Comprehensive management method and system of agricultural Internet of Things

Abstract

The invention discloses a comprehensive management method and system for agricultural Internet of Things, and relates to the technical field of agricultural informatization, and the method comprises the steps: deploying a multi-type sensor array, collecting soil moisture content, environmental parameters and crop growth state data, employing dual-mode communication, and dynamically adjusting the sampling frequency according to the crop growth stage; filtering, calibrating and standardizing the acquired data; mining a multi-parameter association relationship based on a weighted fusion algorithm, a crop growth model and a machine learning algorithm; precise planting decisions such as irrigation, fertilization and pest control are generated, and priorities are sequenced according to crop growth critical periods; the intelligent execution terminal is linked to operate according to the instruction; the state after execution is monitored in real time and fed back to form closed-loop control; whole-process data is encrypted and stored in a distributed database, and tracing and trend analysis are supported. According to the invention, the precision, collaboration and intelligence level of agricultural management are improved, on-demand supply of water and fertilizer and advanced prevention and control of plant diseases and insect pests are realized, data safety is guaranteed, and the system is suitable for multi-scene planting.

Description

Technical Field

[0001] This invention relates to the field of agricultural information technology, and in particular to a comprehensive management method and system for agricultural Internet of Things (IoT). Background Technology

[0002] With the development of large-scale and intensive agriculture, traditional planting and management models rely on manual experience and judgment, resulting in prominent problems such as lagging monitoring, blind decision-making, and resource waste. During crop growth, the dynamic changes in parameters such as soil moisture, ambient temperature and humidity, and light intensity directly affect crop growth. However, manual monitoring is time-consuming and labor-intensive, with low data collection frequency and narrow coverage, making it difficult to accurately capture the spatiotemporal differences in the field microenvironment. For example, traditional irrigation relies on the subjective judgment of growers, which can easily lead to insufficient irrigation causing crop wilting due to water shortage, or excessive irrigation causing soil compaction and nutrient loss. In the process of fertilization, the nitrogen, phosphorus, and potassium ratio lacks scientific basis, which not only reduces fertilizer utilization but may also cause non-point source pollution in farmland.

[0003] While existing agricultural IoT systems have achieved basic data collection, their technical solutions have several limitations. In the data collection phase, the sensors are of limited types, often focusing on single-dimensional parameters of soil or environment, lacking direct monitoring of crop growth status. Furthermore, the sampling frequency is fixed and not dynamically adjusted according to crop growth stages, resulting in insufficient data for critical growth periods or redundant data for non-critical periods. Data preprocessing uses only simple filtering and lacks scientific outlier detection and correction mechanisms designed for the complex environmental interference in agricultural scenarios. Data deviations caused by sensor drift, extreme weather, and other factors directly affect the accuracy of subsequent decisions. In the data fusion and decision-making phase, multi-source data is fused using fixed weights, failing to explore the deep correlation between environment, crop, and soil. Decision generation relies on general models that are not adequately adapted to the growth characteristics of different crops. Pest and disease early warnings are based on single environmental parameters, lacking quantitative analysis and predictive capabilities, leading to delayed control measures.

[0004] In terms of execution control and data management, existing systems primarily rely on independent operation of intelligent devices, lacking a collaborative scheduling mechanism. This leads to conflicting action sequences for irrigation, fertilization, and plant protection equipment, impacting management effectiveness. Furthermore, the absence of a complete closed-loop control chain means that feedback on crop growth status after execution fails to effectively inform decision-making optimization, hindering continuous improvement in management precision. Data storage is fragmented, lacking encryption protection and traceability capabilities, compromising data security. The absence of a systematic trend analysis mechanism makes it difficult to support long-term planting planning with historical data, hindering the transformation of agricultural production from "experience-driven" to "data-driven." Simultaneously, existing systems suffer from poor communication compatibility, often employing a single communication method, which is prone to signal interruptions in complex farmland environments, resulting in delayed equipment response and further reducing the reliability and efficiency of integrated management. Summary of the Invention

[0005] This invention proposes a comprehensive management method and system for agricultural Internet of Things (IoT) to solve the problems mentioned in the prior art.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a comprehensive management method for agricultural Internet of Things, comprising the following steps: Multiple types of sensor arrays are deployed: soil sensors collect humidity, temperature, and EC values; environmental sensors collect light intensity, temperature, humidity, precipitation, and wind speed; crop sensors obtain plant height, leaf area index, and chlorophyll content through image acquisition and spectral analysis; dual-mode communication is adopted, and the sampling frequency is dynamically adjusted according to the crop growth stage. The collected data were preprocessed, and environmental noise was removed by an 8th-order Butterworth low-pass filter. Sensor drift calibration was then performed based on cubic polynomial fitting. The weighted fusion algorithm is used to fuse and analyze multi-source data. The weights are dynamically allocated according to the degree of influence of parameters on crop growth. The WOFOST crop growth model and random forest machine learning algorithm are combined to explore the quantitative correlation between environment, crop and soil. Precise planting decisions are generated based on the results of the fusion analysis, covering irrigation amount, fertilizer amount, pest and disease control plan, and light regulation strategy. The decision priority is sorted according to the key growth period of the crop. It links with the intelligent irrigation execution terminal to execute operations according to decision-making instructions; Real-time monitoring of crop growth status and environmental changes after execution; feedback data sampling frequency is consistent with that before execution; data is transmitted to the fusion analysis module to construct a closed-loop control link of decision-making, execution, and feedback. The entire process data is stored in a distributed database, which employs encrypted transmission and storage. The database supports multi-dimensional retrieval and has data traceability and growth trend analysis functions.

[0007] Furthermore, it also includes a crop water requirement precision prediction unit within the fusion analysis module. This prediction unit calculates water requirement by combining environmental parameters with the crop growth stage, specifically expressed as follows: ,in This refers to the daily water requirement of crops. For crop coefficients, For reference crop evapotranspiration, Leaf area index, Humidity influence coefficient The daily average relative humidity, This is the temperature influence coefficient. This refers to the daily average temperature.

[0008] Furthermore, it also includes an outlier detection unit for data preprocessing. The detection unit uses a box plot method combined with crop growth patterns to screen outlier data. The quartiles are calculated using a sliding window method, and the outlier threshold is [Q1-1.5×IQR, Q3+1.5×IQR]. For data exceeding the threshold, linear interpolation is used in combination with the mean of historical data of the same period for correction. If the corrected data still does not conform to the physiological growth pattern of crops, it is marked as invalid data and a sensor self-test command is triggered.

[0009] Furthermore, it also includes a pest and disease risk early warning unit in the decision generation module. This early warning unit quantifies the risk of pest and disease occurrence, specifically expressed as follows: ,in The risk level for the occurrence of pests and diseases. This represents the coefficient for the types of pests and diseases. This represents the cumulative rainfall over the past 7 days. This is the average temperature over the past 7 days. This is the ventilation impact factor. The average daily wind speed <3 indicates low risk, 3≤ <5 indicates medium risk, 5≤ <8 indicates high risk.

[0010] Furthermore, it also includes a multi-device coordination unit for the execution control module. The coordination unit dynamically allocates tasks based on the priority of decision instructions and the operating status of the equipment. The irrigation equipment adopts a dual-mode switching between drip irrigation and sprinkler irrigation, and the fertilization equipment supports independent control of the three channels of nitrogen, phosphorus and potassium. When the plant protection drone is operating, the flight altitude is adjusted in combination with wind speed data. The sunshade equipment is turned on 1 hour in advance when the average daily temperature is ≥35℃, and the opening degree is adjusted according to the light intensity.

[0011] Furthermore, it also includes an error correction unit for the status feedback module. The correction unit uses the absolute error method to calculate the deviation value Δ = |target value - actual value|, compares the target value of the decision instruction with the actual value of the feedback, and dynamically adjusts the subsequent decision parameters based on the PID algorithm. The proportional coefficient Kp = 0.3-0.5, the integral coefficient Ki = 0.1-0.2, and the derivative coefficient Kd = 0.05-0.1. When the deviation value is ≥10%, PID correction is started, and when the deviation value is <5%, correction is stopped.

[0012] Furthermore, it also includes a trend analysis unit in the data storage module. This analysis unit collects crop growth status data, environmental parameter change trends, and decision execution effects on a weekly, monthly, and quarterly basis. It generates trend curves using cubic polynomial fitting, combines them with the WOFOST crop growth model to predict the growth status of the next stage, and optimizes the planting density and variety selection for the next season based on the quarterly trend curves. This provides data support for long-term planting planning and also supports the retrieval of historical data by plot, crop type, and growth cycle, assisting in the digital accumulation of planting experience.

[0013] Furthermore, it includes the following modules: The data acquisition module integrates three types of sensors: soil, environment, and crop. The soil sensor measures humidity from 0-100%RH, temperature from -20℃ to 60℃, and EC value from 0-20mS / cm. The environment sensor measures light intensity from 0-200000 lux, temperature and humidity from -40℃ to 85℃ / 0-100%RH, and wind speed from 0-60m / s. The crop sensor has a 1920×1080 pixel visual sensor and a 0-100SPAD chlorophyll sensor. It supports LoRa / NB-IoT dual-mode communication. The preprocessing module is equipped with an 8th-order Butterworth low-pass filter circuit, a cubic polynomial fitting calibration algorithm, and a Z-score normalization module. It integrates an outlier detection and correction subunit and supports valid, invalid, and corrected data quality labeling. The fusion analysis module has built-in weighted fusion algorithm, random forest machine learning model and WOFOST crop growth model, supports crop water requirement prediction and pest and disease risk assessment calculation, and has the functions of multi-source data association mining and growth status prediction. The decision generation module includes a crop planting expert knowledge base and supports the generation of irrigation amount, fertilizer amount, pest and disease control plan, and light regulation strategy. The decision priority is dynamically adjusted and the instruction format conforms to the communication standard of agricultural intelligent equipment. The execution control module is equipped with a communication interface, links with intelligent devices, supports device status monitoring and fault alarm, and has a time-series collaborative control function. The status feedback module integrates the data acquisition and deviation calculation sub-modules. The acquisition parameters are consistent with those of the front end, and the absolute error method and PID error correction algorithm are used. The data storage module uses a distributed database, supports encrypted storage and blockchain notarization, and supports multi-dimensional retrieval and trend analysis.

[0014] Furthermore, it also includes an adaptive sampling unit for the data acquisition module, which adjusts the sampling frequency based on the crop water requirement prediction results; monitors the sensor battery power in real time through the voltage detection module; and when there is a conflict between the water requirement and battery power adjustment commands, the sampling frequency corresponding to the water requirement takes precedence.

[0015] Furthermore, it also includes a dynamic adjustment unit for the decision generation module. This unit dynamically adjusts the control plan parameters based on the pest and disease risk level R. When the risk level R ≥ 8, a high concentration of pesticide is used, and the spraying interval is shortened to 72 hours; when the risk level is 5-7, a conventional concentration of pesticide is used, and the spraying interval is 96 hours; when the risk level < 5, a low concentration of pesticide is used, and the spraying interval is 120 hours. The linkage environmental sensor monitors the temperature and humidity every 30 minutes after the pesticide is sprayed.

[0016] Compared with existing technologies, the beneficial effects of this invention are: This invention significantly improves the accuracy, coordination, and intelligence of agricultural IoT management through end-to-end technical solution optimization, providing scientific and efficient technical support for agricultural production. In the data acquisition and preprocessing stage, multi-type sensor arrays achieve comprehensive monitoring of soil, environmental, and crop parameters. Adaptive sampling frequency adjustment balances data accuracy and equipment endurance, while a scientific outlier detection and correction mechanism eliminates interfering data, ensuring high-quality input data and laying a solid foundation for subsequent analysis and decision-making.

[0017] This integrated analysis and decision generation process utilizes a multi-parameter coupled method for water demand prediction and pest and disease risk quantification. Breaking through the limitations of traditional general-purpose models, it achieves dynamic adaptation of decisions to crop growth needs and environmental changes. This makes irrigation, fertilization, and pest and disease control measures more targeted, meeting the individualized needs of crops at different growth stages while reducing resource waste and environmental pollution, thus improving the sustainability of agricultural production. The multi-equipment collaborative scheduling mechanism solves the timing conflict problem of traditional independent equipment operation. Through precise matching of action timing and parameters, it improves the effectiveness of decision execution, ensuring that management measures are effectively implemented.

[0018] The optimization of closed-loop control and data management enables the system to continuously improve itself. The deviation correction mechanism constantly reduces the discrepancies between decision-making and execution, improving management accuracy. Encrypted storage and blockchain evidence preservation ensure data security and traceability, while multi-dimensional trend analysis provides data support for long-term planting planning and helps to digitally accumulate planting experience. Overall, this invention constructs a complete collaborative link of "monitoring-analysis-decision-execution-feedback." The system has strong compatibility and stable and reliable communication, effectively breaking through the limitations of traditional agricultural management, promoting the transformation of agricultural production towards refinement, intelligence, and efficiency, improving crop yield and quality, and reducing planting costs and environmental pressure. Attached Figure Description

[0019] Figure 1 This is a schematic block diagram of an integrated management method for agricultural Internet of Things proposed in this invention; Figure 2 This is a schematic block diagram of a comprehensive management method for agricultural Internet of Things proposed in this invention; Figure 3A bar chart comparing key indicators for different wheat management methods; Figure 4 A map showing the area affected by the 7-day change in pest and disease risk levels. Figure 5 Create a scatter plot to match crop water requirements with irrigation volume. Detailed Implementation

[0020] 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0022] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified. Furthermore, the terms "installed," "connected," and "linked" should be interpreted broadly; for example, they may refer to a fixed connection, a detachable connection, or an integral connection; they may refer to a mechanical connection or an electrical connection; they may refer to a direct connection or an indirect connection through an intermediate medium; and they may refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances. The invention will now be described in further detail with reference to the accompanying drawings.

[0023] Reference Figures 1 to 5 A comprehensive management method for agricultural Internet of Things (IoT) includes the following steps: Multiple types of sensor arrays are deployed. Soil sensors are buried at a depth of 10-20cm to collect humidity, temperature, and EC values. Environmental sensors are installed at a height of 2m to collect light intensity, temperature, humidity, precipitation, and wind speed. Crop sensors obtain plant height, leaf area index, and chlorophyll content through image acquisition and spectral analysis. LoRa / NB-IoT dual-mode communication is adopted, with a transmission delay of ≤500ms. The sampling frequency is dynamically adjusted according to the crop growth stage: once every 2 hours during the seedling stage, once every 30 minutes during the growing stage, and once every 15 minutes during the flowering stage. The collected data is preprocessed by removing environmental noise through an 8th-order Butterworth low-pass filter (cutoff frequency 1Hz), and sensor drift calibration is completed based on cubic polynomial fitting. The data format is standardized using Z-score to achieve a unified standard for the data range (0-100). The weighted fusion algorithm is used to fuse and analyze multi-source data. The weights are dynamically allocated according to the degree of influence of parameters on crop growth. The WOFOST crop growth model and random forest machine learning algorithm are combined to explore the quantitative correlation between environment, crop and soil. Based on the results of the fusion analysis, precise planting decisions are generated, covering irrigation amount (dynamically calculated according to crop water requirements), fertilizer amount (the ratio of nitrogen, phosphorus and potassium is determined based on the soil EC value and the crop nitrogen, phosphorus and potassium absorption coefficient), pest and disease control plan (the type of pesticide is matched with the type of pest and disease, and the spraying time avoids the high temperature and strong light period), and light regulation strategy (the opening and closing degree of the shade canopy is controlled according to the light intensity level). The decision priority is sorted according to the key growth period of the crop, flowering period > growth period > seedling period. It links intelligent irrigation equipment, variable fertilizer applicators, plant protection drones, shade sheds and other execution terminals to execute operations according to decision commands, with irrigation volume control accuracy ≤ ±5%, fertilizer volume control accuracy ≤ ±3%, and pesticide spraying uniformity ≥ 90%; Real-time monitoring of crop growth status and environmental changes after execution; feedback data sampling frequency remains consistent with that before execution; data is transmitted to the fusion analysis module to construct a closed-loop control link of decision-making, execution, and feedback. The entire process of data (collected data, preprocessing results, decision instructions, execution records, and feedback data) is stored in a distributed database, which is encrypted and transmitted and stored using AES-256. The database supports multi-dimensional retrieval by plot, time, and data type, and has data traceability and growth trend analysis functions.

[0024] This invention also includes a crop water requirement accurate prediction unit within a fusion analysis module. This prediction unit calculates water requirement by combining environmental parameters with the crop growth stage, specifically expressed as follows: ,in This represents the daily water requirement of crops (in mm). The crop coefficient is 0.3-0.5 during the seedling stage, 0.7-0.9 during the growing stage, and 1.0-1.2 during the flowering stage. For reference crop evapotranspiration (unit: mm / d, calculated based on the FAO-recommended Penman-Monteith formula). Leaf area index (dimensionless, calculated by semantic segmentation algorithm from images acquired by visual sensors). The humidity influence coefficient (values ​​range from 0.01 to 0.03). The daily average relative humidity (in %). The temperature influence coefficient (values ​​range from 0.02 to 0.04). The daily average temperature (in °C) is used to achieve precise matching between crop water demand and irrigation supply through multi-parameter coupling calculation, so that the irrigation amount and the actual water demand of crops are kept in dynamic balance.

[0025] This invention also includes an outlier detection unit for data preprocessing. This unit uses a box plot method combined with crop growth patterns to screen for abnormal data. The quartiles are calculated using a sliding window method with a window length of 7 days. The outlier threshold is [Q1 - 1.5 × IQR, Q3 + 1.5 × IQR], where IQR = Q3 - Q1, Q1 is the first quartile, and Q3 is the third quartile. For data exceeding the threshold, linear interpolation combined with the mean of historical data from the same period is used for correction. If the corrected data still does not conform to the crop's physiological growth patterns, it is marked as invalid data, triggering a sensor self-check instruction. The self-check instruction includes sensor calibration prompts and a data acquisition restart process, providing high-quality data input for subsequent decision generation.

[0026] This invention also includes a pest and disease risk early warning unit within the decision generation module. This early warning unit quantifies the risk of pest and disease occurrence, specifically expressed as follows: ,in The risk level of pest and disease occurrence (values ​​range from 0 to 10, with higher values ​​indicating higher risk). The coefficients for pest and disease species are: fungi 0.8-1.0, insects 0.6-0.8. The values ​​are preset based on common local pests and diseases and support remote updates. This represents the cumulative rainfall over the past 7 days (in mm). The average temperature over the past 7 days (in °C). The ventilation impact factor (value ranges from 0.005 to 0.01) The average daily wind speed (in m / s). <3 indicates low risk, 3≤ <5 indicates medium risk, 5≤ <8 indicates high risk. A value of ≥8 indicates extremely high risk. Early warning of risk is achieved by quantitatively correlating environmental parameters with the occurrence patterns of pests and diseases, and the warning information is synchronized to the planting management terminal.

[0027] This invention also includes a multi-device coordination unit for the execution control module. This coordination unit dynamically allocates tasks based on the priority of decision commands and the operating status of the equipment. The irrigation equipment adopts a dual-mode switching between drip irrigation and sprinkler irrigation, while the fertilization equipment supports independent control of the nitrogen, phosphorus, and potassium three channels. The execution interval between the two is set to ≥30 minutes to ensure step-by-step absorption of water and fertilizer. When the plant protection drone is operating, its flight altitude is adjusted based on wind speed data: 2-3m when the wind speed is ≤2m / s, and 3-5m when the wind speed is 2-4m / s. The flight altitude adjustment is constrained by a pesticide atomization particle size ≥100μm. The shade canopy equipment is opened one hour in advance when the average daily temperature is ≥35℃. The opening degree is adjusted according to the light intensity (100% opening degree when the light intensity is >150,000 lux, and 70% opening degree when the light intensity is 100,000-150,000 lux), achieving time-series coordination and parameter matching of the actions of each execution device.

[0028] This invention also includes an error correction unit for the status feedback module. The correction unit calculates the deviation value Δ = |target value - actual value| using the absolute error method, compares the target value of the decision instruction with the actual value of the feedback, and dynamically adjusts the subsequent decision parameters based on the PID algorithm. The proportional coefficient Kp = 0.3-0.5, the integral coefficient Ki = 0.1-0.2, and the derivative coefficient Kd = 0.05-0.1. When the deviation value is ≥10%, PID correction is started, and when the deviation value is <5%, correction is stopped. Through the closed-loop correction mechanism, the execution deviation is gradually reduced, so that the execution effect continuously approaches the decision target value, thereby improving the accuracy of comprehensive management.

[0029] This invention also includes a trend analysis unit within the data storage module. This unit statistically analyzes crop growth status data (plant height growth rate is calculated using month-on-month comparisons, and chlorophyll content changes are analyzed using difference analysis) weekly, monthly, and quarterly, along with trends in environmental parameter changes and decision-making effectiveness (irrigation utilization rate = actual crop water consumption / irrigation amount, fertilizer absorption rate = cumulative crop nitrogen, phosphorus, and potassium absorption / fertilizer amount). A trend curve is generated using cubic polynomial fitting, with a goodness-of-fit R0. 2 With a value of ≥0.9, the WOFOST crop growth model is used to predict the growth trend of the next stage. Based on the quarterly trend curve, the planting density and variety selection of the next season are optimized, providing data support for long-term planting planning. At the same time, it supports the retrieval of historical data by plot, crop type and growth cycle, which helps to digitally accumulate planting experience.

[0030] This invention includes the following modules: The data acquisition module integrates soil, environmental, and crop sensors. The soil sensor measures humidity from 0-100%RH with an accuracy of ±2%RH, temperature from -20℃ to 60℃ with an accuracy of ±0.5℃, and EC value from 0-20mS / cm with an accuracy of ±0.1mS / cm. The environmental sensor measures light intensity from 0-200000lux with an accuracy of ±5%FS, temperature and humidity from -40℃ to 85℃ / 0-100%RH with an accuracy of ±0.3℃ / ±2%RH, and wind speed from 0-60m / s with an accuracy of ±0.1m / s. The crop sensor has a 1920×1080 pixel resolution and a 30fps frame rate, while the chlorophyll sensor measures from 0-100SPAD with an accuracy of ±1SPAD. It supports LoRa / NB-IoT dual-mode communication. The LoRa modulation is adjustable from SF7 to SF12 with a communication distance ≥3km. The NB-IoT supports the Cat-NB1 protocol with a transmission latency ≤500ms, and the sampling frequency can be adjusted remotely. The preprocessing module is equipped with an 8th-order Butterworth low-pass filter circuit, a cubic polynomial fitting calibration algorithm, and a Z-score normalization module. It integrates outlier detection and correction subunits, adopts sliding window quartile calculation and linear interpolation correction logic, and has a data processing latency of ≤100ms. The output data format is uniformly JSON, and it supports data quality marking (valid / invalid / corrected). The fusion analysis module has a built-in weighted fusion algorithm (the dynamic weight allocation logic is built based on the analytic hierarchy process), a random forest machine learning model (100 decision trees, 10 layers deep), and a WOFOST crop growth model. It supports the calculation of crop water requirement prediction formula and pest and disease risk assessment formula, with a data processing capacity of ≥1000 records / second. It also has the functions of multi-source data association mining and growth status prediction. The decision generation module includes a crop planting expert knowledge base, covering planting parameter thresholds and management rules for more than 10 crops such as wheat, corn, rice, and tomatoes. It supports the generation of irrigation amount, fertilizer amount, pest and disease control plan, and light regulation strategy. The decision priority is dynamically adjusted according to the key growth period of crops, the decision output delay is ≤200ms, and the instruction format conforms to the communication standard of agricultural intelligent devices. The execution control module is equipped with a LoRa / NB-IoT communication interface, and the communication protocol is compatible with MQTT / HTTP. It can be linked with devices such as smart irrigation valves, variable fertilizer machines, plant protection drones, and shade canopy controllers. It supports real-time monitoring of device status (running / fault / offline) and fault alarms. The execution command response time is ≤50ms, and it has the function of coordinated control of device action sequence. The status feedback module integrates a data acquisition submodule and a deviation calculation submodule. The data acquisition submodule has the same parameters as the front-end data acquisition module. The deviation calculation adopts the absolute error method, supports the deployment of PID error correction algorithm, and the feedback data transmission delay is ≤150ms, providing real-time feedback input for closed-loop control. The data storage module uses the MongoDB distributed database, with a cluster deployment of 3 data nodes and 1 backup node. It supports encrypted data storage (AES-256) and blockchain notarization (consortium blockchain architecture, block generation interval of 10 minutes). The data storage capacity is ≥10TB. It supports multi-dimensional retrieval and trend analysis by land plot, time, and data type. The data backup frequency is once a day, and the backup data is retained for 1 year.

[0031] This invention also includes an adaptive sampling unit for the data acquisition module. This unit adjusts the sampling frequency based on the crop water requirement prediction results. When the water requirement W ≥ 8 mm / d, the sampling frequency is increased to once per 20 minutes; when the water requirement is 4-8 mm / d, it remains at once per 30 minutes; and when the water requirement is < 4 mm / d, it is reduced to once per hour. The voltage detection module monitors the sensor battery level in real time. If the voltage is below 3.6V, it is determined that the remaining power is < 20%, at which point the sampling frequency is reduced by 50%. When there is a conflict between the water requirement and battery level adjustment commands, the sampling frequency corresponding to the water requirement takes precedence, dynamically balancing data acquisition accuracy and device battery life, and extending the sensor's field operating time.

[0032] This invention also includes a dynamic adjustment unit for the decision generation module. This unit dynamically adjusts the control plan parameters based on the pest and disease risk level R. When the risk level R ≥ 8, a high-concentration pesticide (1.2 times the recommended conventional value in the pesticide instructions) is used, and the spraying interval is shortened to 72 hours; when the risk level is 5-7, a conventional concentration pesticide is used, and the spraying interval is 96 hours; when the risk level is < 5, a low-concentration pesticide (0.8 times the recommended conventional value in the pesticide instructions) is used, and the spraying interval is 120 hours. The linkage environmental sensor monitors the temperature and humidity every 30 minutes after the pesticide is sprayed. If the humidity is ≥ 85% and continues for more than 4 hours, the next spraying interval is automatically shortened by 20%, and the shortened spraying interval is no less than 48 hours, thus realizing the dynamic adaptation of the pest and disease control plan to environmental conditions.

[0033] The following two examples further illustrate the specific implementation of this system: Example 1: Implementation of Integrated Internet of Things Management for Tomato Cultivation in Facility Agriculture This embodiment is applied to a 10-acre multi-span greenhouse tomato planting scenario. The tomato variety is pink-fruited, with a growth cycle of 120 days, covering four stages: seedling stage, growth stage, flowering stage, and fruiting stage. The method and system of this invention achieve intelligent management throughout the entire cycle. The specific implementation process is as follows: 1. Sensor Deployment and Data Acquisition The data acquisition module deploys multiple types of sensor arrays according to crop growth needs. Soil sensors employ a three-needle structure, buried 15 cm deep, with one sensor placed every 50 square meters, for a total of 20 sensors. These sensors collect soil moisture, temperature, and EC values, with measurement ranges of 0-100%RH, -20℃ to 60℃, and 0-20 mS / cm, respectively, and accuracies of ±2%RH, ±0.5℃, and ±0.1 mS / cm, respectively. Environmental sensors are installed 2 meters high on the greenhouse pillars, with four sensors per greenhouse, for a total of eight. These sensors collect data on light intensity, air temperature and humidity, precipitation, and wind speed. Light intensity measurement range is 0-200,000 lux with an accuracy of ±5%FS; air temperature measurement range is -40℃ to 85℃ with an accuracy of ±0.3℃; air humidity measurement range is 0-100%RH with an accuracy of ±2%RH; precipitation measurement range is 0-50 mm / h with an accuracy of ±0.1 mm; and wind speed measurement range is 0-60 m / s with an accuracy of ±0.1 m / s. The crop sensors include a visual sensor and a chlorophyll sensor. The visual sensor is installed on the top beam of the greenhouse, with one sensor per 100 square meters, for a total of 10 sensors. It has a resolution of 1920×1080 pixels and a frame rate of 30 frames per second, and captures images of tomato plants by taking pictures from above. The chlorophyll sensor is a handheld portable sensor, which collects 20 samples from 20 plants per week at fixed points. It has a measurement range of 0-100 SPAD and an accuracy of ±1 SPAD.

[0034] The sensor employs LoRa / NB-IoT dual-mode communication. The LoRa modulation mode is set to SF9, with a communication distance of 3.2 kilometers. The NB-IoT uses the Cat-NB1 protocol, with a transmission latency of 420 milliseconds. The sampling frequency is dynamically adjusted according to the growth stage: once every 2 hours during the seedling stage (1-30 days), once every 30 minutes during the growth stage (31-60 days), once every 15 minutes during the flowering stage (61-80 days), and once every 30 minutes during the fruiting stage (81-120 days). Sensor data is transmitted to the local gateway in real time, and after being aggregated by the gateway, it is uploaded to the cloud server.

[0035] 2. Data preprocessing execution The preprocessing module performs multi-step processing on the collected data. Environmental noise removal uses an 8th-order Butterworth low-pass filter with a cutoff frequency of 1 Hz. Fluctuations in data such as wind speed and light intensity are smoothed. For example, wind speed data collected at a certain time are 2.3 m / s, 2.8 m / s, 1.9 m / s, and 2.5 m / s; after filtering, the output is 2.3 m / s, 2.4 m / s, 2.2 m / s, and 2.3 m / s. Sensor drift calibration is based on cubic polynomial fitting, with the fitting formula being... Based on calibration data from a standard laboratory environment, an output of 48 corresponds to 50% RH, 58 to 60% RH, and 68 to 70% RH. Solving for a, b, c, and d yields a = 0.0001, b = -0.002, c = 1.002, and d = -0.002. This is used to correct the humidity data collected on-site. Z-score standardization is performed using the formula... , where μ is the data mean and σ is the standard deviation. For example, the mean soil temperature data is 25℃ and the standard deviation is 3℃. A certain collected value of 28℃ is standardized to 1.0, so that the range of all data is unified to 0-100.

[0036] The outlier detection unit uses a box plot method with a sliding window of 7 days. For a given soil moisture data point, Q1=45, Q3=65, IQR=20, and the outlier threshold is 15 to 95. For data exceeding the threshold (88% RH), linear interpolation combined with the historical average of 62% RH is used for correction, resulting in a corrected value of 65% RH. If this still does not conform to the optimal humidity range of 60-70% RH for tomato growth, the data is marked as invalid, triggering a sensor self-test command. The sensor is then recalibrated and data acquisition resumes.

[0037] 3. Fusion Analysis and Decision Generation The fusion analysis module employs a weighted fusion algorithm, with weights determined using the analytic hierarchy process: soil moisture 0.3, temperature 0.2, light intensity 0.2, leaf area index 0.15, and EC value 0.15. Combining the WOFOST crop growth model with the random forest algorithm, using 100 decision trees at a maximum depth of 10 layers, it mines multi-parameter correlations.

[0038] The precise crop water requirement prediction unit uses a formula Where W is the daily water requirement of the crop in mm, and k1 is the crop coefficient. For reference, crop evapotranspiration is measured in mm / d. Leaf area index is dimensionless, k2 is the humidity influence coefficient, and RH is the daily average relative humidity in percentage. This is the temperature influence coefficient. The unit for daily average temperature is °C. Tomato flowering period. =1.1, ET0 is calculated to be 5.2 mm / d based on the Penman-Monteith formula. The image acquired through a visual sensor was calculated to be 3.5 using a semantic segmentation algorithm. =0.02, RH=75%, k3=0.03, T=28℃, substituting into the formula, we get W=1.1×5.2×3.5×(1+0.02×75-0.03×28)=1.1×5.2×3.5×(1+1.5-0.84)=1.1×5.2×3.5×1.66=1.1×30.452=33.497mm / d, rounded to 33.5mm / d, corresponding to an irrigation amount of 335 cubic meters per mu (1mm=1 cubic meter per mu).

[0039] The pest and disease risk early warning unit targets fungi causing tomato gray mold, using a formula... ,in The risk level for pest and disease occurrence is set on a value from 0 to 10, where α is the pest and disease type coefficient. The unit for the cumulative rainfall over the past 7 days is mm. The average temperature over the past 7 days is expressed in degrees Celsius (°C). This is the ventilation impact factor. The daily average wind speed is expressed in m / s. α = 0.9, representing the cumulative precipitation over the past 7 days. =30mm, average temperature over the past 7 days =26℃, =0.008, daily average wind speed V=1.5m / s, substituting, we get R=0.9×30×26×(1-0.008×1.5)=0.9×30×26×(1-0.012)=0.9×30×26×0.988=0.9×770.88=693.792. After normalizing to 0-10, R=6.9, which is judged as high risk. The prevention and control plan is generated: use boscalid agent, the concentration is 1.0 times the conventional value, and the spraying time is 6-8 am the next day to avoid high temperature and strong light.

[0040] The decision generation module prioritizes flowering period > fruiting period > growth period > seedling period and outputs a comprehensive decision: irrigation amount of 335 cubic meters per mu, executed twice at 9 am and 4 pm; fertilizer application amount of 3:1:2 N:P, 20 kg per mu, 30 minutes apart from irrigation; shade canopy opening degree of 100% when light intensity is >150,000 lux, and 70% when light intensity is 100,000-150,000 lux; gray mold control is carried out according to the early warning plan.

[0041] 4. Execution control and status feedback The execution control module links the intelligent drip irrigation system, variable-rate fertilizer applicator, plant protection drone, and electric shade shed. The drip irrigation system uses pressure-compensated drippers with a flow rate of 2 liters per hour, irrigating twice according to the decision command, each time 167.5 cubic meters per acre, with a control accuracy of ±4%. The variable-rate fertilizer applicator supports independent control of nitrogen, phosphorus, and potassium channels, precisely outputting fertilizer according to the ratio, with a control accuracy of ±2%, and starts 30 minutes after irrigation to avoid water and fertilizer mixing affecting absorption. The plant protection drone operates at a wind speed of 1.5 m / s and a flight altitude of 2.5 meters, with a pesticide atomization particle size of 120 micrometers and a spray uniformity of 92%, covering the entire greenhouse area. The electric shade shed is adjustable according to light intensity levels, with a response time of 40 milliseconds and an opening control accuracy of ±5%.

[0042] The status feedback module collects data in real time after execution, with the sampling frequency consistent with that before execution. Two hours after irrigation, the soil moisture was measured at 65% RH, with a target value of 68% RH and a deviation of Δ = 68 - 65 = 3%. Since this deviation was less than 5%, PID correction was not initiated. After fertilization, the soil EC value was measured at 2.8 mS / cm, within the target range of 2.5-3.0 mS / cm. Three days after pest and disease control, monitoring of tomato leaves using crop sensors showed that the incidence of lesions decreased from 12% to 3%. The feedback data was transmitted to the fusion analysis module, updating the pest and disease risk level R to 2.3, indicating a low risk.

[0043] 5. Data storage and trend analysis The data storage module uses the MongoDB distributed database, with a cluster deployment of 3 data nodes and 1 backup node, providing a data storage capacity of 15TB. All data acquisition, preprocessing results, decision instructions, execution records, and feedback data throughout the entire process are transmitted and stored using AES-256 encryption. The blockchain consortium chain nodes include the planting base, agricultural technology department, and data center, with a block generation interval of 10 minutes to ensure data immutability. The trend analysis unit statistically analyzes the weekly growth rate of tomato plant height and changes in chlorophyll content, using a cubic polynomial fit to generate trend curves with a goodness of fit R0. 2 =0.93, the prediction results show that the fruiting period will begin 20 days after flowering, and the amount of potassium fertilizer applied should be increased appropriately.

[0044] Table 1 shows a comparison of the effects of different management methods in tomato cultivation:

[0045] Table 1 shows that traditional manual management relies on experience-based judgment, resulting in low matching rates between irrigation and fertilization (only 58% and 45% utilization rates, respectively). Delayed pest and disease control leads to an incidence rate of 18%, resulting in low yield and quality compliance rates. This invention, through precise water requirement prediction and pest and disease risk early warning, achieves on-demand water and fertilizer supply and proactive prevention, increasing irrigation utilization to 89%, fertilizer utilization to 78%, reducing pest and disease incidence to 4%, increasing yield by 22.7%, and improving quality compliance by 14.6%. These results fully demonstrate the significant advantages of this invention in resource conservation, risk control, and yield and quality improvement. Through data-driven closed-loop management, it breaks through the limitations of traditional experience-based planting, achieving refined and efficient production of greenhouse tomatoes.

[0046] Example 2: Implementation of Integrated IoT Management for Wheat Planting in Field Agriculture This embodiment is applied to a 100-mu (approximately 6.7 hectares) field wheat planting scenario. The wheat variety is a winter wheat variety with a growth cycle of 230 days, covering the sowing period, seedling stage, tillering stage, jointing stage, flowering stage, grain-filling stage, and maturity stage. The method and system of this invention achieve large-scale and intelligent management. The specific implementation process is as follows: 1. Sensor Deployment and Data Acquisition The data acquisition module deploys sensors according to a field grid layout. Soil sensors are arranged in a 50m x 50m grid, with a total of 40 sensors deployed at a depth of 12cm, collecting soil moisture, temperature, and EC values; parameters are consistent with Example 1. Environmental sensors are deployed along the perimeter and at key points within the field, with a total of 10 sensors installed at a height of 2 meters, collecting light intensity, temperature, humidity, precipitation, and wind speed. Data transmission uses LoRa / NB-IoT dual-mode communication, with a LoRa communication distance of 3.5 km and an NB-IoT transmission latency of 450 milliseconds. Crop sensors include visual sensors and spectral sensors. The visual sensors are installed on a mobile monitoring vehicle, patrolling twice a week to collect images of the wheat population; the spectral sensors are fixed on a drone, taking aerial photos once every two weeks to obtain leaf area index and chlorophyll content data; the measurement range and accuracy meet system requirements.

[0047] Sampling frequency is adjusted according to wheat growth stages: seedling stage (November to March) once every 2 hours; tillering stage (March to April) once every 30 minutes; jointing stage (April to May) once every 30 minutes; flowering stage (mid-May) once every 15 minutes; grain-filling stage (May to June) once every 30 minutes; and maturity stage (late June) once every 2 hours. Sensor data is aggregated by a local gateway and then uploaded to a cloud server via a 4G network to ensure stable data transmission.

[0048] 2. Data preprocessing execution The preprocessing module processes the data according to a standard procedure. An 8th-order Butterworth low-pass filter with a cutoff frequency of 1 Hz smooths the precipitation data. For example, precipitation data collected at a certain time period of 0.5 mm, 0.8 mm, 0.3 mm, and 0.6 mm becomes 0.5 mm, 0.6 mm, 0.5 mm, and 0.5 mm after filtering. A cubic polynomial fitting is used to calibrate the soil temperature sensor. Using standard environmental calibration data, 10℃ corresponds to an output of 10.2, 20℃ to 20.1, and 30℃ to 29.8. Solving for a, b, c, and d yields a = 0.00005, b = -0.001, c = 1.0005, and d = -0.002, which corrects the field-collected data. In Z-score standardization, the average soil moisture during the wheat seedling stage is 40%RH with a standard deviation of 5%RH. A collected value of 35%RH is standardized to -1.0, which is converted to a range of 40 (0-100).

[0049] The outlier detection unit has a sliding window of 7 days. For a given wind speed data point, Q1 = 1.2 m / s, Q3 = 2.5 m / s, and IQR = 1.3 m / s, the outlier threshold ranges from -0.75 to 4.45. For the 5.2 m / s data point exceeding the threshold, linear interpolation combined with the historical average of 1.8 m / s was used to correct it to 2.8 m / s, which aligns with the prevailing wind speed patterns in the field and is therefore retained as valid data.

[0050] 3. Fusion Analysis and Decision Generation The weighted fusion weights for the fusion analysis module are: soil moisture 0.35, temperature 0.2, light intensity 0.15, leaf area index 0.15, and EC value 0.15. The data associations are mined by combining the WOFOST model and the random forest algorithm.

[0051] The precise crop water requirement prediction unit uses a formula ,in The unit for daily water requirement of crops is mm. Here, ET0 is the reference crop evapotranspiration in mm / d, and LAI is the dimensionless leaf area index. Humidity influence coefficient The unit for daily average relative humidity is %. The temperature influence coefficient is given by T, where T is the daily average temperature in °C. (Wheat jointing stage) =0.8, ET0 calculated based on the Penman-Monteith formula is 4.8 mm / d, LAI=2.8, =0.015, RH=65%, =0.025, T=22℃, substituting, we get W=0.8×4.8×2.8×(1+0.015×65-0.025×22)=0.8×4.8×2.8×(1+0.975-0.55)=0.8×4.8×2.8×1.425=0.8×19.368=15.494mm / d, which is 154.94 cubic meters per mu. Rounding up to 155 cubic meters per mu, the irrigation will be carried out in three stages.

[0052] The pest and disease risk early warning unit targets fungal rust diseases in wheat, and the formula is... ,in The risk level for pest and disease occurrence is set on a value from 0 to 10, where α is the pest and disease type coefficient. The unit for the cumulative rainfall over the past 7 days is mm. The average temperature over the past 7 days is expressed in degrees Celsius (°C). This is the ventilation impact factor. The daily average wind speed is expressed in m / s. α = 0.95, representing the cumulative precipitation over the past 7 days. =25mm, average temperature over the past 7 days =20℃, =0.007, daily average wind speed =2.0m / s, substituting, we get =0.95×25×20×(1-0.007×2.0)=0.95×25×20×(1-0.014)=0.95×25×20×0.986=0.95×493=468.35, after normalizing to 0-10 =4.7, judged as medium risk, generate the following prevention and control plan: use triadimefon agent, at a concentration of 0.8 times the conventional value, and spray at an interval of 120 hours.

[0053] The decision generation module outputs decisions based on priority: flowering stage > jointing stage > grain filling stage > tillering stage > seedling stage > maturity stage. For example, irrigation amount is 155 cubic meters per mu, administered three times: at 10:00 AM, 3:00 PM, and 6:00 PM; fertilizer application rate is 15 kg per mu (NPK) with a nitrogen-phosphorus-potassium ratio of 4:1:1, with a 30-minute interval between applications and irrigation; the shade structure does not need to be turned on to maintain the field environment; rust disease control is implemented according to the medium-risk plan.

[0054] 4. Execution control and status feedback The execution control module links the intelligent sprinkler irrigation system, the towed variable displacement fertilizer applicator, and the plant protection machinery. The sprinkler irrigation system uses a central pivot type, with a spray radius of 150 meters, a flow rate of 5 cubic meters per hour, and a control accuracy of ±3%, irrigating in three stages according to the decision command. The variable displacement fertilizer applicator controls the application rate by adjusting the speed of the fertilizer roller, with independent adjustment of the nitrogen, phosphorus, and potassium channels, a control accuracy of ±2%, and starts 30 minutes after irrigation. The plant protection machinery operates at a wind speed of 2.0 m / s, a travel speed of 5 km / h, a spraying height of 3.0 meters, a pesticide atomization particle size of 110 micrometers, and a spray uniformity of 91%.

[0055] The status feedback module collected data after execution. Soil moisture after irrigation was 55%RH, the target value was 58%RH, and the deviation was Δ=3%. Since the deviation was less than 5%, PID correction was not initiated. After fertilization, the soil EC value was 2.2 mS / cm, within the target range of 2.0-2.5 mS / cm. Five days after pest and disease control, the incidence of wheat leaf rust decreased from 8% to 2%, and the feedback data updated the risk level to R=2.1, indicating a low risk. During the grain-filling stage, wheat chlorophyll content was found to be lower than the target value, with a deviation Δ=6%. PID correction was initiated, with Kp=0.4, Ki=0.15, and Kd=0.08. The fertilizer application rate was increased by 5%, and chlorophyll content recovered to the target range after 10 days.

[0056] 5. Data storage and trend analysis The data storage module uses the MongoDB distributed database with a storage capacity of 20TB, and employs AES-256 encryption for transmission and storage. The blockchain consortium nodes include planting cooperatives, agricultural technology extension centers, and data service providers, with a block generation interval of 10 minutes. The trend analysis unit provides monthly statistics on wheat plant height growth rate and grain-filling rate, using a cubic polynomial to fit the trend curve R0. 2 =0.92, predicting that the grouting period will last for 25 days, and the soil moisture should be maintained at 55-60%RH to provide a basis for subsequent management.

[0057] Table 2 compares the effects of different management methods in wheat cultivation:

[0058] Table 2 shows that traditional manual management in field wheat cultivation suffers from serious resource waste and lagging pest and disease control. Irrigation and fertilizer utilization rates are only 52% and 40%, respectively, and the incidence of pests and diseases is 15%, hindering yield and quality improvement. This invention, through multi-source data fusion analysis and precise decision-making, achieves on-demand water and fertilizer supply and early pest and disease control. Irrigation utilization rate increases to 85%, fertilizer utilization rate reaches 72%, the incidence of pests and diseases decreases to 3%, yield increases by 23.6%, and the test weight compliance rate increases by 12%. This result verifies the adaptability and effectiveness of this invention in large-scale field cultivation. Through closed-loop management throughout the entire process, it achieves cost reduction, efficiency improvement, quality enhancement, and increased income in wheat production, providing reliable technical support for the intelligent transformation of field agriculture.

[0059] Reference Figure 3 This chart quantifies and compares the core differences between traditional management and the management methods of this invention from three dimensions: resource utilization, risk prevention and control, and yield and quality, intuitively highlighting the technological advantages. The irrigation and fertilizer utilization rates of this invention are increased by 33 and 32 percentage points respectively compared to traditional management. This is primarily due to the synergistic effect of the water demand prediction formula and precise fertilization decisions, enabling on-demand water and fertilizer supply and avoiding resource waste. The incidence of pests and diseases has decreased from 15% to 3%, confirming the scientific validity of the pest and disease risk warning formula. Intervention 24-48 hours in advance effectively curbs the spread of diseases. The significant improvement in yield and bulk density compliance rates is a comprehensive reflection of the closed-loop management throughout the entire process of "monitoring-analysis-decision-execution-feedback." The chart data fully validates the adaptability and effectiveness of this invention in large-scale field agriculture, providing reliable technical support for cost reduction, efficiency improvement, quality enhancement, and income increase in agriculture, and promoting the transformation of agricultural production from "experience-driven" to "data-driven."

[0060] Reference Figure 4 This graph clearly shows the dynamic trend of the risk level of tomato gray mold. The risk level peaks at 6.9 on day 4, corresponding to the high-risk range. However, the pest and disease early warning mechanism of this invention initiates control measures on day 3, and the risk level gradually decreases thereafter. In traditional management models, pest and disease risks are often only discovered after symptoms appear, leading to delayed control and substantial damage to crops. This invention, however, calculates the risk level in real time by combining a risk assessment formula with environmental parameters such as temperature, humidity, and precipitation, predicting disease trends in advance and allowing sufficient time for the formulation of control measures. The rise and fall of the risk level in the graph intuitively reflects the full-chain control capability of "monitoring-early warning-intervention" of this invention, effectively reducing the degree of damage to crops from pests and diseases, ensuring production stability, and reducing pesticide overuse, which aligns with the concept of green agriculture development and verifies the scientific validity and practicality of the risk quantification formula.

[0061] Reference Figure 5The graph and fitted line visually demonstrate the strict linear correlation between crop water requirement and irrigation amount. The high degree of fit of the fitted line fully proves the accuracy and reliability of the water requirement prediction formula of this invention. In traditional irrigation management, the matching of water requirement and irrigation amount relies on the subjective experience of growers, which easily leads to problems such as insufficient irrigation causing crop wilting due to water shortage, or excessive irrigation causing soil compaction and nutrient loss. In contrast, this invention accurately calculates the crop water requirement at different growth stages through a multi-parameter coupling formula, and then calculates the water requirement according to the formula "1 mm / d water requirement corresponds to 10 m³ / d". 3 The conversion relationship of "irrigation volume per mu" is transformed into execution instructions for intelligent irrigation equipment, achieving precise matching of water and fertilizer supply and demand. Each data point in the chart corresponds to the supply and demand relationship of crops at different growth stages, reflecting the core feature of dynamic adjustment. This not only meets the personalized water needs of crop growth but also maximizes water conservation, demonstrating a key technology for water-saving and efficiency-enhancing agriculture. It also provides scientific data for the precise execution of intelligent irrigation equipment, promoting the transformation of irrigation management from "experience-driven" to "data-driven".

[0062] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A comprehensive management method for agricultural Internet of Things (IoT), characterized in that, Includes the following steps: Deploy multiple types of sensor arrays, with soil sensors collecting humidity, temperature, and EC values; Environmental sensors are installed to collect data on light intensity, temperature, humidity, precipitation, and wind speed; crop sensors acquire plant height, leaf area index, and chlorophyll content through image acquisition and spectral analysis; dual-mode communication is used, and the sampling frequency is dynamically adjusted according to the crop growth stage; The collected data were preprocessed, and environmental noise was removed by an 8th-order Butterworth low-pass filter. Sensor drift calibration was then performed based on cubic polynomial fitting. The weighted fusion algorithm is used to fuse and analyze multi-source data. The weights are dynamically allocated according to the degree of influence of parameters on crop growth. The WOFOST crop growth model and random forest machine learning algorithm are combined to explore the quantitative correlation between environment, crop and soil. Precise planting decisions are generated based on the results of the fusion analysis, covering irrigation amount, fertilizer amount, pest and disease control plan, and light regulation strategy. The decision priority is sorted according to the key growth period of the crop. It links with the intelligent irrigation execution terminal to execute operations according to decision-making instructions; Real-time monitoring of crop growth status and environmental changes after execution; feedback data sampling frequency is consistent with that before execution; data is transmitted to the fusion analysis module to construct a closed-loop control link of decision-making, execution, and feedback. The entire process data is stored in a distributed database, which employs encrypted transmission and storage. The database supports multi-dimensional retrieval and has data traceability and growth trend analysis functions.

2. The integrated management method for agricultural Internet of Things according to claim 1, characterized in that, It also includes a crop water requirement accuracy prediction unit within the fusion analysis module. This prediction unit calculates water requirement by combining environmental parameters with the crop growth stage, specifically expressed as follows: ,in This refers to the daily water requirement of crops. For crop coefficients, For reference crop evapotranspiration, Leaf area index, Humidity influence coefficient The daily average relative humidity, This is the temperature influence coefficient. This refers to the daily average temperature.

3. The integrated management method for agricultural Internet of Things according to claim 1, characterized in that, It also includes an outlier detection unit for data preprocessing. The detection unit uses box plot method combined with crop growth patterns to screen outlier data. The quartiles are calculated using sliding window method, and the outlier threshold is [Q1-1.5×IQR, Q3+1.5×IQR]. For data that exceeds the threshold, linear interpolation is used to correct it in combination with the mean of historical data of the same period. If the corrected data still does not conform to the physiological growth pattern of crops, it is marked as invalid data and a sensor self-test command is triggered.

4. The integrated management method for agricultural Internet of Things according to claim 1, characterized in that, It also includes a pest and disease risk early warning unit in the decision generation module. This early warning unit quantifies the risk of pest and disease occurrence, specifically expressed as follows: ,in The risk level for the occurrence of pests and diseases. This represents the coefficient for the types of pests and diseases. This represents the cumulative rainfall over the past 7 days. This is the average temperature over the past 7 days. This is the ventilation impact factor. The daily average wind speed, <3 indicates low risk, 3≤ <5 indicates medium risk, 5≤ <8 indicates high risk.

5. The integrated management method for agricultural Internet of Things according to claim 1, characterized in that, It also includes a multi-device collaboration unit for the execution control module. The collaboration unit dynamically allocates tasks based on the priority of decision instructions and the operating status of the equipment. The irrigation equipment adopts a dual-mode switching between drip irrigation and sprinkler irrigation. The fertilization equipment supports independent control of the three channels of nitrogen, phosphorus and potassium. When the plant protection drone is operating, the flight altitude is adjusted in combination with wind speed data. The sunshade equipment is turned on 1 hour in advance when the average daily temperature is ≥35℃, and the opening degree is adjusted according to the light intensity.

6. The integrated management method for agricultural Internet of Things according to claim 1, characterized in that, It also includes an error correction unit for the status feedback module. The correction unit calculates the deviation value Δ = |target value - actual value| using the absolute error method, compares the target value of the decision instruction with the actual value of the feedback, and dynamically adjusts the subsequent decision parameters based on the PID algorithm. The proportional coefficient Kp = 0.3-0.5, the integral coefficient Ki = 0.1-0.2, and the derivative coefficient Kd = 0.05-0.

1. When the deviation value is ≥10%, PID correction is started, and when the deviation value is <5%, correction is stopped.

7. The integrated management method for agricultural Internet of Things according to claim 1, characterized in that, It also includes a trend analysis unit for the data storage module. The analysis unit collects crop growth status data, environmental parameter change trends, and decision execution effects on a weekly, monthly, and quarterly basis. It generates trend curves by using cubic polynomial fitting, and predicts the growth status of the next stage by combining the WOFOST crop growth model. Based on the quarterly trend curves, it optimizes the planting density and variety selection for the next season. It also supports the retrieval of historical data by plot, crop type, and growth cycle.

8. A comprehensive management system for agricultural Internet of Things (IoT) according to any one of claims 1-7, characterized in that, Includes the following modules: The data acquisition module integrates three types of sensors: soil, environment, and crop. The soil sensor measures humidity from 0-100%RH, temperature from -20℃ to 60℃, and EC value from 0-20mS / cm. The environment sensor measures light intensity from 0-200000 lux, temperature and humidity from -40℃ to 85℃ / 0-100%RH, and wind speed from 0-60m / s. The crop sensor has a 1920×1080 pixel visual sensor and a 0-100SPAD chlorophyll sensor. It supports LoRa / NB-IoT dual-mode communication. The preprocessing module is equipped with an 8th-order Butterworth low-pass filter circuit, a cubic polynomial fitting calibration algorithm, and a Z-score normalization module. It integrates an outlier detection and correction subunit and supports valid, invalid, and corrected data quality labeling. The fusion analysis module has built-in weighted fusion algorithm, random forest machine learning model and WOFOST crop growth model, supports crop water requirement prediction and pest and disease risk assessment calculation, and has the functions of multi-source data association mining and growth status prediction. The decision generation module includes a crop planting expert knowledge base and supports the generation of irrigation amount, fertilizer amount, pest and disease control plan, and light regulation strategy. The decision priority is dynamically adjusted and the instruction format conforms to the communication standard of agricultural intelligent equipment. The execution control module is equipped with a communication interface, links with intelligent devices, supports device status monitoring and fault alarm, and has a time-series collaborative control function. The status feedback module integrates the data acquisition and deviation calculation sub-modules. The acquisition parameters are consistent with those of the front end, and the absolute error method and PID error correction algorithm are used. The data storage module uses a distributed database, supports encrypted storage and blockchain notarization, and supports multi-dimensional retrieval and trend analysis.

9. A comprehensive management system for agricultural Internet of Things according to claim 8, characterized in that, It also includes an adaptive sampling unit for the data acquisition module, which adjusts the sampling frequency based on the crop water requirement prediction results; monitors the sensor battery power in real time through the voltage detection module; and when there is a conflict between the water requirement and battery power adjustment commands, the sampling frequency corresponding to the water requirement takes precedence.

10. A comprehensive management system for agricultural Internet of Things according to claim 8, characterized in that, It also includes a dynamic adjustment unit for the decision generation module. This unit dynamically adjusts the control plan parameters based on the pest and disease risk level R. When the risk level R ≥ 8, a high concentration of pesticide is used and the spraying interval is shortened to 72 hours; when the risk level is 5-7, a conventional concentration of pesticide is used and the spraying interval is 96 hours; when the risk level is < 5, a low concentration of pesticide is used and the spraying interval is 120 hours. The linkage environmental sensor monitors the temperature and humidity every 30 minutes after the pesticide is sprayed.

Images