sampling point; Represents the RSSI strength value of the i-th sampling point predicted by the model is the partial derivative of the loss function with respect to A; represents the partial derivative of the loss function with respect to η;

[0150] When ▽J reaches the most appropriate value, in this embodiment, when ▽J is closest to 0 (that is, the error is the smallest), the final positioning model is determined:

[0151]

[0152] Where d is the distance between the reader and the tag, m; η C is the attenuation factor of the RF signal based on the orchard environment, which is a fitting constant; A C The RSSI is the strength information obtained by the reader when the distance between the reader and the tag is 1m in the orchard environment. It is a fitting constant. (dBm) The RSSI information is in dBm.

[0153] Optionally, the delayed spray time in step S5 is:

[0154]

[0155] Where: t i represents the spray delay time of the i-th layer, S i represents the spray area of ​​the i-th layer, v represents the moving speed of the spray device, and b represents the overlap rate. Specifically, in this embodiment, the overlap rate of the sprayed area should be controlled between 0.25-0.3, and the overlap rate b=0.3 is set. The extended spray time of each layer is shown as follows:

[0156]

[0157] Among them, t1, t2, t3 and t4 are the spray delay times of the first, second, third and fourth layers respectively, s.

[0158] This embodiment also provides an agricultural precision spraying system based on an RFID positioning model, including a multi-sensor scanning device, a variable spraying device, and a computer system; wherein,

[0159] The multi-sensor scanning device includes an IMU sensor, a 3D laser radar sensor, and a handheld bracket; the IMU sensor is fixed above the 3D laser radar sensor and installed together on the handheld bracket to form a handheld multi-sensor scanning device; it is used to obtain basic information of pre-spray tree planting, perform pre-spray tree planting stratification and obtain prescription values.

[0160] The variable spray device includes a variable spray execution unit, an information collection unit, an energy supply unit and a spray structure; and is used for accurately spraying pre-sprayed tree planting.

[0161] The computer system is the computer system part of the operation process of the agricultural precision spraying system based on the RFID positioning model, and is not limited to a single computer entity. The computer system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the control method of agricultural precision spraying based on the RFID positioning model as described in any one of the above items are implemented.

[0162] Preferably, the information collection unit comprises: an RFID UHF reader / writer, an RFID tag, an infrared sensor array trigger group and an infrared sensor array control group;

[0163] The infrared sensor array trigger group is arranged relative to the infrared sensor array control group and is arranged in front of the infrared sensor array control group.

[0164] The infrared sensor array trigger group includes multiple infrared sensors, which are arranged vertically at intervals to determine the canopy and trigger the RFID reader to determine the target tree planting and obtain prescription information. The number of the infrared sensors can be set according to actual conditions, such as two, three or more. The spacing between each sensor is set according to actual conditions, without specific restrictions, as long as the canopy can be effectively detected.

[0165] The infrared sensor array control group includes n infrared sensors, wherein n is the number of set layers, which is used to detect the canopy layer triggering judgment to achieve layered spraying. In this embodiment, n is 4. In other embodiments, the number of layers can be set according to actual conditions.

[0166] Optionally, the variable spray execution unit includes a nozzle group, which is provided with n nozzles, and the n nozzles are arranged vertically and spaced apart to spray each set layer, wherein n is the number of set layers;

[0167] The spacing between the nozzles is d:

[0168]

[0169] Among them, H is the tree height, that is, the height of the target spraying area; L is the spray distance of the spray system; d is the nozzle spacing; β is the nozzle spray angle; h is the spray width of a single nozzle; r is the overlap rate of the spray areas of two adjacent nozzles, r∈[0.25, 0.3] ensures the effectiveness of edge spray while also ensuring the system's drug-saving situation, and n is the number of set layers.

[0170] Among them, the above d is obtained by combining the following formula.

[0171]

[0172] Among them, H is the tree height, that is, the height of the target spraying area, m; L is the spray distance of the spray system, m; d is the nozzle spacing, m; β is the nozzle spray angle, °; h is the spray width of a single nozzle, m; a is the width of the spray overlap area of ​​two adjacent nozzles, m; r is the overlap rate of the spray area of ​​two adjacent nozzles.

[0173] Preferably, the spray structure includes a frame, and a mobile device is arranged under the frame, and the mobile device is driven by a motor; the operator can remotely control the robot to move or navigate autonomously to realize the autonomous movement of the variable spray device in the pre-spraying area. The frame is constructed of aluminum profiles and can be customized according to different factors such as the structure, size and height of the pre-sprayed trees to ensure that the variable spray device can adapt to various orchard environments and provide the best operation effect; the mobile chassis adopts a crawler chassis, which is powered by a dual-channel brushless DC motor.

[0174] The following is a specific operation process to illustrate an agricultural precision spraying system based on an RFID positioning model in this embodiment.

[0175] like Figure 3 - Figure 8 As shown, a system for agricultural precision spraying based on an RFID positioning model includes a multi-sensor scanning device and a variable spraying device;

[0176] like Figure 8 As shown, the multi-sensor scanning device is mainly composed of an IMU sensor 1, a 3D laser radar sensor 2, an embedded computer 3, a handheld bracket 4 and a PC client 5. The IMU sensor 1 is fixed above the 3D laser radar sensor 2 and installed together on the handheld bracket 4 to form a handheld multi-sensor scanning device. The 3D laser radar sensor 2 and the IMU sensor 1 communicate with the embedded computer through Ethernet and RS232 serial ports respectively. The embedded computer receives sensor information and uses the WIFI signal network to display the collected point cloud information and IMU information in real time on the PC client 5. At the same time, control instructions can be sent to the embedded computer through the command window of the PC client 5.

[0177] The variable spray device is mainly composed of a variable spray execution unit, an information collection unit, an information processing unit and a spray architecture. The variable spray execution unit includes a six-way relay 6, a medicine box 7, a one-way valve 8, a diaphragm pump 9, a three-way overflow valve 10, a digital pressure transmitter 11, a vortex flowmeter 12, a nozzle group 13, a two-way solenoid valve group 14 and an explosion-proof water pipe 15; the information collection unit includes an RFID ultra-high frequency reader 16, an RFID tag 17, an infrared sensor array trigger group 18 and an infrared sensor array control group 19; the information processing unit includes an STM32 controller 20 and a host computer (PC); the spray architecture includes an aluminum profile frame 21, a crawler chassis 22, and a dual-channel DC brushless motor driver 23.

[0178] Furthermore, the infrared sensor array trigger group 18 is composed of three trigger infrared sensors, which are arranged vertically, with each sensor spaced about 10 cm apart, and the overall arrangement is perpendicular to the ground. This arrangement is intended to cover the wider crown area in the middle of the fruit tree canopy, and the three trigger sensors can most effectively detect the key positions of the canopy to ensure triggering efficiency.

[0179] Furthermore, the infrared sensor array control group 19 is composed of 4 control infrared sensors, the nozzle group 13 is composed of 4 conical nozzles; and the solenoid valve group is composed of 4 solenoid valves.

[0180] like Figure 7As shown, the fruit trees are divided into four areas according to their height, namely, V1, V2, V3 and V4. The spraying range V1 is controlled by the corresponding infrared sensor, and the corresponding electromagnetic valve and the corresponding nozzle spray the spray;

[0181] V2, V3, V4 and so on. A is a point on one side of the boundary of the V1 region, and A' is a point on the other side of the boundary of the V1 region; B is a point on one side of the boundary of the V2 region, and B' is a point on the other side of the boundary of the V2 region; C is a point on one side of the boundary of the V3 region, and C' is a point on the other side of the boundary of the V3 region; D is a point on one side of the boundary of the V4 region, and D' is a point on the other side of the boundary of the V4 region.

[0182] Each spray head, a solenoid valve and a control infrared sensor together form a spraying unit, the spray head is clipped on the solenoid valve, and the infrared sensor is fixed above the solenoid valve. The four spraying units are independent of each other and arranged vertically with a spacing of the spray head spacing d.

[0183] During the operation of the system, the variable spray device moves forward in the orchard at a preset speed, the infrared sensor array trigger group 18 detects the tree crown, triggers the RFID UHF reader 16 to collect the EPC information and RSSI intensity information in the RFID tag 17 in the canopy of the fruit tree, and sends it to the information processing unit. After receiving the tag information, the host computer 5 first uses a filtering algorithm to pre-process each group of original intensity information, excludes singular values, and obtains the RSSI intensity representative value of each tag in this frame; then substitutes each intensity value into the RSSI logarithmic ranging positioning model, calculates the distance between each tag and the RFID reader, and compares the distance value between each tag and the RFID reader. The fruit tree corresponding to the tag with the smallest distance value is determined as the target fruit tree, and the system obtains the EPC information of the target tag and the prescription information stored in the user area, and sends the information to the STM32 controller 20 through serial communication.

[0184] Specifically, the initialization content of the RFID reader includes defining the RFID reader client object, setting related macro definitions, and declaring the function interface required for the reader operation; by declaring and using four callback functions (i.e., start reading, stop reading, RFID reader GPIO trigger start, and RFID reader GPIO trigger stop), the reading and triggering control events of the RFID reader are realized. Specifically, the start reading callback function is used to start the reading function of the RFID reader, the stop reading callback function is used to terminate the reading operation of the reader, the GPI trigger start callback function is used to respond to the GPIO trigger signal to start the reading operation, and the GPI trigger stop callback function is used to respond to the GPIO trigger signal to terminate the reading operation. It should be noted that, if Figure 4 As shown, various control instructions of the RFID reader must be sent and received in sequence, otherwise the RFID reader cannot be correctly controlled.

[0185] When the RFID tag information returned by the callback function is received for the first time, the program creates a structure stack to manage the received information. The received RFID tag 17 information is saved in the buffer of the host computer and parsed, and the parsed data is pushed into the structure stack to form a complete frame of prescription map interpretation information. The prescription map interpretation information frame is dynamically managed through the pointer (Ptr). Among them, in order to ensure that a frame of prescription map interpretation information contains at least 30 intensity information (including no less than 10 target canopy information), a GPIO trigger window time is set to 400ms;

[0186] Optionally, the specific steps for interpreting information of a frame of prescription map are:

[0187] Prescription interpretation information call: call the structure stack through the Ptr pointer to extract all EPC values ​​and their corresponding RSSI strength information in the prescription map interpretation information of the frame; based on the extracted EPC information, classify all RSSI data in the frame interpretation information to form the original strength information group of each tag;

[0188] The original intensity information of each group is preprocessed using Gaussian filtering algorithm and mean filtering to obtain the RSSI intensity value represented by each tag. These intensity values ​​are substituted into the aforementioned RSSI logarithmic ranging positioning model to calculate the distance between each tag and the RFID reader. The distance values ​​between each tag and the RFID reader are compared, and the fruit tree corresponding to the tag with the smallest distance value is determined as the target fruit tree, and the EPC information of the tag is obtained.

[0189] Based on the EPC information of the target tag, quickly extract the information of the user area in the tag memory and send it to the lower computer in the corresponding format;

[0190] Use the pointer (Ptr) to clear all prescription map interpretation information in the structure stack to prepare for the next trigger window cycle;

[0191] Before the spray execution unit performs variable spray operation, it is necessary to start the diaphragm pump 9 first to extract liquid medicine from the medicine box 7. The water pressure regulation of the diaphragm pump 9 depends on the reading of the digital pressure transmitter 11, and the water pressure of the pipeline between the diaphragm pump 9 and the two-way solenoid valve group 14 is maintained by the control of the one-way valve 8 to ensure that the pressure requirements for the spray operation can be met. In addition, the components are connected by explosion-proof water pipes 15 to ensure the stable transmission of high-pressure water flow, avoid safety accidents caused by water pipe rupture or leakage, ensure the stability of fluid pressure during the spraying process, and ensure that the spraying effect is optimal.

[0192] When the STM32 controller 20 receives the information, it divides the information into a duty cycle sequence and a crown width sequence. The duty cycle sequence is used to control the spray flow rate; the crown width sequence is used to calculate the delay sequence, thereby extending the spray time. When each sensor in the infrared sensor array control group 19 detects the crown of the target fruit tree, the system generates a signal to start spraying and enters the precision spraying stage; the STM32 controller 20 determines which infrared sensor the signal comes from, activates the relay 6, and transmits the duty cycle signal to the two-way solenoid valve in charge of the corresponding area; the solenoid valve controls the opening and closing according to the received duty cycle signal, thereby adjusting the flow of the liquid medicine and starting the spray; and atomizes through the nozzle group 13 to form fine spray droplets, which are accurately sprayed to the target canopy; when each sensor of the infrared sensor array control group 19 leaves the corresponding regional canopy, a signal to extend the stop spraying is generated, and the system enters the delay stop stage; the STM32 determines the source of the detection signal, transmits the delay information to the preset timer, and the system continues to spray until the timing ends; the spraying stops, and the system's spray equipment sprays the pre-sprayed tree planting layer. The schematic diagram of the process is as follows Figure 6 .

[0193] After the sprayer has finished spraying one target fruit tree, it continues to move forward in the preset direction, and the system waits for the signal from the infrared sensor array trigger group 18 to start the spraying operation of the next target fruit tree.

[0194] The above are only specific application examples of the present invention and do not constitute any limitation on the protection scope of the present invention. In addition to the above embodiments, the present invention may also have other implementations. Any technical solution formed by equivalent replacement or equivalent transformation falls within the scope of protection claimed by the present invention.

Claims

1. A method for agricultural precision spraying based on RFID positioning model, characterized in that: Includes steps: S1, generate a prescription map with the prescription values ​​of each layer of pre-spray tree planting, including: S11, obtaining a three-dimensional point cloud map of the pre-spraying area based on a multi-sensor scanning device; S12, processing by a point cloud processing algorithm to obtain a single pre-sprayed tree planting point cloud; S13, stratifying the pre-sprayed tree planting according to the set height, obtaining information on the volume and canopy width of each layer, and obtaining the prescription value of each layer of a single tree planting through a spray unit model; S14, number all pre-sprayed trees to form an ordered prescription map code; S2, arranging RFID tag cards for the pre-spray trees in the pre-spray area, including: writing the number of the pre-spray trees, the prescription information of each layer and the volume of each layer into the RFID tag card; wherein each pre-spray tree is configured with one RFID tag card, and the RFID tag cards have the same horizontal height; S3, build an RSSI positioning model based on RFID, including: S31, constructing a RSSI-distance relationship model; S32, collection and analysis of RSSI data; S33, RSSI data filtering and tag model fitting: Gaussian filtering and mean filtering are performed on the RSSI data obtained in step S32, and the RSSI logarithmic ranging model of n tags is fitted using the least squares method; S34, the difference between the RSSI value predicted by the model and the actually measured RSSI value is measured through the loss function, and the gradient of the loss function is calculated to optimize the model parameters and obtain the final positioning model: Where, d is the distance between the reader and the tag; η C A is the attenuation factor of the RF signal based on the pre-spray area environment; C The strength information obtained by the reader when the distance between the reader and the tag is 1m based on the pre-spray area environment; RSSI (dBm) It is the strength RSSI information; S4, the spraying device identifies the pre-spraying tree; during the spraying operation, the spraying device uses the RFID reader to obtain the tree corresponding to the RFID tag card with the smallest distance value as the target tree according to the RFID-based RSSI positioning model, and obtains the information in the RFID tag card; S5, the spray equipment sprays the pre-spray tree planting layer; according to the information in the RFID tag card of the target tree planting, based on the duty cycle sequence and crown width sequence information of each layer of the target tree planting, when the sensors of each layer corresponding to the spray equipment sense the tree crown, spray according to the corresponding duty cycle information, and when the sensors of each layer sense that they have left the tree crown, they enter delayed spraying until they stop and move to the next working position; wherein, the duration of the delayed spraying is calculated by the crown width sequence information.

2. The method for agricultural precision spraying based on RFID positioning model according to claim 1, characterized in that: The step S13 specifically includes: S131, single pre-spray tree value point cloud stratification: The single pre-sprayed tree planting point cloud is divided into u layers according to height, and the volume of each layer is V u =(V1, V2, ..., V n ), the maximum crown width of each layer is S u =(S1, S2, ..., S n ). S132, spray volume calculation: Among them, Q i is the target spray volume of the i-th layer; q is the spray volume per unit time; S i is the target canopy width of the i-th layer; v is the forward speed of the spray system; V i is the canopy volume of the i-th layer; m a The spray volume required per unit canopy volume; S133, based on the nozzle flow control relationship, calculate the PWM duty cycle under the prescription information q=k flow ×D+b Where q is the spray volume per unit time; D is the PWM duty cycle; k flow is the proportional factor of the nozzle flow expression; b is the intercept constant; Among them, D i is the PWM duty cycle of the i-th layer; V i is the volume of the i-th layer, S i is the maximum width of the i-th layer; k flow is the proportional factor of the nozzle flow expression; b is the intercept constant.

3. The method for agricultural precision spraying based on RFID positioning model according to claim 1, characterized in that: The RSSI-distance relationship model in step S31 includes: Among them, RSSI is the received signal strength value; η is the attenuation factor of the radio frequency signal; d is the distance between the RFID reader and the RFID tag; A is the strength information obtained by the reader when the distance between the RFID reader and the RFID tag is 1m.

4. The method for agricultural precision spraying based on RFID positioning model according to claim 3, characterized in that: The step S32 specifically includes: keeping the RFID reader and the RFID tag at the same level in the pre-spraying area, setting multiple sampling points, collecting a sufficient number of RSSI values ​​within a GPIO trigger window time at the sampling points, and obtaining a set of original RSSI sequences: Among them, R is a frame of RSSI strength information set; R n is the RSSI value sequence of tag n; RSSI nt is the RSSI value of tag n received by the reader at time t within the GPIO trigger window.

5. The method for agricultural precision spraying based on RFID positioning model according to claim 4, characterized in that: The step S33 comprises: Perform Gaussian filtering on the RSSI value data: Where f(x) is the Gaussian probability density function; F(x) is the Gaussian cumulative distribution function; μ and σ represent the original strength information of a set of RFID, R1, R2, ... R n The respective means and variances of ; x represents the intensity information; The RSSI values ​​with confidence values ​​within the set interval are retained, and the rest are discarded; Among them, R filtered is the filtered RSSI strength information set; is the RSSI value of the tag after Gaussian filtering; The intensity values ​​of each group after Gaussian filtering are processed by mean filtering to obtain: Where R processed For the collected data set, is the RSSI value sequence of tag n after processing, is the value of tag 1 at 1 meter after being processed by the original data processing algorithm. is the value of label n at 10 meters after being processed by the original data processing algorithm. The n RSSI logarithmic ranging models corresponding to the n tags fitted by the least squares method are shown in the following formula: Where A1 represents the strength information obtained by the reader when the distance between the reader and tag 1 is 1m; η1 represents the attenuation factor of the RF signal fitted by tag 1; A n It represents the strength information obtained by the reader when the distance between the reader and tag N is 1m; η n Indicates the attenuation factor of the RF signal fitted by tag N.

6. The method for agricultural precision spraying based on RFID positioning model according to claim 5, characterized in that: The loss function and the gradient of the loss function in step S34 are as shown in the following formula: Where J represents the loss function based on the orchard logarithmic ranging model; ▽J represents the gradient of the loss function; N represents the number of RFID tags; Indicates the RSSI strength value actually measured at the i-th sampling point; Represents the RSSI strength value of the i-th sampling point predicted by the model is the partial derivative of the loss function with respect to A; represents the partial derivative of the loss function with respect to η; When ▽J reaches the most appropriate value, the final positioning model is determined.

7. The method for agricultural precision spraying based on RFID positioning model according to claim 1, characterized in that: The delayed spraying time in step S5 is: Where: t i represents the spray delay time of the i-th layer, S i represents the spray area of ​​the i-th layer, v represents the speed of the spray device, and b represents the overlap rate.

8. An agricultural precision spraying system based on RFID positioning model, characterized in that: It includes a multi-sensor scanning device, a variable spray device and a computer system; wherein, The multi-sensor scanning device includes an IMU sensor, a 3D laser radar sensor, and a handheld bracket; the IMU sensor is fixed above the 3D laser radar sensor and installed together on the handheld bracket to form a handheld multi-sensor scanning device; The variable spray device includes a variable spray execution unit, an information collection unit, an energy supply unit and a spray architecture; The computer system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the control method for agricultural precision spraying of the RFID positioning model as described in any one of claims 1 to 7 are implemented.

9. The agricultural precision spraying system based on RFID positioning model according to claim 8, characterized in that: The information collection unit includes: an RFID UHF reader / writer, an RFID tag, an infrared sensor array trigger group and an infrared sensor array control group; The infrared sensor array trigger group includes a plurality of infrared sensors arranged vertically and spaced apart to determine the canopy layer to trigger and activate the RFID reader to determine the target tree planting and obtain prescription information; The infrared sensor array control group includes n infrared sensors, wherein n is the number of set layers, which are used to detect the canopy layer triggering judgment to achieve stratified spraying.

10. The agricultural precision spraying system based on RFID positioning model according to claim 9, characterized in that: The variable spray execution unit includes a nozzle group, which is provided with n nozzles, and the n nozzles are arranged vertically and spaced apart to spray each set layer, wherein n is the number of set layers; The distance between the nozzles is d; Among them, H is the tree height, that is, the height of the target spraying area; L is the spray distance of the spray system, d is the nozzle spacing; β is the nozzle spray angle; h is the spray width of a single nozzle; r is the overlap rate of the spray areas of two adjacent nozzles, r∈[0.25, 0.3]; n is the number of set layers.

Images