Remote monitoring and analysis method for agricultural planting environment data

Abstract

The invention belongs to the technical field of agricultural planting, and discloses a remote monitoring and analyzing method for agricultural planting environment data. Comprising the following steps: S1, acquiring environment data of unit areas in each farmland through acquisition equipment; the method comprises the following steps: planning a spraying path, dynamically dividing an area into different grades, dividing operation blocks, automatically generating an anti-drift and high-efficiency flight path, starting a dual-coverage strategy for advanced insect pest blocks, ensuring the control effect of a core area, and in a spraying decision-making link, setting a pesticide spraying amount prediction model so as to ensure the control effect of the core area. According to the method, basic elements such as the area of a to-be-treated area and the insect pest level are considered, variables such as crop canopy density, real-time wind speed and temperature and the insect pest resistance risk level calculated through historical pesticide application records are integrated in real time, and compared with a traditional fixed dose mode, the method can effectively improve the control efficiency on the premise of guaranteeing the control effect, and the control cost is reduced. And waste and pollution caused by excessive use of pesticides are obviously reduced.

Description

Technical Field

[0001] This invention relates to the field of agricultural planting technology, and more specifically, to a method for remote monitoring and analysis of agricultural planting environment data. Background Technology

[0002] As modern agriculture develops towards precision and intelligence, using information technology to remotely monitor and analyze the planting environment has become a key means to improve agricultural production efficiency and resource utilization efficiency. Traditional agricultural environmental monitoring systems are usually limited to the collection and remote display of basic data such as soil temperature and humidity and meteorological parameters. The depth of data utilization is insufficient and it is difficult to directly support production decisions.

[0003] Currently, in crop health assessment, existing technologies mostly rely on single or a few environmental indicators, lacking a multi-dimensional and comprehensive evaluation of crop physiological status. In pest and disease early warning, existing solutions primarily rely on visual detection based on image recognition, which is only effective after pest symptoms appear, resulting in delayed warnings. Other solutions are based on statistical prediction models using environmental data, but these often directly link meteorological factors to pest occurrence, neglecting the intrinsic regulatory role of crop health on pest occurrence and development, leading to insufficient universality and foresight in the prediction models. Regarding precision operation execution, drone spraying technology is widely used, but path planning is often based on fixed plot shapes or simple area markings, failing to deeply integrate with the spatial distribution of pests. Spraying decisions are often based on experience or fixed standards, unable to adjust in real-time according to dynamically changing crop canopy structure, real-time weather conditions, and pest resistance levels, resulting in low pesticide utilization and unstable control effects. Summary of the Invention

[0004] To address the problems in the background art, this invention proposes a method for remote monitoring and analysis of agricultural planting environment data.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for remote monitoring and analysis of agricultural planting environment data, comprising the following steps: S1: Collect environmental data for each unit area within the farmland using data acquisition equipment, and normalize the environmental data. S2: Generate growth scores for farmland unit areas based on the normalized data, and derive the growth score change rate of unit areas based on multiple growth scores; S3: Obtain historical pest quantity data and build a pest quantity prediction model. Use the pest quantity prediction model to predict the pest quantity in the unit area and classify the unit area according to the pest quantity in the unit area. S4: Based on the unit area classification results, plan the spraying path for the drone to spray pesticides. The drone sprays pesticides along the spraying path for the intermediate pest unit area and the advanced pest unit area. S5: Obtain historical pesticide spraying data and construct a pesticide spraying prediction model. Use the pesticide spraying prediction model to predict the pesticide spraying amount required for all intermediate-level pest unit areas and advanced-level pest unit areas. S6: Determine the spraying time based on the predicted pesticide spraying amount, and decide on the drone's spraying strategy based on the spraying time, flight time, and the amount of pesticide stored in the drone.

[0006] Furthermore, specific crops are planted in the agricultural planting area, which is divided into multiple farmlands. Each farmland is evenly divided into multiple unit areas. The environmental data includes soil moisture, soil temperature, average crop height, soil conductivity, and soil organic matter content. Soil moisture is measured within a unit area using a soil moisture sensor employing either time-domain reflectometry or frequency-domain reflectometry. Data is collected at regular intervals, and the daily average value is calculated. The soil moisture content is then normalized. The optimal moisture range for soil for crops is — The soil's tolerance range for crop moisture is — , This represents the measured soil moisture. when < < Ms= ; when < < Ms=1; when < < Ms= ; when < or > Ms=0; In the formula, Ms is the normalized soil moisture, and a and b are calibration coefficients obtained through training with historical laboratory data; Soil temperature within a unit area is measured using a thermistor or thermocouple soil temperature sensor. Data is collected at regular intervals, and the daily average value is calculated. The soil temperature is then normalized. The optimal temperature range for soil for crops is — The soil's temperature tolerance range for crops is — , This represents the measured soil temperature. when < < T= ; when < < T=1; when < < T= ; when < or > T=0; In the formula, T is the normalized soil temperature, and c and d are calibration coefficients, which are obtained through training with historical laboratory data. The height of crops within a unit area is measured using LiDAR or a binocular stereo vision camera. The average height of crops is then calculated by averaging the values, and the average height is normalized.

[0007] In the formula, To normalize the average height of crops, This represents the measured average height of crops. This represents the theoretical maximum height of the crop at this growth stage. The soil conductivity was obtained by measuring four points within a unit area using a four-electrode EC sensor and taking the average value. The soil conductivity was then normalized. The optimal electrical conductivity range of soil for crops is: — The soil's tolerance range for agricultural electrical conductivity is [range missing]. — , This represents the measured soil conductivity. when < < , = ; when < < , =1; when < < , = ; when < or > , =0.1; In the formula, The normalized soil conductivity is represented by e and f, which are calibration coefficients obtained through training with historical laboratory data. The soil organic matter content within a unit area was measured using a near-infrared spectroscopy soil sensor, and the soil organic matter content was then normalized. The optimal range of organic matter content in soil for crops is as follows: — The soil's tolerance range for crop organic matter content is... — , This represents the measured soil organic matter content. when < < , = ; when < < , =1; when < < , = ; when < or > , =0.2; In the formula, To normalize soil organic matter content, These are calibration coefficients, obtained through training with historical laboratory data.

[0008] Furthermore, the process of generating growth scores for farmland unit areas based on the normalized data includes: Growth score S:

[0009] In the formula, , , , and These are weighting coefficients, obtained through training based on historical data.

[0010] Furthermore, the process of deriving the growth score change rate of a unit region based on multiple growth scores includes: A growth scoring evaluation cycle is set. At each growth scoring evaluation cycle, the corresponding growth score of each unit area is obtained. The growth scores of the previous 2n growth scoring evaluation cycles are taken for each unit area. The 2n growth scores are arranged in chronological order. The average of the first n growth scores is calculated to obtain the first growth score average. The average of the last n growth scores is calculated to obtain the second growth score average. The second growth score average is subtracted from the first growth score average to obtain the score difference. The score difference is divided by the first growth score average to obtain the growth score change rate.

[0011] Furthermore, the process of acquiring historical pest quantity data and constructing a pest quantity prediction model, and then using this model to predict the pest quantity within a unit area, includes: The number of pests refers to the number of pests within a unit area. Factors influencing the number of pests within a unit area include: real-time growth score, growth score change rate, microbial diversity index, pesticide residue index, number of natural enemies of pests, and historical pest population. Real-time growth score refers to the latest growth score of the unit region; Microbial biomass based on PLFA analysis yields a microbial diversity index; The pesticide residue index is calculated by the pesticide degradation rate and application time. The number of natural enemies of pests captured per unit time using sex pheromones and sticky traps; The historical pest baseline was obtained per unit time through sex pheromone traps and yellow sticky traps; The historical pest quantity data of a single unit area is obtained. The historical pest quantity data includes the real-time growth score, growth score change rate, microbial diversity index, pesticide residue index, number of natural enemies of pests, historical pest base, and historical pest quantity of the single unit area. Based on the real-time growth score, growth score change rate, microbial diversity index, pesticide residue index, number of natural enemies of pests, historical pest base and corresponding historical pest number in the corresponding unit area of ​​different historical pest quantity data, a pest quantity prediction set is generated and divided into the first training set and the first test set. Construct a first convolutional neural network, using real-time growth score, growth score change rate, microbial diversity index, pesticide residue index, number of natural enemies of pests and historical pest base from different historical pest quantity data in the first training set as the input data of the first convolutional neural network, and using the corresponding historical pest quantity in the first training set as the output data of the first convolutional neural network. The first convolutional neural network is trained to obtain the first initial convolutional neural network. The first initial convolutional neural network is validated using the first test set. The first initial convolutional neural network, whose output is less than or equal to the preset first test error threshold, is used as the pest quantity prediction model. The real-time growth score, growth score change rate, microbial diversity index, pesticide residue index, number of natural enemies of pests, and historical pest population of each unit area are input into the pest population prediction model to obtain the predicted pest population of each unit area.

[0012] Furthermore, the process of classifying unit areas based on the number of pests within each unit area includes: The predicted number of pests in each unit area is obtained based on the pest quantity prediction model. An appropriate first pest quantity threshold and second pest quantity threshold are set based on historical pest quantity data. The historical pest quantity data refers to the data set of pest quantities in previous unit areas. The number of pests in each unit area is compared with the two pest quantity thresholds. If the number of pests in a unit area is less than the first pest number threshold, the unit area is judged as a low-level pest unit area and no intervention is required for this unit area. When the number of the first pest is less than the number of pests in the unit area, which is less than the number of the second pest, the unit area is determined to be a medium-level pest unit area, and pesticide spraying by drone is required. If the number of pests in a unit area exceeds the threshold for the second type of pest, the unit area is identified as a high-level pest unit area and requires pesticide spraying by drone.

[0013] Furthermore, based on the unit area classification results, the spraying path for drone pesticide spraying is planned. The process of drones spraying pesticides along the spraying path on intermediate-level and advanced-level pest unit areas includes: Based on the pest level classification results of the unit area, the drone spraying path planning is performed. The unit areas marked as medium and high pests are spatially clustered. The clustering radius is 1.5 times the drone spraying width. Spatially adjacent unit areas with the same or similar pest levels are merged into continuous operation blocks. The start and end points of the drone are marked. Starting from the start point, the nearest block is iteratively selected to be added to the path until all blocks are covered. Finally, the path is connected to the end point to form a complete path, thus obtaining the drone pesticide spraying path. For a single operational block, the optimal flight direction is determined based on the real-time wind direction. A flight direction perpendicular to the wind direction is preferred to reduce droplet drift. When the wind speed exceeds 5 m / s, the flight direction is switched to tailwind. A reciprocating sweeping path is used to achieve full coverage within the block. The track interval N = 0.8B is calculated based on the effective spray width B of the UAV at the set flight altitude to ensure a 20% overlap between adjacent tracks. For blocks with advanced pest units, a dual coverage strategy is implemented. After the first regular sweep, a second enhanced spray is carried out in an orthogonal direction targeting the core pest sub-area.

[0014] Furthermore, the process of acquiring historical pesticide spraying data and constructing a pesticide spraying prediction model, and then using this model to predict the required pesticide spraying amounts for all intermediate and advanced pest unit areas, includes: The amount of pesticide sprayed refers to the amount of pesticide required for the unit area to be sprayed, and the unit area to be sprayed refers to all intermediate pest unit areas and advanced pest unit areas. Factors affecting the amount of pesticides required for all intermediate and advanced pest units include: the number of intermediate pest units, the number of advanced pest units, real-time wind speed, spray loss, crop canopy density, and real-time temperature. Real-time wind speed is monitored in real time using a wind speed sensor; The spraying loss is calculated based on equipment parameters and experience. Crop canopy density is calculated based on canopy porosity from lidar point clouds; Real-time temperature is monitored by a temperature sensor. Obtain historical pesticide spraying data for the unit area to be sprayed. The historical pesticide spraying data includes the number of intermediate pest units, the number of advanced pest units, real-time wind speed, spraying loss, crop canopy density, real-time temperature, and the historical pesticide spraying amount for that single unit area. Based on the number of intermediate pest units, the number of advanced pest units, real-time wind speed, spray loss, crop canopy density, real-time temperature and corresponding historical pesticide spraying data of different historical pesticide spraying amounts, a pesticide spraying amount prediction set is generated and divided into a second training set and a second test set. A second convolutional neural network is constructed. The number of intermediate pest units, the number of advanced pest units, real-time wind speed, spray loss, crop canopy density, and real-time temperature in the different historical pesticide spraying data in the second training set are used as the input data of the second convolutional neural network, and the corresponding historical pesticide spraying amount in the second training set is used as the output data of the second convolutional neural network. The second convolutional neural network is trained to obtain the second initial convolutional neural network. The second initial convolutional neural network is validated using the second test set. The second initial convolutional neural network that outputs a second test error threshold less than or equal to the preset second test error threshold is used as the pesticide spraying amount prediction model. The number of intermediate pest units, the number of advanced pest units, real-time wind speed, spray loss, crop canopy density, and real-time temperature of the area to be sprayed are input into the pesticide spraying amount prediction model to obtain the predicted pesticide spraying amount for the area to be sprayed.

[0015] Furthermore, the process of determining the spraying time based on the predicted pesticide spraying volume, and then deciding on the drone's spraying strategy based on the spraying time, flight time, and the drone's pesticide storage capacity, includes: The predicted pesticide spraying amount is obtained based on the pesticide spraying amount prediction model. The optimal pesticide spraying speed of the drone is set according to the real-time wind speed. The spraying time is obtained by dividing the predicted pesticide spraying amount by the optimal pesticide spraying speed. The remaining flight time is set according to the flight speed of the drone. The remaining flight time refers to the flight time of the drone when it is not spraying. If the predicted pesticide spraying amount is less than the pesticide storage capacity of the drone, and the spraying time plus the remaining flight time is less than the endurance time, then the drone will spray according to the predetermined spraying path. If the predicted pesticide spraying amount is less than the pesticide storage capacity of the drone, and the spraying time plus the remaining flight time is greater than the endurance time, the drone will spray according to the predetermined spraying path. If the drone's battery level drops below the alarm value midway, it needs to return to recharge and then return to spray according to the spraying path again. If the number of times it returns to recharge is greater than the set number, a specific number of drones need to be dispatched to assist in spraying before spraying, and the spraying path of each drone needs to be replanned. The specific number is related to the number of times it returns to recharge, so that the number of times each drone returns to recharge is less than the set number. If the predicted pesticide spraying amount is greater than the pesticide storage capacity of the drone, and the spraying time plus the remaining flight time is less than the endurance time, the drone will spray according to the predetermined spraying path. The drone needs to return midway to replenish the pesticide, and after replenishment, it will return again to spray according to the spraying path. If the predicted pesticide spraying volume exceeds the drone's pesticide storage capacity, and the spraying time plus the remaining flight time exceeds the endurance time, the drone will spray according to the predetermined spraying path. The drone needs to return midway to replenish pesticides, and after replenishment, it will return again to spray according to the spraying path. If the drone's battery level drops below the alarm value midway, it needs to return to recharge and then return again to spray according to the spraying path. If the number of times it returns to recharge exceeds the set number, a specific number of drones needs to be dispatched to assist in spraying before spraying, and the spraying path of each drone needs to be replanned. The specific number is related to the number of times it returns to recharge, so that the number of times each drone returns to recharge is less than the set number.

[0016] The technical effects and advantages of the remote monitoring and analysis method for agricultural planting environment data of this invention are as follows: (1) By planning the spraying path, the area is dynamically divided into different levels and the operation blocks are divided. The energy consumption of UAV turning, the operation priority of different severity levels, and the influence of real-time wind direction and speed on spray drift are comprehensively considered. The anti-drift and high-efficiency flight path is automatically generated. For high-level pest blocks, a dual-coverage strategy is launched to ensure the control effect in the core area. In the spraying decision-making stage, by setting a pesticide spraying amount prediction model, not only are basic elements such as the area to be treated and the pest level considered, but also variables such as crop canopy density, real-time wind speed and temperature, and pest resistance risk level calculated from historical pesticide application records are integrated in real time. Compared with the traditional fixed dosage mode, this method can significantly reduce the waste and pollution caused by excessive use of pesticides while ensuring the control effect.

[0017] (2) By setting up growth scores and pest population prediction models, the depth of agricultural environmental monitoring and the foresight of decision-making have been significantly improved. At the crop growth assessment level, for parameters such as soil temperature, humidity, and electrical conductivity that have clear optimal ranges, piecewise functions are used for normalization, which accurately reflects the linear or nonlinear stress caused by the deviation of parameters from the optimal value on growth. This makes the final growth score more representative of the true physiological health status of crops, rather than a simple summation of environmental conditions. At the pest prediction level, a coupled prediction model of "crop health-environmental drive-pest population" has been constructed. This model takes the growth score reflecting the crop's intrinsic resistance and its dynamic change rate as one of the core inputs, and also incorporates key environmental driving factors that directly affect the development and migration of pests, such as cumulative temperature and rainfall, to achieve a more scientific and forward-looking risk warning. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0020] Reference Figure 1 A method for remote monitoring and analysis of agricultural planting environment data includes the following steps: S1: Collect environmental data for each unit area within the farmland using data acquisition equipment, and normalize the environmental data. S2: Generate growth scores for farmland unit areas based on the normalized data, and derive the growth score change rate of unit areas based on multiple growth scores; S3: Obtain historical pest quantity data and build a pest quantity prediction model. Use the pest quantity prediction model to predict the pest quantity in the unit area and classify the unit area according to the pest quantity in the unit area. S4: Based on the unit area classification results, plan the spraying path for the drone to spray pesticides. The drone sprays pesticides along the spraying path for the intermediate pest unit area and the advanced pest unit area. S5: Obtain historical pesticide spraying data and construct a pesticide spraying prediction model. Use the pesticide spraying prediction model to predict the pesticide spraying amount required for all intermediate-level pest unit areas and advanced-level pest unit areas. S6: Determine the spraying time based on the predicted pesticide spraying amount, and decide on the drone's spraying strategy based on the spraying time, flight time, and the amount of pesticide stored in the drone.

[0021] It should be further explained that, in the specific implementation process, the agricultural planting area is planted with specific crops, and the agricultural planting area is divided into multiple farmlands. Each farmland is evenly divided into multiple unit areas. The environmental data includes soil moisture, soil temperature, average crop height, soil conductivity, and soil organic matter content. Soil moisture is measured within a unit area using a soil moisture sensor employing either time-domain reflectometry or frequency-domain reflectometry. Data is collected at regular intervals, and the daily average value is calculated. The soil moisture content is then normalized. The optimal moisture range for soil for crops is — Specifically, the range is 18%–25%, and the soil's tolerance for crop moisture content is within this range. — Specifically, it ranges from 10% to 35%. This represents the measured soil moisture. when < < Ms= ; when < < Ms=1; when < < Ms= ; when < or > Ms=0; In the formula, Ms is the normalized soil moisture, and a and b are calibration coefficients, which are obtained through training with historical laboratory data, specifically 0.2 and 0.8; Soil temperature within a unit area is measured using a thermistor or thermocouple soil temperature sensor. Data is collected at regular intervals, and the daily average value is calculated. The soil temperature is then normalized. The optimal temperature range for soil for crops is — Specifically, the temperature tolerance range for soil to crops is 20℃-28℃. — Specifically, the temperature ranges from 5℃ to 35℃. This represents the measured soil temperature. when < < T= ; when < < T=1; when < < T= ; when < or > T=0; In the formula, T is the normalized soil temperature, and c and d are calibration coefficients, which are obtained through training with historical laboratory data, specifically 0.2 and 0.8; The height of crops within a unit area is measured using LiDAR or a binocular stereo vision camera. The average height of crops is then calculated by averaging the values, and the average height is normalized.

[0022] In the formula, To normalize the average height of crops, This represents the measured average height of crops. This is the theoretical maximum height of the crop at this growth stage, specifically 80cm; The soil conductivity was obtained by measuring four points within a unit area using a four-electrode EC sensor and taking the average value. The soil conductivity was then normalized. The optimal electrical conductivity range of soil for crops is: — Specifically, the conductivity tolerance range of the soil for crops is 1.5 dS / m to 2.5 dS / m. — Specifically, it ranges from 0.5 dS / m to 4 dS / m. This represents the measured soil conductivity. when < < , = ; when < < , =1; when < < , = ; when < or > , =0.1; In the formula, To normalize the soil conductivity, e and f are calibration coefficients, obtained through training with historical laboratory data, specifically 0.3 and 0.7; The soil organic matter content within a unit area was measured using a near-infrared spectroscopy soil sensor, and the soil organic matter content was then normalized. The optimal range of organic matter content in soil for crops is as follows: — Specifically, it is 3%–5%, and the soil's tolerance range for crop organic matter content is [missing information]. — Specifically, it ranges from 1% to 8%. This represents the measured soil organic matter content. when < < , = ; when < < , =1; when < < , = ; when < or > , =0.2; In the formula, To normalize soil organic matter content, These are calibration coefficients, obtained through training with historical laboratory data, specifically 0.4 and 0.6.

[0023] It should be further explained that, in the specific implementation process, the process of generating growth scores for farmland unit areas based on the normalized data includes: Growth score S:

[0024] In the formula, , , , and These are weighting coefficients, obtained from training on historical data, and set to 0.3, 0.25, 0.2, 0.15, and 0.1 respectively. If the environmental data for a certain unit area is: It is 20%. It is 25℃. It is 40cm. It is 2dS / m. If the growth score is 9%, then the crop growth score S in this unit area is 0.82.

[0025] It should be further explained that, in the specific implementation process, the process of deriving the growth score change rate of a unit region based on multiple growth scores includes: A growth scoring evaluation cycle is set. At each growth scoring evaluation cycle, the corresponding growth score of each unit area is obtained. The growth scores of the previous 2n growth scoring evaluation cycles are taken for each unit area. The 2n growth scores are arranged in chronological order. The average of the first n growth scores is calculated to obtain the first growth score average. The average of the last n growth scores is calculated to obtain the second growth score average. The second growth score average is subtracted from the first growth score average to obtain the score difference. The score difference is divided by the first growth score average to obtain the growth score change rate.

[0026] It should be further explained that, in the specific implementation process, the process of obtaining historical pest quantity data and constructing a pest quantity prediction model, and then using the pest quantity prediction model to predict the pest quantity within a unit area, includes: The number of pests refers to the number of pests within a unit area. Factors influencing the number of pests within a unit area include: real-time growth score, growth score change rate, microbial diversity index, pesticide residue index, number of natural enemies of pests, and historical pest population. Real-time growth score refers to the latest growth score of the unit region; Microbial biomass based on PLFA analysis yields a microbial diversity index; The pesticide residue index is calculated by the pesticide degradation rate and application time. The number of natural enemies of pests captured per unit time using sex pheromones and sticky traps; The historical pest baseline was obtained per unit time through sex pheromone traps and yellow sticky traps; The historical pest quantity data of a single unit area is obtained. The historical pest quantity data includes the real-time growth score, growth score change rate, microbial diversity index, pesticide residue index, number of natural enemies of pests, historical pest base, and historical pest quantity of the single unit area. Based on the real-time growth score, growth score change rate, microbial diversity index, pesticide residue index, number of natural enemies of pests, historical pest base and corresponding historical pest number in the corresponding unit area of ​​different historical pest quantity data, a pest quantity prediction set is generated and divided into the first training set and the first test set. Construct a first convolutional neural network, using real-time growth score, growth score change rate, microbial diversity index, pesticide residue index, number of natural enemies of pests and historical pest base from different historical pest quantity data in the first training set as the input data of the first convolutional neural network, and using the corresponding historical pest quantity in the first training set as the output data of the first convolutional neural network. The first convolutional neural network is trained to obtain the first initial convolutional neural network. The first initial convolutional neural network is validated using the first test set. The first initial convolutional neural network, whose output is less than or equal to the preset first test error threshold, is used as the pest quantity prediction model. The real-time growth score, growth score change rate, microbial diversity index, pesticide residue index, number of natural enemies of pests, and historical pest population of each unit area are input into the pest population prediction model to obtain the predicted pest population of each unit area. In an embodiment of the present invention, the predicted number of pests in all unit areas is obtained through a pest quantity prediction model. The predicted number of pests is related to real-time growth score, growth score change rate, microbial diversity index, pesticide residue index, number of natural enemies of pests, and historical pest population. The real-time growth score directly affects the predicted number of pests. The higher the real-time growth score, the healthier the crop. Healthy crops are more resistant to pests, and the fewer pests are predicted. Therefore, the real-time growth score is negatively correlated with the number of pests. The growth score change rate directly affects the predicted number of pests. The higher the growth score change rate, the better the crop growth and the fewer the predicted pests. Therefore, the growth score change rate is negatively correlated with the number of pests. The level of the microbial diversity index directly affects the predicted number of pests. The higher the microbial diversity index, the more natural enemies of the pests there are, and the lower the predicted number of pests. Therefore, the microbial diversity index is negatively correlated with the number of pests. The level of pesticide residue index directly affects the predicted number of pests. The higher the pesticide residue index, the lower the predicted number of pests. Therefore, the pesticide residue index and the number of pests are negatively correlated. The number of natural enemies of pests directly affects the predicted number of pests. The more natural enemies there are, the fewer pests are predicted. Therefore, the number of natural enemies of pests is negatively correlated with the number of pests. The size of the historical pest population directly affects the predicted pest population. The larger the historical pest population, the more pests reproduce, and the higher the predicted pest population. Therefore, the historical pest population and the pest population are positively correlated.

[0027] It should be further explained that, in the specific implementation process, the process of classifying unit areas based on the number of pests within the unit area includes: The predicted number of pests in each unit area is obtained based on the pest quantity prediction model. An appropriate first pest quantity threshold and second pest quantity threshold are set based on historical pest quantity data. The historical pest quantity data refers to the data set of pest quantities in previous unit areas. The number of pests in each unit area is compared with the two pest quantity thresholds. If the number of pests in a unit area is less than the first pest number threshold, the unit area is judged as a low-level pest unit area and no intervention is required for this unit area. When the number of the first pest is less than the number of pests in the unit area, which is less than the number of the second pest, the unit area is determined to be a medium-level pest unit area, and pesticide spraying by drone is required. If the number of pests in a unit area exceeds the threshold for the second type of pest, the unit area is identified as a high-level pest unit area and requires pesticide spraying by drone.

[0028] It should be further explained that, in the specific implementation process, the spraying path for drone pesticide spraying is planned according to the unit area classification results. The process of drone spraying pesticides along the spraying path on intermediate-level and advanced-level pest unit areas includes: Based on the pest level classification results of the unit area, the drone spraying path planning is performed. The unit areas marked as medium and high pests are spatially clustered. The clustering radius is 1.5 times the drone spraying width. Spatially adjacent unit areas with the same or similar pest levels are merged into continuous operation blocks. The start and end points of the drone are marked. Starting from the start point, the nearest block is iteratively selected to be added to the path until all blocks are covered. Finally, the path is connected to the end point to form a complete path, thus obtaining the drone pesticide spraying path. For a single operational block, the optimal flight direction is determined based on the real-time wind direction. A flight direction perpendicular to the wind direction is preferred to reduce droplet drift. When the wind speed exceeds 5 m / s, the flight direction is switched to tailwind. A reciprocating sweeping path is used to achieve full coverage within the block. The track interval N = 0.8B is calculated based on the effective spray width B of the UAV at the set flight altitude to ensure a 20% overlap between adjacent tracks. For blocks with advanced pest units, a dual coverage strategy is implemented. After the first regular sweep, a second enhanced spray is carried out in an orthogonal direction targeting the core pest sub-area.

[0029] It should be further explained that, in the specific implementation process, the process of acquiring historical pesticide spraying data and constructing a pesticide spraying prediction model, and then using this model to predict the required pesticide spraying amounts for all intermediate and advanced pest unit areas, includes: The amount of pesticide sprayed refers to the amount of pesticide required for the unit area to be sprayed, and the unit area to be sprayed refers to all intermediate pest unit areas and advanced pest unit areas. Factors affecting the amount of pesticides required for all intermediate and advanced pest units include: the number of intermediate pest units, the number of advanced pest units, real-time wind speed, spray loss, crop canopy density, and real-time temperature. Real-time wind speed is monitored in real time using a wind speed sensor; The spraying loss is calculated based on equipment parameters and experience. Crop canopy density is calculated based on canopy porosity from lidar point clouds; Real-time temperature is monitored by a temperature sensor. Obtain historical pesticide spraying data for the unit area to be sprayed. The historical pesticide spraying data includes the number of intermediate pest units, the number of advanced pest units, real-time wind speed, spraying loss, crop canopy density, real-time temperature, and the historical pesticide spraying amount for that single unit area. Based on the number of intermediate pest units, the number of advanced pest units, real-time wind speed, spray loss, crop canopy density, real-time temperature and corresponding historical pesticide spraying data of different historical pesticide spraying amounts, a pesticide spraying amount prediction set is generated and divided into a second training set and a second test set. A second convolutional neural network is constructed. The number of intermediate pest units, the number of advanced pest units, real-time wind speed, spray loss, crop canopy density, and real-time temperature in the different historical pesticide spraying data in the second training set are used as the input data of the second convolutional neural network, and the corresponding historical pesticide spraying amount in the second training set is used as the output data of the second convolutional neural network. The second convolutional neural network is trained to obtain the second initial convolutional neural network. The second initial convolutional neural network is validated using the second test set. The second initial convolutional neural network that outputs a second test error threshold less than or equal to the preset second test error threshold is used as the pesticide spraying amount prediction model. The number of intermediate pest units, the number of advanced pest units, real-time wind speed, spray loss, crop canopy density, and real-time temperature of the unit to be sprayed are input into the pesticide spraying amount prediction model to obtain the predicted pesticide spraying amount for the unit to be sprayed. In an embodiment of the present invention, the predicted pesticide spraying amount for the unit area to be sprayed is obtained by a pesticide spraying amount prediction model. The predicted pesticide spraying amount is related to the number of intermediate pest unit areas, the number of advanced pest unit areas, real-time wind speed, spraying loss, crop canopy density and real-time temperature. The number of intermediate-level and advanced-level pest units directly affects the amount of pesticides to be sprayed. The more intermediate-level and advanced-level pest units there are, the more pesticides need to be sprayed, and the more pesticides are predicted to be sprayed. Therefore, the number of intermediate-level and advanced-level pest units is positively correlated with the amount of pesticides to be sprayed. The magnitude of real-time wind speed directly affects the amount of pesticides to be sprayed. The higher the real-time wind speed, the more pesticides need to be used to compensate for drift loss, and the more pesticides are predicted to be sprayed. Therefore, real-time wind speed and pesticide spraying amount are positively correlated. The amount of spray loss directly affects the amount of pesticide to be sprayed. The more spray loss, the more pesticide to be sprayed. Therefore, spray loss and pesticide spraying amount are positively correlated. The density of crop canopy directly affects the amount of pesticides that can be sprayed. The higher the crop canopy density, the more pesticides are needed to reach the lower layers, and the more pesticides can be sprayed. Therefore, crop canopy density and pesticide spraying amount are positively correlated. The real-time temperature directly affects the amount of pesticide to be sprayed. The higher the real-time temperature, the more pesticides volatilize, requiring an increase in pesticide dosage and thus a higher predicted amount of pesticide to be sprayed. Therefore, real-time temperature and pesticide spraying amount are positively correlated.

[0030] It should be further explained that, in the specific implementation process, the process of determining the spraying time based on the predicted pesticide spraying volume, and deciding on the drone's spraying strategy based on the spraying time, flight time, and the drone's pesticide storage capacity includes: The predicted pesticide spraying amount is obtained based on the pesticide spraying amount prediction model. The optimal pesticide spraying speed of the drone is set according to the real-time wind speed. The spraying time is obtained by dividing the predicted pesticide spraying amount by the optimal pesticide spraying speed. The remaining flight time is set according to the flight speed of the drone. The remaining flight time refers to the flight time of the drone when it is not spraying. If the predicted pesticide spraying amount is less than the pesticide storage capacity of the drone, and the spraying time plus the remaining flight time is less than the endurance time, then the drone will spray according to the predetermined spraying path. If the predicted pesticide spraying amount is less than the pesticide storage capacity of the drone, and the spraying time plus the remaining flight time is greater than the endurance time, the drone will spray according to the predetermined spraying path. If the drone's battery level drops below the alarm value midway, it needs to return to recharge and then return to spray according to the spraying path again. If the number of times it returns to recharge is greater than the set number, a specific number of drones need to be dispatched to assist in spraying before spraying, and the spraying path of each drone needs to be replanned. The specific number is related to the number of times it returns to recharge, so that the number of times each drone returns to recharge is less than the set number. If the predicted pesticide spraying amount is greater than the pesticide storage capacity of the drone, and the spraying time plus the remaining flight time is less than the endurance time, the drone will spray according to the predetermined spraying path. The drone needs to return midway to replenish the pesticide, and after replenishment, it will return again to spray according to the spraying path. If the predicted pesticide spraying volume exceeds the drone's pesticide storage capacity, and the spraying time plus the remaining flight time exceeds the endurance time, the drone will spray according to the predetermined spraying path. The drone needs to return midway to replenish pesticides, and after replenishment, it will return again to spray according to the spraying path. If the drone's battery level drops below the alarm value midway, it needs to return to recharge and then return again to spray according to the spraying path. If the number of times it returns to recharge exceeds the set number, a specific number of drones needs to be dispatched to assist in spraying before spraying, and the spraying path of each drone needs to be replanned. The specific number is related to the number of times it returns to recharge, so that the number of times each drone returns to recharge is less than the set number.

[0031] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0032] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for remote monitoring and analysis of agricultural planting environment data, characterized in that, Includes the following steps: S1: Collect environmental data for each unit area within the farmland using data acquisition equipment, and normalize the environmental data. S2: Generate growth scores for farmland unit areas based on the normalized data, and derive the growth score change rate of the unit areas based on multiple growth scores; S3: Obtain historical pest quantity data and build a pest quantity prediction model. Use the pest quantity prediction model to predict the pest quantity in the unit area and classify the unit area according to the pest quantity in the unit area. S4: Based on the unit area classification results, plan the spraying path for the drone to spray pesticides. The drone sprays pesticides along the spraying path for the intermediate pest unit area and the advanced pest unit area. S5: Obtain historical pesticide spraying data and construct a pesticide spraying prediction model. Use the pesticide spraying prediction model to predict the pesticide spraying amount required for all intermediate-level pest unit areas and advanced-level pest unit areas. S6: Determine the spraying time based on the predicted pesticide spraying amount, and decide on the drone's spraying strategy based on the spraying time, flight time, and the amount of pesticide stored in the drone.

2. The method for remote monitoring and analysis of agricultural planting environment data according to claim 1, characterized in that, The agricultural planting area is planted with specific crops. The agricultural planting area is divided into multiple farmlands, and each farmland is evenly divided into multiple unit areas. The environmental data includes soil moisture, soil temperature, average crop height, soil conductivity, and soil organic matter content. Soil moisture is measured within a unit area using a soil moisture sensor employing either time-domain reflectometry or frequency-domain reflectometry. Data is collected at regular intervals, and the daily average value is calculated. The soil moisture content is then normalized. The optimal moisture range for soil for crops is — The soil's tolerance range for crop moisture is — , This represents the measured soil moisture. when < < Ms= ; when < < Ms=1; when < < Ms= ; when < or > Ms=0; In the formula, Ms is the normalized soil moisture, and a and b are calibration coefficients obtained through training with historical laboratory data; Soil temperature within a unit area is measured using a thermistor or thermocouple soil temperature sensor. Data is collected at regular intervals, and the daily average value is calculated. The soil temperature is then normalized. The optimal temperature range for soil for crops is — The soil's temperature tolerance range for crops is — , This represents the measured soil temperature. when < < T= ; when < < T=1; when < < T= ; when < or > T=0; In the formula, T is the normalized soil temperature, and c and d are calibration coefficients obtained through training with historical laboratory data; The height of crops within a unit area is measured using LiDAR or a binocular stereo vision camera. The average height of crops is then calculated by averaging the values, and the average height is normalized. In the formula, To normalize the average height of crops, This represents the measured average height of crops. This represents the theoretical maximum height of the crop at this growth stage. The soil conductivity was obtained by measuring four points within a unit area using a four-electrode EC sensor and taking the average value. The soil conductivity was then normalized. The optimal electrical conductivity range of soil for crops is: — The soil's tolerance range for agricultural electrical conductivity is [range missing]. — , This represents the measured soil conductivity. when < < , = ; when < < , =1; when < < , = ; when < or > , =0.1; In the formula, The normalized soil conductivity is represented by e and f, which are calibration coefficients obtained through training with historical laboratory data. The soil organic matter content within a unit area was measured using a near-infrared spectroscopy soil sensor, and the soil organic matter content was then normalized. The optimal range of organic matter content in soil for crops is as follows: — The soil's tolerance range for crop organic matter content is... — , This represents the measured soil organic matter content. when < < , = ; when < < , =1; when < < , = ; when < or > , =0.2; In the formula, To normalize soil organic matter content, These are calibration coefficients, obtained through training with historical laboratory data.

3. The method for remote monitoring and analysis of agricultural planting environment data according to claim 2, characterized in that, The process of generating growth scores for farmland unit areas based on normalized data includes: Growth score S: In the formula, , , , and These are weighting coefficients, obtained through training based on historical data.

4. The method for remote monitoring and analysis of agricultural planting environment data according to claim 3, characterized in that, The process of deriving the growth score change rate of a unit region based on multiple growth scores includes: A growth score evaluation cycle is set. At each growth score evaluation cycle, the corresponding growth score of each unit area is obtained. The growth scores of the previous 2n growth score evaluation cycles are taken for each unit area. The 2n growth scores are arranged in chronological order. The average of the first n growth scores is calculated to obtain the first growth score average. The average of the last n growth scores is calculated to obtain the second growth score average. The second growth score average is subtracted from the first growth score average to obtain the score difference. The score difference is divided by the first growth score average to obtain the growth score change rate.

5. The method for remote monitoring and analysis of agricultural planting environment data according to claim 4, characterized in that, The process of acquiring historical pest population data and constructing a pest population prediction model, and then using this model to predict the pest population within a unit area, includes: The number of pests refers to the number of pests within a unit area. Factors influencing the number of pests within a unit area include: real-time growth score, growth score change rate, microbial diversity index, pesticide residue index, number of natural enemies of pests, and historical pest population. Real-time growth score refers to the latest growth score of the unit region; Microbial biomass based on PLFA analysis yields a microbial diversity index; The pesticide residue index is calculated by the pesticide degradation rate and application time. The number of natural enemies of pests captured per unit time using sex pheromones and sticky traps; The historical pest baseline was obtained per unit time through sex pheromone traps and yellow sticky traps; The historical pest quantity data of a single unit area is obtained. The historical pest quantity data includes the real-time growth score, growth score change rate, microbial diversity index, pesticide residue index, number of natural enemies of pests, historical pest base, and historical pest quantity of the single unit area. Based on the real-time growth score, growth score change rate, microbial diversity index, pesticide residue index, number of natural enemies of pests, historical pest base and corresponding historical pest number in the corresponding unit area of ​​different historical pest quantity data, a pest quantity prediction set is generated and divided into the first training set and the first test set. Construct a first convolutional neural network, using real-time growth score, growth score change rate, microbial diversity index, pesticide residue index, number of natural enemies of pests and historical pest base from different historical pest quantity data in the first training set as the input data of the first convolutional neural network, and using the corresponding historical pest quantity in the first training set as the output data of the first convolutional neural network. The first convolutional neural network is trained to obtain the first initial convolutional neural network. The first initial convolutional neural network is validated using the first test set. The first initial convolutional neural network, whose output is less than or equal to the preset first test error threshold, is used as the pest quantity prediction model. The real-time growth score, growth score change rate, microbial diversity index, pesticide residue index, number of natural enemies of pests, and historical pest population of each unit area are input into the pest population prediction model to obtain the predicted pest population of each unit area.

6. The method for remote monitoring and analysis of agricultural planting environment data according to claim 5, characterized in that, The process of classifying a unit area based on the number of pests within that unit area includes: The predicted number of pests in each unit area is obtained based on the pest quantity prediction model. An appropriate first pest quantity threshold and second pest quantity threshold are set based on historical pest quantity data. The historical pest quantity data refers to the data set of pest quantities in previous unit areas. The number of pests in each unit area is compared with the two pest quantity thresholds. If the number of pests in a unit area is less than the first pest number threshold, the unit area is judged as a low-level pest unit area and no intervention is required for this unit area. When the number of the first pest is less than the number of pests in the unit area, which is less than the number of the second pest, the unit area is determined to be a medium-level pest unit area, and pesticide spraying by drone is required. If the number of pests in a unit area exceeds the threshold for the second type of pest, the unit area is identified as a high-level pest unit area and requires pesticide spraying by drone.

7. The method for remote monitoring and analysis of agricultural planting environment data according to claim 6, characterized in that, Based on the unit area classification results, the spraying path for drone pesticide spraying is planned. The process of drone spraying pesticides along the spraying path for intermediate and advanced pest unit areas includes: Based on the pest level classification results of the unit area, the drone spraying path planning is performed. The unit areas marked as medium and high pests are spatially clustered. The clustering radius is 1.5 times the drone spraying width. Spatially adjacent unit areas with the same or similar pest levels are merged into continuous operation blocks. The start and end points of the drone are marked. Starting from the start point, the nearest block is iteratively selected to be added to the path until all blocks are covered. Finally, the path is connected to the end point to form a complete path, thus obtaining the drone pesticide spraying path. For a single operational block, the optimal flight direction is determined based on the real-time wind direction. A flight direction perpendicular to the wind direction is preferred to reduce droplet drift. When the wind speed exceeds 5 m / s, the flight direction is switched to tailwind. A reciprocating sweeping path is used to achieve full coverage within the block. The track interval N = 0.8B is calculated based on the effective spray width B of the UAV at the set flight altitude to ensure a 20% overlap between adjacent tracks. For blocks with advanced pest units, a dual coverage strategy is implemented. After the first regular sweep, a second enhanced spray is carried out in an orthogonal direction targeting the core pest sub-area.

8. The method for remote monitoring and analysis of agricultural planting environment data according to claim 7, characterized in that, The process of acquiring historical pesticide spraying data and constructing a pesticide spraying prediction model, and then using this model to predict the required pesticide spraying amounts for all intermediate and advanced pest unit areas, includes: The amount of pesticide sprayed refers to the amount of pesticide required for the unit area to be sprayed, and the unit area to be sprayed refers to all intermediate pest unit areas and advanced pest unit areas. Factors affecting the amount of pesticides required for all intermediate and advanced pest units include: the number of intermediate pest units, the number of advanced pest units, real-time wind speed, spray loss, crop canopy density, and real-time temperature. Real-time wind speed is monitored in real time using a wind speed sensor; The spraying loss is calculated based on equipment parameters and experience. Crop canopy density is calculated based on canopy porosity from lidar point clouds; Real-time temperature is monitored by a temperature sensor. Obtain historical pesticide spraying data for the unit area to be sprayed. The historical pesticide spraying data includes the number of intermediate pest units, the number of advanced pest units, real-time wind speed, spraying loss, crop canopy density, real-time temperature, and the historical pesticide spraying amount for that single unit area. Based on the number of intermediate pest units, the number of advanced pest units, real-time wind speed, spray loss, crop canopy density, real-time temperature and corresponding historical pesticide spraying data of different historical pesticide spraying amounts, a pesticide spraying amount prediction set is generated and divided into a second training set and a second test set. A second convolutional neural network is constructed. The number of intermediate pest units, the number of advanced pest units, real-time wind speed, spray loss, crop canopy density, and real-time temperature in the different historical pesticide spraying data in the second training set are used as the input data of the second convolutional neural network, and the corresponding historical pesticide spraying amount in the second training set is used as the output data of the second convolutional neural network. The second convolutional neural network is trained to obtain the second initial convolutional neural network. The second initial convolutional neural network is validated using the second test set. The second initial convolutional neural network that outputs a second test error threshold less than or equal to the preset second test error threshold is used as the pesticide spraying amount prediction model. The number of intermediate pest units, the number of advanced pest units, real-time wind speed, spray loss, crop canopy density, and real-time temperature of the area to be sprayed are input into the pesticide spraying amount prediction model to obtain the predicted pesticide spraying amount for the area to be sprayed.

9. The method for remote monitoring and analysis of agricultural planting environment data according to claim 8, characterized in that, The process of determining the spraying time based on the predicted pesticide spraying volume, and deciding on the drone's spraying strategy based on the spraying time, flight time, and the amount of pesticide stored in the drone, includes: The predicted pesticide spraying amount is obtained based on the pesticide spraying amount prediction model. The optimal pesticide spraying speed of the drone is set according to the real-time wind speed. The spraying time is obtained by dividing the predicted pesticide spraying amount by the optimal pesticide spraying speed. The remaining flight time is set according to the flight speed of the drone. The remaining flight time refers to the flight time of the drone when it is not spraying. If the predicted pesticide spraying amount is less than the pesticide storage capacity of the drone, and the spraying time plus the remaining flight time is less than the endurance time, then the drone will spray according to the predetermined spraying path. If the predicted pesticide spraying amount is less than the pesticide storage capacity of the drone, and the spraying time plus the remaining flight time is greater than the endurance time, the drone will spray according to the predetermined spraying path. If the drone's battery level drops below the alarm value midway, it needs to return to recharge and then return to spray according to the spraying path again. If the number of times it returns to recharge is greater than the set number, a specific number of drones need to be dispatched to assist in spraying before spraying, and the spraying path of each drone needs to be replanned. The specific number is related to the number of times it returns to recharge, so that the number of times each drone returns to recharge is less than the set number. If the predicted pesticide spraying amount is greater than the pesticide storage capacity of the drone, and the spraying time plus the remaining flight time is less than the endurance time, the drone will spray according to the predetermined spraying path. The drone needs to return midway to replenish the pesticide, and after replenishment, it will return again to spray according to the spraying path. If the predicted pesticide spraying volume exceeds the drone's pesticide storage capacity, and the spraying time plus the remaining flight time exceeds the endurance time, the drone will spray according to the predetermined spraying path. The drone needs to return midway to replenish pesticides, and after replenishment, it will return again to spray according to the spraying path. If the drone's battery level drops below the alarm value midway, it needs to return to recharge and then return again to spray according to the spraying path. If the number of times it returns to recharge exceeds the set number, a specific number of drones needs to be dispatched to assist in spraying before spraying, and the spraying path of each drone needs to be replanned. The specific number is related to the number of times it returns to recharge, so that the number of times each drone returns to recharge is less than the set number.

Images