Dioscorea opposita water, fertilizer and pesticide accurate management and control method and management and control platform based on big data analysis

Abstract

The invention belongs to the technical field of agricultural informatization, and more specifically relates to a Chinese yam water, fertilizer and pesticide accurate management and control method and management and control platform based on big data analysis. The method comprises the following steps: acquiring Chinese yam growth data through each acquisition device; performing data cleaning on the collected data to remove abnormal data and performing data classified storage; a precise water control strategy is established to control the watering amount in the current period according to the water deficit, a precise fertilizer control strategy is established to control the fertilizer supply amount in the current period according to the historical fertilizer application amount and fertilizer deficit judgment on the Chinese yam image, and a precise pesticide control strategy is established to give an alarm according to disease and pest recognition on the Chinese yam image; a user can check conditions in real time through a visual interface, and manual management and control can be achieved. The method solves the problems that a traditional management mode is extensive, improper use of watering, fertilization, pesticide and the like affects growth and development of Chinese yams, damages the ecological environment, lacks real-time monitoring and accurate regulation and control and the like.

Description

Technical Field

[0001] The present invention belongs to the technical field of agricultural informatization, and more specifically, relates to a method and platform for precise control of yam water, fertilizer and pesticide based on big data analysis. Background Art

[0002] As an important crop with both medicinal and edible properties, yam cultivation requires careful management of water, fertilizer, and pesticides, directly impacting yield and quality. Traditional cultivation methods rely on experience and lack precise control, leading to resource waste, environmental pollution, and low returns. While precision agriculture is becoming a trend with the development of the Internet of Things and big data technologies, a precise control platform specifically for yam cultivation remains lacking.

[0003] Chinese patent document CN202011401190.X discloses an intelligent management and control platform for agricultural plant protection drones based on big data, including a main sprayer, a side sprayer, a management and control platform end, and a scanner. The main sprayer includes a main spraying unit and a main dosage adjustment unit, and the side sprayer includes a side spraying unit and a side dosage adjustment unit. The management and control platform end includes a flight command module, a spraying type control module, and a spraying dosage control module. The scanner includes a plant image extraction module, a type determination module, a maturity determination module, a type model storage module, and a maturity model storage module. The flight command module is electrically connected to the main sprayer, the side sprayer, and the scanner. The main spray unit and the side spray unit are electrically connected to the spray type control module. The main dosage adjustment unit and the side dosage adjustment unit are electrically connected to the spraying dosage control module.

[0004] The current water, fertilizer and pesticide management methods for yam have the following problems: 1. Traditional extensive management methods: Traditional water, fertilizer, and pesticide management in yam cultivation often relies on farmers' experience and judgment, lacking a scientific basis. Watering times and amounts are often based on intuition, resulting in either excessive soil moisture, which can cause root diseases, or insufficient soil moisture, which can affect yam growth and development. Excessive or irrational fertilization is common, resulting in fertilizer waste and increased production costs, while also potentially leading to soil compaction and environmental pollution. Pesticide use is also imprecise, with an overreliance on highly toxic pesticides and inappropriate application timing, which not only affects yam quality but also damages the ecological environment. 2. Lack of real-time monitoring and precise regulation: The growing environment of yam is complex and ever-changing, with factors such as soil fertility, moisture content, and weather conditions constantly influencing its growth. Traditional management methods struggle to capture real-time changes in these environmental parameters, making it difficult to precisely regulate water, fertilizer, and pesticide use based on the yam's growth. For example, in the event of a sudden drought or rainfall, irrigation and fertilization plans cannot be adjusted quickly, negatively impacting yam growth. Summary of the Invention

[0005] The present invention aims to overcome at least one defect of the above-mentioned prior art and provide a method and platform for precise control of yam water, fertilizer and pesticide based on big data analysis, so as to solve the problems of extensive traditional management methods, improper use of watering, fertilizing, pesticides, etc., which affect the growth and development of yam and cause damage to the ecological environment, and lack of real-time monitoring and precise control.

[0006] The detailed technical solutions of the present invention are as follows: A method for precise control of yam water, fertilizer and pesticide based on big data analysis, the method comprising: S1. Collecting yam growth data through various collection devices, including but not limited to weather data, environmental data, and yam image data. The environmental data includes air temperature and humidity, soil temperature and humidity, light and rainfall; S2. Clean the collected data to remove abnormal data and classify and store the data; First, the original data is cleaned to eliminate abnormal data. The methods used include but are not limited to the mean filling algorithm, isolation forest algorithm, etc., and then classified, managed and stored according to data type and collection time.

[0007] S3. Establish a precise water management strategy to control the amount of watering required for the current cycle based on the water shortage. Establish a precise fertilizer management strategy to control the fertilizer supply within the current cycle based on historical fertilizer application amounts and fertilizer shortage determination based on yam images. Establish a precise pesticide management strategy to issue alarms based on pest and disease identification based on yam images. S4. Users can view weather data, environmental data, yam image data and alarm status in real time through the visual interface, and can manually control water management, fertilizer management and pesticide management.

[0008] Furthermore, a collection interval time point is set. Before the collection interval time point is reached, each collection device is in a deep sleep state, and only the heartbeat packet is kept to keep the device online; When the set collection time point is reached, the sensor data collected at the current collection time point is compared with the data collected at the previous collection time point. When the data change exceeds the set difference threshold, the device will be woken up and the collected data will be uploaded; Specifically, the soil temperature and humidity sensor collects and transmits data every 10 minutes, the light sensor collects and transmits data every 5 minutes, the air temperature and humidity sensor collects and transmits data every 15 minutes, and the rainfall sensor records and transmits rainfall in real time during rainfall. The image acquisition device captures and uploads images of yam growth every 30 minutes. Sensors use the object model JSON format to connect to data acquisition devices. The object model JSON includes three types of JSON: attributes, services, and events. Attribute JSON is used to upload sensor data. Service JSON supports remote device upgrades, remote ad hoc networking, and remote parameter modifications. Event JSON is used for alarms, where the alarm threshold can be modified according to needs. When the environmental parameters detected by the sensor exceed the normal range, the corresponding equipment will issue an early warning message to alert the management personnel.

[0009] Furthermore, the precise water management and control strategy introduces a periodic water balance model based on a historical curve, combines real-time data and image data visual feedback closed-loop control mechanism, and calculates the weather data Q1(t) in the period, the amount of water applied Q2(t) in the period, the amount of evaporated water Q3(t), and the amount of water shortage Qf(t) fed back by visual images based on the watering data Q(in) in the current period to determine the current amount of water to be applied. The calculation formula for the amount of water to be applied in the current period of the precise water management and control strategy is: + dt(1) In formula (1), Q(out) is the amount of water that should be applied in this cycle, Q(in) is the amount of water applied in this cycle according to the historical curve, and t is the time of this cycle; i is the serial number of the water factor, Qi(t) is the amount of water applied on the day of the water factor, including Q1(t), Q2(t), and Q3(t), where Q1(t) is the weather data in the cycle, Q2(t) is the amount of water applied in this cycle, and Q3(t) is the amount of evaporated water; Ci is the weight of the water factor, including C1, C2, and C3, where C1 is the weight of the weather data in the cycle, C2 is the weight of the amount of water applied in this cycle, and C3 is the weight of the amount of evaporated water; Qf(t) is the amount of water shortage reported by the visual image, and Cf is the weight of the amount of water shortage reported by the visual image; In formula (1), Q(in) is the amount of watering in the current cycle of the historical curve. Based on the historical data, the daily watering amount calculation formula is P=P1+P2-P3, where P1 is the rainfall collected on the day, P2 is the amount of watering on the day, and P3 is the amount of evaporated water on the day. The watering amount calculation formula for the cycle is: Q(in)= )dt(2); In formula (1), Q1(t) is the expected rainfall in the weather forecast, and the calculation formula is: Q1(t)= - ), where M(t) is the expected rainfall in the future weather forecast within the period, and N(t) is the rainfall collected by the rain sensor on the past days within the period; In formula (1), Q2(t) is the amount of water applied during this period, which is directly obtained from the operation data; In formula (1), Q3(t) is the expected amount of evaporated water, and the calculation formula is: Q3(t)=k*f(Ta,Ha)*g(Ts,Hs)*h(I)*S(3); In formula (3), S is the benchmark evaporation water per mu. The method for obtaining it is to collect 1 square meter of land in the yam planting area every day, weigh it before and after 24 hours to obtain the daily evaporation water, take the average of the daily evaporation water during the yam growth period, and multiply the average by 666.67 to obtain the benchmark evaporation water per mu; k is the comprehensive correction coefficient, which is set according to the specific land and crop conditions; f(Ta,Ha) is the air temperature and humidity function, expressed as: f(Ta,Ha)=(Ta+273.15) / (1-Ha / 100), where Ta is the air temperature and Ha is the relative humidity; g(Ts,Hs) is the soil temperature and moisture function, expressed as: f(Ts,Hs)=(Ta+273.15) / (Hs+1) where Ts is the soil temperature and Hs is the soil volumetric water content; h(I) is the light intensity function, expressed as: h(I)=I / I0, where I is the actual light intensity, unit: lx, and I0 is the reference light intensity value, unit: lx, which is the light intensity at noon on a sunny and cloudless day. Qf(t) is the amount of water shortage reported by visual images. Based on the color changes of yam leaves, the color histogram of the leaf area is statistically analyzed to calculate the variance of the bright green portion of the green channel. This is then compared with the variance of the bright green portion of the green channel in historical image data. A color ratio threshold is set to determine whether the color feature exceeds the normal range. The water shortage of the yam is then determined and divided into seven levels: excessive, normal, water shortage level 1, water shortage level 2, water shortage level 3, water shortage level 4, and water shortage level 5. Furthermore, when the water shortage level is greater than or equal to level 3, an alarm message is sent via the cloud platform and the user is notified of the current abnormal state of the yam.

[0010] Furthermore, C1 is the weight of weather data within the period, C2 is the weight of watering in this period, and C3 is the weight of evaporation. The weights are determined using an orthogonal rotation experiment. An orthogonal rotation experiment is a method of arranging experiments using an orthogonal rotation table to ensure that the weights are evenly distributed and orthogonally rotatable: Using L9 (3 4 ) Orthogonal rotation table, 9 experiments are conducted in each cycle, covering 4 factors and 3 levels each. The yam experiments in different cycles are independent of each other. The 4 factors are set as C1, C2, C3, and Cf, and the related benefits are set as y. According to the design principle of orthogonal rotation table, the benefit calculation formula is: y=β0+β1C1+β2C2+β3C3+β4C f +β 11 C1C1+β 22 C2C2+β33 C3C3+β 44 C f C f +β 12 C1C2+β 13 C1C3+β 14 C1C f +β 23 C2C f +β 24 C2C f +β 34 C3C f (4); In formula (4), β0, β1, β2, β3, β4, β 11 , β 22 , β 33 , β 44 , β 12 , β 13 , β 14 , β 23 , β 24 , β 34 is the undetermined regression coefficient. By substituting the relevant experimental data, the optimal regression coefficient is calculated according to the least square method, that is, The minimum regression coefficient value, among which the optimal regression coefficient β 0m , β 1m , β 2m , β 3m , β 4m , β 11m , β 22m , β 33m , β 44m , β 12m , β 13m , β 14m , β 23m , β 24m , β 34m To calculate the matrix formula (C T C)β=C T Y is obtained, C is the matrix of experimental data of orthogonal rotation table, C T is the transposed matrix of C, β is the optimal regression coefficient matrix, Y is the profit matrix in the orthogonal rotation table experimental data, yj is the profit of the orthogonal rotation experimental table, and ymax is the optimal profit obtained by substituting the optimal regression coefficient.

[0011] Furthermore, the precise fertilizer management and control strategy first queries the historical fertilization data in the current cycle, then uses the yam image recognition technology with multi-feature fusion to analyze whether the yam is lacking fertilizer, and sets the fertilizer supply in this cycle through closed-loop control; The multi-feature fusion yam image recognition technology includes leaf color analysis based on leaf RGB values, YOLO V7-based leaf and stem feature recognition, and comparison of the identified features with a feature library of normal growth and different fertility deficiency states to determine whether the yam is deficient in fertilizer and the type of fertilizer being deficient. The precise fertilizer control strategy first performs YOLOV7 semantic segmentation on the images captured by the camera to cut out all leaf areas, then performs RGB color analysis and feature analysis. When the leaf color analysis ratio is greater than the historical normal yam leaf color analysis ratio, and the feature statistics are greater than the normal yam feature statistics in the historical image data, fertilizer deficiency adjustment is performed; Leaf color analysis is to count the proportion of bright green, yellow, dark green, yellow-white, and white in the leaf area. Feature statistics are to identify leaf thinning, purple-red stems, burn marks, white spots, and reticular patterns, and perform feature statistics. The calculation formula is: (5); In formula (5), G(x) is the fertilizer deficiency index corresponding to yam, where G(1) is the nitrogen deficiency index, G(2) is the phosphorus deficiency index, G(3) is the potassium deficiency index, G(4) is the calcium deficiency index, G(5) is the magnesium deficiency index, and G(6) is the iron deficiency index. V(x) represents the corresponding characteristic number of the current yam, where V(1) is the leaf thinning characteristic number, V(2) is the stem purple-red characteristic number, V(3) is the burn mark characteristic number, V(4) is the white spot characteristic number, V(5) is the mesh pattern characteristic number, and V(6) is the overall yellow-white characteristic number of the leaf. s represents the characteristic number corresponding to V(x), and sm represents the corresponding V(x). The maximum number of features, n represents the area of ​​yam leaves in the current image; Vm(x) represents the number of features corresponding to yam in the historical image, where Vm(1) is the number of features of thinning leaves in the historical image, Vm(2) is the number of features of purple-red stems in the historical image, Vm(3) is the number of features of burn marks in the historical image, Vm(4) is the number of features of white spots in the historical image, Vm(5) is the number of features of mesh patterns in the historical image, Vm(6) is the number of features of yellow-white leaves as a whole in the historical image, l represents the feature number corresponding to Vm(x), lm represents the maximum number of features corresponding to Vm(x), and m represents the area of ​​yam leaves in the historical image.

[0012] Furthermore, the precise pesticide control strategy uses YOLO V7 to train a variety of pest and disease images, identify relevant pest and disease characteristics, and issue alarms for pest and disease. The pest and disease images include but are not limited to anthrax, brown spot, root rot, wilt, yam sawfly, white grub, cutworm, and nematode. The original pest and disease image dataset is preprocessed by contrast enhancement, brightness enhancement, and color enhancement before model training.

[0013] On the other hand, the present invention also includes a control platform for realizing a method for accurately controlling yam water, fertilizer and pesticide based on big data analysis, comprising: The data acquisition module 101 comprises a sensor network, a high-definition camera and an online weather platform deployed in the yam planting area, and is used to obtain environmental data, image data and weather data. The environmental data, image data and weather data are transmitted to the data transmission and storage module 102 via various communication methods. The data transmission and storage module 102 is used to convert the protocol and interface of the data collected by the data acquisition module 101, transmit the data to the data analysis and decision module 103 via Ethernet, remove abnormal data from the data, use the mean filling algorithm to fill the removed data, and finally transmit it to the MySQL database for classification, management, and storage of the data according to data type and collection time; at the same time, it receives the water, fertilizer, and pesticide strategy generated by the data analysis and decision module 103 and sends it to the intelligent control module 104 to realize precise control of water, fertilizer, and pesticide; The data analysis and decision module 103 integrates historical sensor data, historical weather data, historical growth image data, and historical operation data to implement precise water control strategies, precise fertilizer control strategies, and precise pesticide control strategies. The data analysis and decision module also includes server software that reads data stored in the MySQL database and calculates and issues relevant control instructions based on relevant models, which are sent to the intelligent control module 104 through the data transmission and storage module 102. The intelligent control module 104 sends corresponding instructions to the irrigation, fertilization and spraying equipment according to the received control instructions. The intelligent control module 104 and the data transmission and storage module 102 communicate using Ethernet; The user interaction and monitoring module 105 is used for users to view data, manually operate, and issue real-time alarms. The user interaction and monitoring module 105 and the data transmission and storage module 102 communicate using Ethernet and 4G mobile phone communication networks.

[0014] Furthermore, the data acquisition module 101 includes an air temperature and humidity sensor 201, a soil temperature and humidity sensor 202, a light sensor 203, a rainfall sensor 204, a camera 205, and weather data acquisition software 206. A plurality of soil moisture sensors 202 are arranged in the yam planting area, one installed at a set distance, to collect soil moisture data in real time. A weather station is installed near the yam planting area, and the weather station is connected to sensors including the air temperature and humidity sensor 201, the soil temperature and humidity sensor 202, the light sensor 203, and the rainfall sensor 204. Weather data is obtained from the Internet via the OpenWeatherMap API. Furthermore, the data analysis and decision module 103 includes a historical data acquisition module 303, a yam water, fertilizer and pesticide precision control model 302, and a MySQL database 301; the historical data acquisition module 303 is used to acquire historical sensor data, historical weather data, historical image data, and historical operation data and store them in the MySQL database 301; the MySQL database is used to store historical data and real-time acquired data; the yam water, fertilizer and pesticide precision control model 302 includes the implementation of a precision water control module 304, a precision fertilizer control module 305, and a precision pesticide control module 306; finally, combined with real-time sensor data and real-time images, the intelligent control module can be controlled to achieve relevant water, fertilizer and pesticide precision control, namely watering, fertilizing, and spraying; The intelligent control module 104 includes a main controller 401. The main controller 401 controls the flow sensor 402, the liquid level sensor 403, the stirring motor 404, the water pump 405, the fertilizer pump 406, and the medicine pump 407 according to different instructions to remotely control the valve 408. The valve 408 is used to control the flow of the water, fertilizer, and medicine pipelines. The water pump 405 uses a frequency converter to further control the water volume. The user interaction and monitoring module 105 is used to view weather data, environmental data, yam image data and alarm conditions in real time, and can manually control water management, fertilizer management and pesticide management.

[0015] On the other hand, the present invention also includes a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of a method for precise control of yam water, fertilizer and pesticide based on big data analysis.

[0016] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention provides a method and platform for precise control of yam water, fertilizer and pesticide based on big data analysis. The introduction of the MQTT protocol greatly improves the wireless communication standby time of the yam sensor network. The introduction of the physical model concept can more conveniently expand the yam sensor network and more comprehensively reflect the sensor status.

[0017] (2) The present invention provides a method and platform for precise control of yam water, fertilizer and pesticide based on big data analysis, which integrates environmental data, weather data, growth image data and historical operation data, and uses a convolutional neural network to perform convolution calculation on the water volume variance during the yam growth cycle, so as to rationally allocate irrigation resources.

[0018] (3) The present invention provides a method and platform for precise control of yam water, fertilizer and pesticide based on big data analysis. It uses multi-feature fusion yam fertilizer deficiency image recognition technology to analyze plant features in the image to determine the fertilizer deficiency situation, and uses a machine vision model training program to identify related pests and disease scars, so as to provide timely alarms and spray pesticides.

[0019] (4) The present invention provides a method and platform for precise control of yam water, fertilizer and pesticide based on big data analysis, and a closed-loop feedback control system based on machine vision: the model can provide real-time feedback based on machine vision images and real-time data from sensor networks, adjust watering according to the water shortage level, and adjust fertilizer and spraying according to yam images, which can better adapt to the yam planting needs in different regions, varieties and climatic conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a flow chart of the method for precise control of yam water, fertilizer and pesticide based on big data analysis described in the present invention.

[0021] Figure 2 This is a schematic diagram of the periodic curve water balance calculation in Example 1 of the present invention.

[0022] Figure 3 This is a framework diagram of a precise control platform for yam water, fertilizer and pesticide based on big data analysis described in the present invention.

[0023] Figure 4 This is a schematic diagram of the data acquisition module in Example 2 of the present invention.

[0024] Figure 5 This is a schematic diagram of the data analysis and decision-making module in Example 2 of the present invention.

[0025] Figure 6 This is a schematic diagram of the intelligent control module in Example 2 of the present invention.

[0026] The following are the descriptions of the reference numerals: 101-data acquisition module, 102-data transmission and storage module, 103-data analysis and decision module, 104-intelligent control module, 105-user interaction and monitoring module; 201-air temperature and humidity sensor, 202-soil moisture sensor, 203-light sensor, 204-rainfall sensor, 205-camera, 206-weather data acquisition software; 301-MySQL database, 302-Yam water, fertilizer and pesticide precision control model, 303-Historical data acquisition module, 304-Precision water control module, 305-Precision fertilizer control module; 306-Precision pesticide control module; 401-main controller, 402-flow sensor, 403-liquid level sensor, 404-stirring motor, 405-water pump, 406-fertilizer pump, 407-medicine pump, 408-valve. DETAILED DESCRIPTION

[0027] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0028] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0029] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0030] In the absence of conflict, the embodiments of the present invention and the features thereof may be combined with each other.

[0031] Example 1 Ginseng Figure 1 This embodiment provides a method for precise control of yam water, fertilizer and pesticide based on big data analysis, the method comprising: S1. Collect yam growth data through various collection devices, including but not limited to weather data, environmental data, and yam image data. The environmental data includes air temperature and humidity, soil temperature and humidity, light and rainfall.

[0032] Preferably, a collection interval time point is set. Before the collection interval time point is reached, each collection device is in a deep sleep state, and only the heartbeat packet is kept to keep the device online; data collection uses the MQTT protocol, and through the deep sleep and MQTT protocol heartbeat packet management mechanism, the deep sleep and data change exceeding the threshold active wake-up method greatly reduces the sensor power consumption.

[0033] When the set collection time point is reached, the sensor data collected at the current collection time point is compared with the data collected at the previous collection time point. When the data change exceeds the set difference threshold, the device will be woken up and the collected data will be uploaded; Specifically, the soil temperature and humidity sensor collects and transmits data every 10 minutes, the light sensor collects and transmits data every 5 minutes, the air temperature and humidity sensor collects and transmits data every 15 minutes, and the rainfall sensor records and transmits rainfall in real time during rainfall. The image acquisition device captures and uploads images of yam growth every 30 minutes. Sensors use the object model JSON format to connect to data acquisition devices. The object model JSON includes three types of JSON: attributes, services, and events. Attribute JSON is used to upload sensor data. Service JSON supports remote device upgrades, remote ad hoc networking, and remote parameter modifications. Event JSON is used for alarms, where the alarm threshold can be modified according to needs. When the environmental parameters detected by the sensor exceed the normal range, the corresponding device will issue an early warning message to alert the management personnel; By introducing the concept of physical models, the platform can remotely modify the sensor collection interval; it can enable sensor nodes to have automatic network configuration and automatic address allocation functions, and users only need to scan the device QR code to automatically configure the network for use, greatly improving the scalability and maintainability of the sensor network; when the sensor itself fails, the platform can also obtain the fault identification code in time and process it; when the environmental parameters detected by the sensor exceed the normal range, the user interaction and monitoring module will issue an early warning message to remind managers to take corresponding measures.

[0034] S2. Clean the collected data to remove abnormal data and classify and store the data; First, the original data is cleaned to eliminate abnormal data. The methods used include but are not limited to the mean filling algorithm and the isolation forest algorithm. Then, the data is classified, managed, and stored according to the data type and collection time. At the same time, the data transmission and storage module also obtains local weather data in real time through the relevant weather forecast platform interface, stores the relevant weather, rainfall, etc., and provides it to the data analysis and decision-making module for processing.

[0035] S3. Establish precise water control strategies, precise fertilizer control strategies, and precise pesticide control strategies; Establish a precise water management strategy to control the amount of watering in the current cycle based on the amount of water shortage. Establish a precise fertilizer management strategy to control the fertilizer supply in this cycle based on the historical amount of fertilizer applied and the fertilizer shortage judgment of the yam image. Establish a precise pesticide management strategy to issue an alarm based on the identification of pests and diseases in the yam image.

[0036] The precise water management and control strategy can be formulated based on the amount of water required for the current growth cycle of yam and the rainfall in the current cycle, that is, the amount of water to be irrigated in this cycle = the amount of water required in the current cycle - the rainfall in the current cycle. In this embodiment, preferably, the precise water management and control strategy introduces a periodic water balance model based on a historical curve, combines real-time data and image data visual feedback closed-loop control mechanism, and calculates the weather data Q1(t) in the cycle, the amount of water irrigated in this cycle Q2(t), the amount of evaporated water Q3(t), and the visual image feedback water shortage Qf(t) based on the watering data Q(in) in this cycle, and judges the current amount of water to be irrigated. The calculation formula for the amount of water to be irrigated in the current cycle of the precise water management and control strategy is: + dt(1) In formula (1), Q(out) is the amount of water that should be applied in this cycle, Q(in) is the amount of water applied in this cycle according to the historical curve, and t is the time of this cycle; i is the serial number of the water factor, Qi(t) is the amount of water applied on the day of the water factor, including Q1(t), Q2(t), and Q3(t), where Q1(t) is the weather data in the cycle, Q2(t) is the amount of water applied in this cycle, and Q3(t) is the amount of evaporated water; Ci is the weight of the water factor, including C1, C2, and C3, where C1 is the weight of the weather data in the cycle, C2 is the weight of the amount of water applied in this cycle, and C3 is the weight of the amount of evaporated water; Qf(t) is the amount of water shortage reported by the visual image, and Cf is the weight of the amount of water shortage reported by the visual image; In formula (1), Q(in) is the amount of watering in the current cycle of the historical curve. Based on the historical data, the daily watering amount calculation formula is P=P1+P2-P3, where P1 is the rainfall collected on the day, P2 is the amount of watering on the day, and P3 is the amount of evaporated water on the day. The watering amount calculation formula for the cycle is: Q(in)= )dt(2); In formula (1), Q1(t) is the expected rainfall in the weather forecast, and the calculation formula is: Q1(t)= - ), where M(t) is the expected rainfall in the future weather forecast within the period, and N(t) is the rainfall collected by the rain sensor on the past days within the period; In formula (1), Q2(t) is the amount of water applied during this period, which is directly obtained from the operation data; In formula (1), Q3(t) is the expected amount of evaporated water, and the calculation formula is: Q3(t)=k*f(Ta,Ha)*g(Ts,Hs)*h(I)*S(3); In formula (3), S is the benchmark evaporation water per mu. The method for obtaining it is to collect 1 square meter of land in the yam planting area every day, weigh it before and after 24 hours to obtain the daily evaporation water, take the average of the daily evaporation water during the yam growth period, and multiply the average by 666.67 to obtain the benchmark evaporation water per mu; k is a comprehensive correction factor that takes into account the impact of other factors on evaporation, such as land type and vegetation cover. Its value range is usually between (0.5-1.5) and is set according to the specific land and crop conditions; f(Ta,Ha) is the air temperature and humidity function, which is used to calculate the effect of air temperature and humidity on evaporation. It is expressed as: f(Ta,Ha)=(Ta+273.15) / (1-Ha / 100), where Ta is the air temperature, unit: ℃; Ha is the relative humidity, unit: %; g(Ts,Hs) is the soil temperature and humidity function, which is used to calculate the influence of air temperature and humidity on evaporation. It is expressed as: f(Ts,Hs)=(Ta+273.15) / (Hs+1), where Ts is the soil temperature, unit: ℃; Hs is the soil volumetric water content, unit: %; h(I) is the light intensity function used to calculate the effect of light on evaporation, expressed as: h(I) = I / I0, where I is the actual light intensity, in lx, and I0 is the reference light intensity value, in lx, which is the light intensity at noon on a sunny and cloudless day. Qf(t) represents the water deficit measured by visual image feedback. Based on the color changes of yam leaves, the color histogram of the leaf region is statistically analyzed to calculate the variance of the bright green portion of the green channel. This variance is then compared with historical image data. A color ratio threshold is set to determine whether the color features exceed the normal range. The yam water deficit is then determined, with seven levels classified: excess, normal, water deficit 1, water deficit 2, water deficit 3, water deficit 4, and water deficit 5. Images are categorized into seven levels based on different growth cycles, allowing real-time assessment of the current yam water deficit status in the area. The water deficit level is then comprehensively determined by comparing it with current sensor data. If the water deficit level is equal to or greater than level 3, an alarm is sent via the cloud platform, notifying the user of the abnormal yam status.

[0037] Furthermore, when the water shortage level is greater than or equal to level 3, an alarm message is sent via the cloud platform and the user is notified of the current abnormal state of the yam.

[0038] Furthermore, C1 is the weight of weather data within the period, C2 is the weight of watering within the period, and C3 is the weight of evaporated water. A relatively excellent weight combination can be directly calculated using the Design-Expert 10.0 software.

[0039] In this embodiment, the weights are preferably determined by an orthogonal rotation experiment. The orthogonal rotation experiment refers to arranging the experiment using an orthogonal rotation table to ensure that the weights are evenly distributed and have orthogonal rotation properties: Using L9 (3 4 ) Orthogonal rotation table, 9 experiments are conducted in each cycle, covering 4 factors and 3 levels each. The yam experiments in different cycles are independent of each other. The 4 factors are set as C1, C2, C3, and Cf, and the related benefits are set as y. According to the design principle of orthogonal rotation table, the benefit calculation formula is: y=β0+β1C1+β2C2+β3C3+β4C f +β 11 C1C1+β 22 C2C2+β 33 C3C3+β 44 C f C f +β12 C1C2+β 13 C1C3+β 14 C1C f +β 23 C2C f +β 24 C2C f +β 34 C3C f (4); In formula (4), β0, β1, β2, β3, β4, β 11 , β 22 , β 33 , β 44 , β 12 , β 13 , β 14 , β 23 , β 24 , β 34 is the undetermined regression coefficient. By substituting the relevant experimental data, the optimal regression coefficient is calculated according to the least square method, that is, the maximum limit meets the min The minimum regression coefficient value, among which the optimal regression coefficient β 0m , β 1m , β 2m , β 3m , β 4m , β 11m , β 22m , β 33m , β 44m , β 12m , β 13m , β 14m , β 23m , β 24m , β 34m To calculate the matrix formula (C T C)β=C T Y is obtained, C is the matrix of experimental data of orthogonal rotation table, C T is the transposed matrix of C, β is the optimal regression coefficient matrix, Y is the profit matrix in the orthogonal rotation table experimental data, yj is the profit of the orthogonal rotation experimental table, and ymax is the optimal profit obtained by substituting the optimal regression coefficient.

[0040] Significance testing is used to determine which weights have a significant impact on the final return and which factors can be ignored. Significance testing involves substituting different weights into a computer and using a lookup table to analyze the degree of impact of different weights on the final return.

[0041] The precise fertilizer control strategy can use YOLO V7 to identify the features of yam leaves and stems to determine whether the yam is lacking fertilizer and the type of fertilizer it is lacking. It can then set a fixed amount of fertilizer to be added at a time, and perform a test once per cycle. In this embodiment, preferably, the precise fertilizer control strategy first queries the historical fertilization data for the current cycle, then uses multi-feature fusion yam image recognition technology on the yam image to analyze whether the yam is lacking fertilizer, and sets the fertilizer supply amount for this cycle through closed-loop control. The multi-feature fusion yam image recognition technology includes leaf color analysis based on leaf RGB values, YOLO V7-based leaf and stem feature recognition, and comparison of the identified features with a feature library of normal growth and different fertility deficiency states to determine whether the yam is deficient in fertilizer and the type of fertilizer being deficient. Analysis of relevant nutrient deficiency colors and nutrient deficiency characteristics includes but is not limited to the following: 1) Nitrogen deficiency: leaves lose their green color, yellow parts increase, and leaves become thinner; 2) Phosphorus deficiency: leaves are dark green and stems are purple-red; 3) Potassium deficiency: yellowing of leaf edges, burn marks, and plant lodging; 4) Calcium deficiency: the edges of the leaves turn yellow and become transparent white, and white spots appear on the leaves 5) Magnesium deficiency causes clear reticular patterns on the leaves; 6) Iron deficiency: the leaves are generally yellow-white in color.

[0042] The precise fertilizer control strategy first performs YOLO V7 semantic segmentation on the images captured by the camera to cut out all leaf areas. RGB color analysis and feature analysis are then performed. When the leaf color analysis ratio is greater than the historical normal yam leaf color analysis ratio, and the feature statistics are greater than the normal yam feature statistics in the historical image data, fertilizer deficiency adjustment is performed. Leaf color analysis is to count the proportion of bright green, yellow, dark green, yellow-white, and white in the leaf area. Feature statistics are to identify leaf thinning, purple-red stems, burn marks, white spots, and reticular patterns, and perform feature statistics. The calculation formula is: (5); In formula (5), G(x) is the fertilizer deficiency index corresponding to yam, where G(1) is the nitrogen deficiency index, G(2) is the phosphorus deficiency index, G(3) is the potassium deficiency index, G(4) is the calcium deficiency index, G(5) is the magnesium deficiency index, and G(6) is the iron deficiency index. V(x) represents the corresponding characteristic number of the current yam, where V(1) is the leaf thinning characteristic number, V(2) is the stem purple-red characteristic number, V(3) is the burn mark characteristic number, V(4) is the white spot characteristic number, V(5) is the mesh pattern characteristic number, and V(6) is the overall yellow-white characteristic number of the leaf. s represents the characteristic number corresponding to V(x), and sm represents the corresponding V(x). The maximum number of features, n represents the area of ​​yam leaves in the current image; Vm(x) represents the number of features corresponding to yam in the historical image, where Vm(1) is the number of features of thinning leaves in the historical image, Vm(2) is the number of features of purple-red stems in the historical image, Vm(3) is the number of features of burn marks in the historical image, Vm(4) is the number of features of white spots in the historical image, Vm(5) is the number of features of mesh patterns in the historical image, Vm(6) is the number of features of yellow-white leaves as a whole in the historical image, l represents the feature number corresponding to Vm(x), lm represents the maximum number of features corresponding to Vm(x), and m represents the area of ​​yam leaves in the historical image.

[0043] Furthermore, the precise pesticide control strategy uses YOLO V7 to train a variety of pest and disease images, identify relevant pest and disease characteristics, and issue alarms for pest and disease. The pest and disease images include but are not limited to anthrax, brown spot, root rot, wilt, yam sawfly, white grubs, cutworms, and nematodes, totaling 1,300 images. The original pest and disease image dataset is preprocessed by contrast enhancement, brightness enhancement, and color enhancement before model training.

[0044] S4, users view and manually control the management in real time; Users can view weather data, environmental data, yam image data and alarm status in real time through the visual interface, and can manually control water management, fertilizer management and pesticide management.

[0045] Through the visual interface of the user interaction module, growers can view environmental data such as soil temperature and humidity, light, air temperature and humidity in real time, intuitively understand the growth image of the yam, and view the water, fertilizer and pesticide control plan generated by the system and the equipment operation status. If the growth of yam in a certain area is found to be abnormal, such as yellowing leaves or slow growth, the manual control function can be used to manually increase the amount of fertilizer or adjust the spray range for that area. The user interaction module has a data report generation function. Growers can generate data reports for different time periods such as days, weeks, and months according to their needs to summarize their planting experience and analyze problems and results in the planting process. The user interaction module also has an alarm function, which will sound an alarm when the yam growth status is abnormal or there are pests and diseases.

[0046] like Figure 2 The diagram of the water balance calculation of the periodic curve is shown as follows, which is a diagram of the water calculation part of the data analysis and decision-making module. First, a historical watering curve is constructed based on historical sensor data, historical weather data, historical growth image data, and historical operation data. According to the current time point, the historical water volume within the period should be determined based on the historical curve. Convolutional neural network operations are performed on the expected rainfall, the amount of water applied by the equipment, and the expected evaporation water. Based on feedback from the visual system, the relevant watering amount is intelligently adjusted to achieve water balance. The relevant water calculation formula is: + dt(6); In formula (6), Q(out) is the amount of water to be applied during the current cycle, Q(in) is the amount of water applied during the current cycle according to the historical curve, t is the time of the current cycle, i is the water factor number, Qi(t) is the amount of water applied on the current day, and includes Q1(t), Q2(t), and Q3(t). Q1(t) is the weather data for the current cycle, Q2(t) is the amount of water applied during the current cycle, and Q3(t) is the amount of evaporated water. Ci is the water factor weight, which includes C1, C2, and C3, where C1 is the weight of the weather data for the current cycle, C2 is the weight of the amount of water applied during the current cycle, and C3 is the weight of the amount of evaporated water. Qf(t) is the amount of water shortage as reflected by the visual image, and Cf is the weight of the amount of water shortage as reflected by the visual image. The correlation weights were determined using an orthogonal rotation experiment. Different weights were used for orthogonal comparisons of different batches of yam during the germination period, the vine growth period, the tuber expansion period, and the dormancy period. The final correlation factor weights were determined by maximizing the final profit level.

[0047] Taking the yam germination period as an example, yam seedlings were placed in stages in an artificial climate chamber during the germination period, with the incubation temperature maintained at 25°C. The fruits were harvested after the dormancy period. Yield per mu was calculated using the data proportionally. During the experiment, data such as rainfall, watering volume, and soil evaporation were provided by corresponding sensors. Yield was calculated based on the local market price of yam after harvest. Using a four-factor, three-level orthogonal rotation experiment for the yam germination period as an example, some of the designed final yield levels and related factor weight combinations are shown in Table 1.

[0048] Table 1: Final return level and relevant factor weight combination plan

[0049] The quadratic polynomial regression model of parameters such as C1, C2, C3, and Cf was established using Design-Expert10.0 software, and a relatively excellent weight combination was calculated.

[0050] In this embodiment, for example, the historical water volume is 5 cubic meters, the expected precipitation is 2 cubic meters, the watering has been 2 cubic meters, the evaporation is 1 cubic meter, the visual judgment water shortage level is 2, the cycle is set to 7 days, and according to the relevant weight factors, watering should be done within 7 days. =5+ - - - , it can be seen that the water shortage is more serious and 7.59 cubic meters of water should be applied within 7 days.

[0051] Example 2 On the other hand, this embodiment provides a control platform for realizing a method for accurately controlling yam water, fertilizer and pesticide based on big data analysis, such as Figure 3 Shown, including: The data acquisition module 101 comprises a sensor network, a high-definition camera and an online weather platform deployed in the yam planting area, and is used to obtain environmental data, image data and weather data. The environmental data, image data and weather data are transmitted to the data transmission and storage module 102 via various communication methods. The data transmission and storage module 102 is used to convert the protocol and interface of the data collected by the data acquisition module 101, transmit the data to the data analysis and decision module 103 via Ethernet, remove abnormal data from the data, use the mean filling algorithm to fill the removed data, and finally transmit it to the MySQL database for classification, management, and storage of the data according to data type and collection time; at the same time, it receives the water, fertilizer, and pesticide strategy generated by the data analysis and decision module 103 and sends it to the intelligent control module 104 to realize precise control of water, fertilizer, and pesticide; The data analysis and decision module 103 integrates historical sensor data, historical weather data, historical growth image data, and historical operation data to implement precise water control strategies, precise fertilizer control strategies, and precise pesticide control strategies. The data analysis and decision module also includes server software that reads data stored in the MySQL database and calculates and issues relevant control instructions based on relevant models, which are sent to the intelligent control module 104 through the data transmission and storage module 102. The intelligent control module 104 sends corresponding instructions to the irrigation, fertilization and spraying equipment according to the received control instructions. The intelligent control module 104 and the data transmission and storage module 102 communicate using Ethernet; The user interaction and monitoring module 105 is used for users to view data, manually operate, and issue real-time alarms. The user interaction and monitoring module 105 and the data transmission and storage module 102 communicate using Ethernet and 4G mobile phone communication networks.

[0052] Furthermore, the data acquisition module 101 includes an air temperature and humidity sensor 201, a soil temperature and humidity sensor 202, a light sensor 203, a rainfall sensor 204, a camera 205, and weather data acquisition software 206. Figure 4 As shown; multiple soil moisture sensors 202 are arranged in the yam planting area, one is installed at a set distance, and soil moisture data is collected in real time; a weather station is installed near the yam planting area, and the weather station is connected to sensors including air temperature and humidity sensor 201, soil temperature and humidity sensor 202, light sensor 203, and rainfall sensor 204; weather data is obtained from the Internet through the OpenWeatherMap API.

[0053] The soil moisture sensor 202 is used to collect soil data corresponding to the farmland area; the weather station is used to collect meteorological data corresponding to the yam planting area; the camera 205 is used to collect crop image data corresponding to the yam planting area, and the OpenWeatherMap API is used to obtain weather data, such as rainfall, snowfall, etc.

[0054] Furthermore, the data analysis and decision module 103 is as follows: Figure 5 As shown, it includes a historical data acquisition module 303, a yam water, fertilizer and medicine precision control model 302, and a MySQL database 301; the historical data acquisition module 303 is used to obtain historical sensor data, historical weather data, historical image data, and historical operation data and store them in the MySQL database 301; the MySQL database is used to store historical data and real-time acquired data; the yam water, fertilizer and medicine precision control model 302 includes a precision water control module 304, a precision fertilizer control module 305, and a precision pesticide control module 306; finally, combined with real-time sensor data and real-time images, it can control the intelligent control module to achieve relevant water, fertilizer and medicine precision control, namely watering, fertilizing, and spraying, such as controlling irrigation time, irrigation amount, fertilizer type, fertilizer amount, spraying time, spraying type, and spraying amount; Intelligent control module 104 is as follows Figure 6As shown, the system includes a main controller 401. According to various instructions, the main controller 401 controls a flow sensor 402, a liquid level sensor 403, a stirring motor 404, a water pump 405, a fertilizer pump 406, and a pesticide pump 407 to remotely control a valve 408. Valve 408 controls the flow rate in the water, fertilizer, and pesticide pipelines, and water pump 405 uses a frequency converter to further control the water volume. For example, when soil moisture falls below a threshold set by the model, the drip irrigation system is activated, and water pump 405 irrigates at a set flow rate. When the yam reaches a specific growth stage, the fertilizer pump 406 and water pump 405 precisely apply the appropriate amount of fertilizer after mixing and proportioning according to the model's calculations. When pests and diseases are detected, pesticide pumps 407 and water pump 405 mix and proportion the pesticides according to the set dosage and range. Operations on different plots can be controlled by remotely controlling valve 408. For example, if a pest and disease occurs in Plot 1, Plot 1 will be sprayed for 10 minutes.

[0055] The user interaction and monitoring module 105 is used to view weather data, environmental data, yam image data and alarm conditions in real time, and can manually control water management, fertilizer management and pesticide management.

[0056] Specifically, in late April, the camera in the data acquisition module captured images of yam, and the data analysis and decision module judged based on YOLO V7 that the yam had begun to enter the vine-spinning period. The vine-spinning period was determined to be 30-40 days, and one of the cycles was set to 7 days. The relevant historical watering curve was formed based on historical data. According to the historical curve, 5 cubic meters of water should be applied in this cycle, and 7.59 cubic meters of water should be applied according to the formula. According to historical records, fertilization operations were performed during this cycle, and 10 kg of urea was applied. The leaf color was analyzed through the RGB value of the leaves, and it was found that the leaves lost their green color and turned yellow. YOLO V7 recognized the thinning characteristics of the leaves and added 2 kg of urea. YOLO V7 identified the appearance of brown spots at the roots of the plants, and some plants expanded into brown rotten spots, which triggered an alarm to the user, reminding the user that some areas had poor drainage and too much soil moisture. After calculation by the data analysis and decision module, the following operations were performed: 1. Water 7.59 cubic meters within 7 days.

[0057] 2. Apply 12 kg of urea fertilizer within 7 days.

[0058] 3. Alarm the user to remind him that the drainage in a specific area is poor and the soil moisture is too high.

[0059] The intelligent control module receives the relevant control instructions and, based on them, averages watering, fertilizing, and spraying operations over a seven-day period. Each day within this cycle, based on changes in weather and environmental data, as well as visual feedback, the precision water and fertilizer control strategies adjust watering and fertilizing rates and update the relevant control instructions. After receiving an alarm on the mobile client, users can view relevant environmental data and manually drain specific areas.

[0060] Example 3 This embodiment further provides a computer-readable storage medium storing executable instructions, which, when executed, enable the machine to perform the above-mentioned photovoltaic module intelligent defect detection method.

[0061] Specifically, a system or device equipped with a readable storage medium can be provided, on which software program codes that implement the functions of any of the above-mentioned embodiments are stored, and a computer or processor of the system or device can read and execute instructions stored in the readable storage medium.

[0062] In this case, the program code itself read from the computer-readable medium can realize the function of any one of the above embodiments, and thus the computer-readable code and the computer-readable storage medium storing the computer-readable code constitute part of this specification.

[0063] Examples of readable storage media include floppy disks, hard disks, magneto-optical disks, optical disks (e.g., CD-ROMs, CD-Rs, CD-RWs, DVD-ROMs, DVD-RAMs, DVD-RWs, DVD-RWs), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, the program code may be downloaded from a server computer or a cloud via a communication network.

[0064] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0065] Obviously, the above embodiments of the present invention are merely examples for the purpose of clearly illustrating the technical solutions of the present invention, and are not intended to limit the specific implementation methods of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the claims of the present invention shall be included within the scope of protection of the claims of the present invention.

Claims

1. A method for precise control of yam water, fertilizer and pesticide based on big data analysis, characterized in that: The method comprises: S1. Collecting yam growth data through various collection devices, including but not limited to weather data, environmental data, and yam image data. The environmental data includes air temperature and humidity, soil temperature and humidity, light and rainfall; S2. Clean the collected data to remove abnormal data and classify and store the data; S3. Establish a precise water management strategy to control the amount of watering required for the current cycle based on the water shortage. Establish a precise fertilizer management strategy to control the amount of fertilizer supplied during the current cycle based on historical fertilizer application rates and fertilizer shortage determination based on yam images. Establish a precise pesticide management strategy to generate alarms based on pest and disease identification based on yam images. S4. Users can view weather data, environmental data, yam image data and alarm status in real time through the visual interface, and can manually control water management, fertilizer management and pesticide management.

2. The method for precise control of yam water, fertilizer and pesticide based on big data analysis according to claim 1, characterized in that: The precise water control strategy is based on the watering data Q(in) in this cycle, and calculates the weather data Q1(t) in the cycle, the amount of water applied Q2(t) in this cycle, the amount of evaporated water Q3(t), and the amount of water shortage Qf(t) fed back by visual images to determine the current amount of water to be applied. The calculation formula for the amount of water to be applied in the current cycle of the precise water control strategy is: + dt(1) In formula (1), Q(out) is the amount of water that should be applied in this cycle, Q(in) is the amount of water applied in this cycle according to the historical curve, and t is the time of this cycle; i is the serial number of the water factor, Qi(t) is the amount of water applied on the day of the water factor, including Q1(t), Q2(t), and Q3(t), where Q1(t) is the weather data in the cycle, Q2(t) is the amount of water applied in this cycle, and Q3(t) is the amount of evaporated water; Ci is the weight of the water factor, including C1, C2, and C3, where C1 is the weight of the weather data in the cycle, C2 is the weight of the amount of water applied in this cycle, and C3 is the weight of the amount of evaporated water; Qf(t) is the amount of water shortage reported by the visual image, and Cf is the weight of the amount of water shortage reported by the visual image; In formula (1), Q(in) is the amount of watering in the current cycle of the historical curve. Based on the historical data, the daily watering amount calculation formula is P=P1+P2-P3, where P1 is the rainfall collected on the day, P2 is the amount of watering on the day, and P3 is the amount of evaporated water on the day. The watering amount calculation formula for the cycle is: Q(in)= )dt(2); In formula (1), Q1(t) is the expected rainfall in the weather forecast, and the calculation formula is: Q1(t)= - ), where M(t) is the expected rainfall in the future weather forecast within the period, and N(t) is the rainfall collected by the rain sensor on the past days within the period; In formula (1), Q2(t) is the amount of water applied during this period, which is directly obtained from the operation data; In formula (1), Q3(t) is the expected amount of evaporated water, and the calculation formula is: Q3(t)=k*f(Ta,Ha)*g(Ts,Hs)*h(I)*S(3); In formula (3), S is the benchmark evaporation water per mu. The method for obtaining it is to collect 1 square meter of land in the yam planting area every day, weigh it before and after 24 hours to obtain the daily evaporation water, take the average of the daily evaporation water during the yam growth period, and multiply the average by 666.67 to obtain the benchmark evaporation water per mu; k is the comprehensive correction coefficient, which is set according to the specific land and crop conditions; f(Ta,Ha) is the air temperature and humidity function, expressed as: f(Ta,Ha)=(Ta+273.15) / (1-Ha / 100), where Ta is the air temperature and Ha is the relative humidity; g(Ts,Hs) is the soil temperature and moisture function, expressed as: f(Ts,Hs)=(Ta+273.15) / (Hs+1) where Ts is the soil temperature and Hs is the soil volumetric water content; h(I) is the light intensity function, expressed as: h(I)=I / I0, where I is the actual light intensity, unit: lx, and I0 is the reference light intensity value, unit: lx, which is the light intensity at noon on a sunny and cloudless day. Qf(t) is the amount of water shortage reported by visual images. Based on the color changes of yam leaves, the color histogram of the leaf area is statistically analyzed to calculate the variance of the bright green portion of the green channel. This is then compared with the variance of the bright green portion of the green channel in historical image data. A color ratio threshold is set to determine whether the color feature exceeds the normal range. The water shortage of the yam is determined and divided into seven levels: excessive, normal, water shortage level 1, water shortage level 2, water shortage level 3, water shortage level 4, and water shortage level 5.

3. The method for precise control of yam water, fertilizer and pesticide based on big data analysis according to claim 2, characterized in that: When the water shortage level is greater than or equal to level 3, an alarm message will be sent via the cloud platform to notify the user of the current abnormal state of the yam.

4. The method for precise control of yam water, fertilizer and pesticide based on big data analysis according to claim 2, characterized in that: C1 is the weight of weather data within the period, C2 is the weight of watering amount within the period, and C3 is the weight of evaporation water. The weights are determined using orthogonal rotation experiment: Using L9 (3 4 ) Orthogonal rotation table, 9 experiments are conducted in each cycle, covering 4 factors and 3 levels each. The yam experiments in different cycles are independent of each other. The 4 factors are set as C1, C2, C3, and Cf, and the related benefits are set as y. According to the design principle of orthogonal rotation table, the benefit calculation formula is: y=β0+β1C1+β2C2+β3C3+β4C f +b 11 C1C1+β 22 C2C2+β 33 C3C3+β 44 C f C f +b 12 C1C2+β 13 C1C3+β 14 C1C f +b 23 C2C f +b 24 C2C f +b 34 C3C f (4); In formula (4), β0, β1, β2, β3, β4, β 11 , β 22 , β 33 , β 44 , β 12 , β 13 , β 14 , β 23 , β 24 , β 34 is the undetermined regression coefficient. By substituting the relevant experimental data, the optimal regression coefficient is calculated according to the least square method, that is, The minimum regression coefficient value, Among them, the optimal regression coefficient β 0m , β 1m , β 2m , β 3m , β 4m , β 11m , β 22m , β 33m , β 44m , β 12m , β 13m , β 14m , β 23m , β 24m , β 34m To calculate the matrix formula (C T C)β=C T Y is obtained, C is the matrix of experimental data of orthogonal rotation table, C T is the transposed matrix of C, β is the optimal regression coefficient matrix, Y is the profit matrix in the orthogonal rotation table experimental data, yj is the profit of the orthogonal rotation experimental table, and ymax is the optimal profit obtained by substituting the optimal regression coefficient.

5. The method for precise control of yam water, fertilizer and pesticide based on big data analysis according to claim 1, characterized in that: The precise fertilizer management strategy first queries the historical fertilization data in the current cycle, then uses the yam image recognition technology with multi-feature fusion to analyze the yam fertilizer deficiency, and sets the fertilizer supply in this cycle through closed-loop control; The multi-feature fusion yam image recognition technology includes leaf color analysis based on leaf RGB values, YOLO V7 recognition of yam leaf and stem features, and comparison of the identified features with a feature library of normal growth and different fertility deficiency states to determine whether the yam is lacking fertilizer and the type of fertilizer being lacking. The precise fertilizer control strategy first performs YOLOV7 semantic segmentation on the images captured by the camera to cut out all leaf areas, then performs RGB color analysis and feature analysis. When the leaf color analysis ratio is greater than the historical normal yam leaf color analysis ratio, and the feature statistics are greater than the normal yam feature statistics in the historical image data, fertilizer deficiency adjustment is performed; Leaf color analysis is to count the proportion of bright green, yellow, dark green, yellow-white, and white in the leaf area. Feature statistics are to identify leaf thinning, purple-red stems, burn marks, white spots, and reticular patterns, and perform feature statistics. The calculation formula is: （5）； In formula (5), G(x) is the fertilizer deficiency index corresponding to yam, where G(1) is the nitrogen deficiency index, G(2) is the phosphorus deficiency index, G(3) is the potassium deficiency index, G(4) is the calcium deficiency index, G(5) is the magnesium deficiency index, and G(6) is the iron deficiency index. V(x) represents the corresponding characteristic number of the current yam, where V(1) is the leaf thinning characteristic number, V(2) is the stem purple-red characteristic number, V(3) is the burn mark characteristic number, V(4) is the white spot characteristic number, V(5) is the mesh pattern characteristic number, and V(6) is the overall yellow-white characteristic number of the leaf. s represents the characteristic number corresponding to V(x), and sm represents the corresponding V(x). The maximum number of features, n represents the area of ​​yam leaves in the current image; Vm(x) represents the number of features corresponding to yam in the historical image, where Vm(1) is the number of features of thinning leaves in the historical image, Vm(2) is the number of features of purple-red stems in the historical image, Vm(3) is the number of features of burn marks in the historical image, Vm(4) is the number of features of white spots in the historical image, Vm(5) is the number of features of mesh patterns in the historical image, Vm(6) is the number of features of yellow-white leaves as a whole in the historical image, l represents the feature number corresponding to Vm(x), lm represents the maximum number of features corresponding to Vm(x), and m represents the area of ​​yam leaves in the historical image.

6. The method for precise control of yam water, fertilizer and pesticide based on big data analysis according to claim 3 is characterized in that: The precise pesticide control strategy uses YOLO V7 to train a variety of pest and disease images, identify relevant pest and disease characteristics, and issue alarms for pest and disease. The pest and disease images include but are not limited to anthrax, brown spot, root rot, wilt, yam sawfly, white grub, cutworm, and nematode. The original pest and disease image dataset is preprocessed by contrast enhancement, brightness enhancement, and color enhancement before model training.

7. The method for precise control of yam water, fertilizer and pesticide based on big data analysis according to claim 1, characterized in that: Set the collection interval time point. Before the collection interval time point is reached, each collection device is in deep sleep state, and only the heartbeat packet is kept to keep the device online; When the set collection time point is reached, the sensor data collected at the current collection time point is compared with the data collected at the previous collection time point. When the data change exceeds the set difference threshold, the device will be woken up and the collected data will be uploaded; Sensors use the object model JSON format to connect to data acquisition devices. The object model JSON includes three types of JSON: attributes, services, and events. Attribute JSON is used to upload sensor data. Service JSON supports remote device upgrades, remote ad hoc networking, and remote parameter modifications. Event JSON is used for alarms, where the alarm threshold can be modified according to needs. When the environmental parameters detected by the sensor exceed the normal range, the corresponding equipment will issue an early warning message to alert the management personnel.

8. A precise control platform for yam water, fertilizer and pesticide based on big data analysis, characterized by: The platform includes: A data acquisition module (101) is configured to include a sensor network, a high-definition camera, and a network weather platform deployed in the yam planting area, and is configured to acquire environmental data, image data, and weather data. The environmental data, image data, and weather data are transmitted to a data transmission and storage module (102) via a variety of communication methods. The data transmission and storage module (102) is used to convert the protocol and interface of the data collected by the data acquisition module (101), transmit the data to the data analysis and decision module (103) via Ethernet, remove abnormal data generated in the data, use the mean filling algorithm to fill the removed data, and finally transmit it to the MySQL database to classify and manage the data according to the data type and collection time. At the same time, it receives the water, fertilizer and medicine strategy generated by the data analysis and decision module (103) and sends it to the intelligent control module (104) to realize the precise control of water, fertilizer and medicine. The data analysis and decision module (103) integrates historical sensor data, historical weather data, historical growth image data, and historical operation data to implement precise water control strategies, precise fertilizer control strategies, and precise pesticide control strategies. The data analysis and decision module also includes server software that reads data stored in the MySQL database and calculates and issues relevant control instructions based on relevant models and sends them to the intelligent control module (104) through the data transmission and storage module (102). The intelligent control module (104) sends corresponding instructions to the irrigation, fertilization and spraying equipment according to the received control instructions. The intelligent control module (104) and the data transmission and storage module (102) communicate using Ethernet. The user interaction and monitoring module (105) is used for users to view data, perform manual operations, and issue real-time alarms. The user interaction and monitoring module (105) and the data transmission and storage module (102) communicate using Ethernet and 4G mobile phone communication networks.

9. The platform for precise management and control of yam water, fertilizer and pesticide based on big data analysis according to claim 8, characterized in that: The data acquisition module (101) includes an air temperature and humidity sensor (201), a soil temperature and humidity sensor (202), a light sensor (203), a rainfall sensor (204), a camera (205), and weather data acquisition software (206); a plurality of soil moisture sensors (202) are arranged in a yam planting area, one of which is installed at a set distance, to collect soil moisture data in real time; a weather station is installed near the yam planting area, and the weather station is connected to sensors including the air temperature and humidity sensor (201), the soil temperature and humidity sensor (202), the light sensor (203), and the rainfall sensor (204); weather data is obtained from the Internet via the OpenWeatherMap API; The data analysis and decision module (103) includes a historical data acquisition module (303), a yam water, fertilizer and pesticide precision control model (302), and a MySQL database (301); the historical data acquisition module (303) is used to acquire sensor data, weather data, image data, and historical operation data over the years and store them in the MySQL database (301); the MySQL database (301) is used to store historical data and real-time acquired data; the yam water, fertilizer and pesticide precision control model (302) includes the implementation of a precision water control module (304), a precision fertilizer control module (305), and a precision pesticide control module (306); finally, in combination with real-time sensor data and real-time images, the intelligent control module is controlled to realize precision control of water, fertilizer and pesticide, i.e., watering, fertilizing, and spraying; The intelligent control module (104) includes a main controller (401), and the main controller (401) controls a flow sensor (402), a liquid level sensor (403), a stirring motor (404), a water pump (405), a fertilizer pump (406), and a medicine pump (407) according to different instructions to remotely control a valve (408); the valve (408) is used to control the flow of the water, fertilizer, and medicine pipeline, and the water pump (405) uses a frequency converter to further control the water volume; The user interaction and monitoring module (105) is used to view weather data, environmental data, yam image data and alarm conditions in real time, and can manually control water management, fertilizer management and pesticide management.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images