Farmland water and fertilizer coordination method based on edge computing
By deploying sensing terminals and edge nodes in farmland using edge computing for data processing and decision-making, the latency and security issues of traditional water and fertilizer management are solved, achieving efficient and secure coordinated control of water and fertilizer. This approach is suitable for large-scale farmland and remote environments with weak network connectivity.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HENAN HAOQI TECHNOLOGY CO LTD
- Filing Date
- 2026-03-01
- Publication Date
- 2026-05-29
AI Technical Summary
Traditional farmland water and fertilizer management relies on manual experience, which leads to problems such as unreasonable water and fertilizer ratios, delayed supply timing, serious resource waste, and environmental pollution. Furthermore, cloud computing is prone to control delays and data security risks in unstable network environments.
An edge computing-based method for coordinated water and fertilizer management in farmland is adopted. By deploying sensing terminals in the farmland to collect data in real time, preprocessing, data fusion, and lightweight decision-making are performed at the edge nodes to achieve local data processing and rapid issuance of control commands. Combined with a closed-loop feedback control mechanism, water and fertilizer supply is dynamically optimized.
It achieves synchronization and precision in water and fertilizer synergistic control, reduces network transmission pressure and storage costs, improves data security, reduces resource waste, and ensures the stability and adaptability of the system in unstable network environments. It is suitable for large-scale farmland and remote weak network scenarios.
Smart Images
Figure CN122114518A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent farmland management technology, and more specifically, to a method for coordinated water and fertilizer management in farmland based on edge computing. Background Technology
[0002] Farmland water and fertilizer management is a core element in ensuring high crop yields and efficient resource utilization. Traditional water and fertilizer management methods rely heavily on manual experience, leading to problems such as unreasonable water and fertilizer ratios, delayed supply timing, and serious waste of water and fertilizer resources. Furthermore, it easily causes ecological and environmental problems such as soil compaction and groundwater pollution. With the development of smart agriculture technology, cloud-based water and fertilizer coordinated control methods have emerged. These methods collect soil, weather, and crop status data through field sensors, upload them to a cloud server for data processing and decision-making, and then issue control commands to the water and fertilizer execution equipment.
[0003] Shortcomings of existing technology: First, the complex field environment and unstable network signals lead to high delays in data upload and command issuance, which can easily result in untimely water and fertilizer control and affect the synergistic effect. Second, the simultaneous upload of a large amount of raw sensing data to the cloud not only increases the pressure on network transmission but also raises the cost of data storage and poses a risk of data leakage. Third, cloud-based decision-making relies on a stable network connection. When there is a network outage or a weak network, the entire water and fertilizer control system will be paralyzed, unable to operate autonomously, and has poor adaptability.
[0004] To address the above problems, this invention proposes a solution. Summary of the Invention
[0005] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a farmland water and fertilizer coordination method based on edge computing, which solves the problems mentioned in the background art.
[0006] To achieve the above objectives, the present invention provides the following technical solution: The edge computing-based method for coordinated water and fertilizer management in farmland includes the following steps: Deploy sensing terminals in farmland areas to collect farmland soil parameters, meteorological parameters, and crop growth parameters in real time, forming multi-source raw sensing data; The multi-source raw sensing data is transmitted to the edge nodes, and the edge nodes preprocess the multi-source raw sensing data to obtain standardized processed data. The edge node has a built-in data fusion algorithm to perform fusion analysis on standardized data. It combines preset crop growth period parameters, soil basic parameters and environmental parameters to extract soil water and fertilizer demand characteristics, crop growth demand characteristics and environmental impact characteristics to obtain fused feature data. The edge node has a built-in lightweight water and fertilizer synergistic decision-making model. Based on the fused feature data combined with the preset water and fertilizer control threshold and crop water and fertilizer demand model, the optimal water and fertilizer synergistic control parameters are dynamically calculated. The optimal water and fertilizer synergistic control parameters are converted into control commands and sent to the water and fertilizer execution equipment in real time, so that the water and fertilizer execution equipment can operate according to the preset strategy. Real-time data on farmland status after water and fertilizer application is collected and transmitted to edge nodes. The edge nodes compare and analyze the data after application with preset target parameters and make corrections to the water and fertilizer application equipment based on the comparison and analysis results.
[0007] In a preferred embodiment, the sensing terminal includes a soil sensor, a weather sensor, and a crop growth sensor.
[0008] In a preferred embodiment, the edge node preprocesses the multi-source raw sensing data as follows: By removing invalid data, eliminating abnormal fluctuations caused by random interference from sensors, removing abnormal data that exceeds physical meaning, and mapping parameters of different dimensions to the [0,1] interval, standardized data is obtained.
[0009] In a preferred embodiment, the process of fusing and analyzing standardized data is as follows: The mean values of farmland soil parameters, meteorological parameters, and crop growth parameters were fused using the following formula:
[0010] in It is the fusion value of the i-th type of parameter; It is the number of parameters under the i-th type of parameter; It is the i-th type of parameter. Sub-weights of each parameter; It is the i-th type of parameter. Normalized values of the parameters; Based on the fusion of each type of parameter, and taking into account the weights of each parameter, the initial fusion value is calculated by weighted summation, as shown in the following formula:
[0011] in This is the initial fusion value, ranging from [0,1]. It is the weight of the i-th type of parameter; It is the fusion value of the i-th type of parameter.
[0012] In a preferred embodiment, the crop growth period parameters include the crop growth period growth coefficient, the soil basic parameters include the soil type target fusion value, and the environmental parameters include the evaporation correction coefficient and the wind speed correction coefficient.
[0013] In a preferred embodiment, the process of extracting soil water and fertilizer requirements, crop growth requirements, and environmental impact characteristics to obtain fused feature data is as follows: The edge nodes automatically retrieve the corresponding crop growth period parameters, soil basic parameters, and environmental parameters based on the crop type and current growth period preset by the user, and extract soil water and fertilizer demand characteristics, crop growth demand characteristics, and environmental impact characteristics based on the crop growth period parameters, soil basic parameters, and environmental parameters. The formula for extracting soil water and fertilizer demand characteristics is as follows:
[0014] in These are the characteristic values of soil water and fertilizer requirements; It is the fusion value of soil type; It is the preset fusion value for soil-related targets; The formula for extracting crop growth requirement characteristics is as follows:
[0015] in These are characteristic values representing crop growth requirements; It is the fusion value of the crop category; It is the crop growth coefficient during the growing season; The formula for extracting environmental impact characteristics is as follows:
[0016] in These are environmental impact characteristic values; It is a fusion value related to meteorology; It is the evaporation correction factor. It is the wind speed correction factor; After extracting the three core features, the feature values are first validated for validity, and then organized into structured fused feature data.
[0017] In a preferred embodiment, the process of dynamically calculating the optimal water and fertilizer synergistic control parameters is as follows: The edge node reads the pre-deployed lightweight water and fertilizer collaborative decision-making model, verifies the model's integrity, and confirms that the model weights and activation function parameters are not missing. After the verification is passed, the pre-trained model weights that match the current plot are loaded, and the model inference parameters are initialized. Edge nodes read the fusion feature data output from the preceding data fusion step, perform validity checks on the fusion feature data, and correct any feature value that exceeds the [0,1] interval by using a pruning method; if there are missing values, they are supplemented by the mean of features in the same dimension; the verified fusion feature data is normalized and corrected twice, and the corrected data is used to obtain the model input vector. Edge nodes read preset water and fertilizer control thresholds from the local parameter library, convert the water and fertilizer control thresholds into model inference constraints, and clarify the upper and lower bounds of the parameters; Load a crop water and fertilizer requirement model that matches the current crop and growth stage. This model includes the basic water and fertilizer requirements for different growth stages; the optimal ratio of nitrogen, phosphorus, and potassium; correction coefficients for temperature, light, and evaporation; and crop water and fertilizer sensitivity coefficients. Transform crop water and fertilizer demand models into reasoning rules or mathematical expressions and embed them into the decision-making process; The edge nodes use a lightweight water and fertilizer synergy decision-making model, which takes the model input vector as input and outputs candidate water and fertilizer synergy control parameters. The optimization objectives are set as minimizing the deviation between soil water and fertilizer supply and demand, achieving optimal water and fertilizer conservation, and minimizing energy consumption. The weighted scoring method is used to integrate the optimization objectives with the preset weights to obtain a comprehensive score. The group with the highest comprehensive score among the candidate water and fertilizer synergistic control parameters is selected as the optimal water and fertilizer synergistic control parameters, and finally the optimal water and fertilizer synergistic control parameters are output.
[0018] In a preferred embodiment, the edge nodes use a lightweight water and fertilizer co-processing decision-making model. The process of outputting candidate water and fertilizer co-processing control parameters with the model input vector as input is as follows: The edge nodes use a lightweight water and fertilizer collaborative decision-making model, which takes fused feature data as input and combines the crop water and fertilizer demand model to obtain preliminary water and fertilizer demand predictions under the constraint of water and fertilizer control threshold. The water and fertilizer control threshold is used to perform interval cropping and safety constraints on the preliminary water and fertilizer demand forecast. If the preliminary water and fertilizer demand forecast is greater than the upper limit of the threshold, the upper limit is taken as the preliminary water and fertilizer demand forecast. If the preliminary water and fertilizer demand forecast is less than the lower limit of the threshold, then the lower limit shall be taken as the preliminary water and fertilizer demand forecast. If the preliminary water and fertilizer demand forecast exceeds the safe range, it is marked and the constrained parameters are corrected based on the crop water and fertilizer demand model, and candidate water and fertilizer synergistic control parameters are output.
[0019] In a preferred embodiment, the optimal water and fertilizer synergistic control parameters are converted into control commands and sent to the water and fertilizer execution equipment in real time. The operation of the water and fertilizer execution equipment is controlled as follows: The edge node reads the optimal water and fertilizer synergistic control parameters, performs a validity check on the read optimal parameters, and if the check fails, it immediately calls the candidate optimal parameters output in the previous step and re-executes the check. If two consecutive verifications fail, a local audible and visual alarm is triggered at the edge node, the command conversion process is paused, a fault log is recorded and pushed to the cloud backend. The edge node has a built-in instruction conversion module that converts the optimal parameters into digital control instructions one by one according to the communication protocol of the water and fertilizer execution equipment. After the instruction conversion is completed, the edge node performs CRC check on all control instructions. After the check passes, the instructions are grouped and sorted according to the supply sequence. After receiving control commands, the water and fertilizer execution equipment parses the command content and executes them step by step according to the grouped commands to achieve synchronous supply of water and nutrients to the farmland.
[0020] In a preferred embodiment, the process of modifying the water and fertilizer application equipment based on the comparative analysis results is as follows: After receiving the execution completion confirmation frame returned by the water and fertilizer execution device, the edge node sends a data collection command to the sensing terminal deployed in the farmland, triggering the sensing terminal to collect the current farmland status data; The edge nodes compare and analyze the received farmland status data with the preset target parameters, and calculate the deviation value of each parameter; If the deviation exceeds the preset allowable range, the parameters of the water and fertilizer synergy decision model will be adjusted, the optimal water and fertilizer synergy control parameters will be regenerated, and the parameters will be sent to the water and fertilizer execution equipment for correction. If the deviation is within the preset allowable range, the current water and fertilizer synergistic control parameters will be maintained, and the next round of data acquisition and synergistic control cycle will begin.
[0021] The technical effects and advantages of the edge computing-based farmland water and fertilizer coordination method of this invention are as follows: 1. This invention, relying on an edge computing architecture, migrates core tasks such as data preprocessing, multi-source data fusion, and water and fertilizer collaborative decision-making to edge nodes. This enables real-time local data processing and rapid issuance of control commands, with data processing latency controlled at the millisecond level. This effectively solves the problems of high latency and untimely control in cloud computing, ensuring the synchronization and accuracy of water and fertilizer collaborative control. By preprocessing and storing multi-source raw sensing data locally at edge nodes, only key data is uploaded to the cloud, significantly reducing network transmission pressure and cloud storage costs. Simultaneously, the use of encrypted transmission enhances data security and privacy, solving the problems of large data transmission volumes, high storage costs, and high data leakage risks associated with existing cloud computing.
[0022] 2. This invention employs a closed-loop feedback control mechanism to collect real-time farmland status data after water and fertilizer application, dynamically optimizing water and fertilizer synergistic control parameters to achieve dynamic adjustment of water and fertilizer supply. This ensures that the water and fertilizer ratio and timing of supply align with crop growth needs and environmental changes, effectively reducing water and fertilizer waste. Water savings can reach 20%-40%, and fertilizer savings can reach 15%-30%, while avoiding soil pollution and compaction, thus combining economic and ecological benefits. It achieves collaboration between autonomous decision-making at edge nodes and remote monitoring in the cloud. Edge nodes can operate autonomously in environments with no or weak network connectivity, ensuring the stability and adaptability of the water and fertilizer control system. It is suitable for various scenarios, including large-scale farmland and remote farmland with weak network connectivity. The cloud platform enables data backup, remote monitoring, and parameter adjustment, improving system operability and management efficiency. The lightweight data processing algorithm and water and fertilizer synergistic decision-making model built into the edge nodes are adapted to the computing power requirements of the edge nodes, requiring no high-performance hardware support, reducing system deployment costs, and facilitating large-scale application. Furthermore, the algorithm can be flexibly adjusted according to different crops and soil types, demonstrating strong versatility. Attached Figure Description
[0023] Figure 1 This is a schematic diagram of the structure of the edge computing-based farmland water and fertilizer coordination method of the present invention. Detailed Implementation
[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.
[0025] Example 1, Figure 1 This invention presents a method for coordinated water and fertilizer management in farmland based on edge computing.
[0026] Deploy sensing terminals in farmland areas to collect farmland soil parameters, meteorological parameters, and crop growth parameters in real time, forming multi-source raw sensing data; The sensing terminals are deployed in zones according to farmland plots, with one sensing terminal cluster (containing various types of sensors) set up every 5-10 mu. The sensors in the cluster adopt a star topology and are uniformly connected to the sensing gateway of the edge node. The sensing terminals include soil sensors, meteorological sensors, and crop growth sensors. The sensing gateway sends collection instructions to each sensor periodically, presets warning thresholds for each parameter, and immediately triggers real-time collection and reporting when the sensor detects that a parameter exceeds the threshold.
[0027] The multi-source raw sensing data is transmitted to the edge nodes, and the edge nodes preprocess the multi-source raw sensing data to obtain standardized processed data. The original sensing data from multiple sources is first cached in the sensing gateway, and then packaged into batches of 10 data items and transmitted to the edge nodes to avoid frequent communication and bandwidth consumption. Before transmission, the data is checked with CRC32 to ensure data integrity. Preprocessing includes four sub-steps: data cleaning, noise reduction, anomaly detection, and normalization. All of these steps are completed locally at the edge nodes. This includes removing invalid data (such as null values or garbled characters caused by sensor communication interruptions), eliminating abnormal fluctuations caused by random sensor interference, removing abnormal data that exceeds physical meaning, and mapping parameters of different dimensions to the [0,1] interval to obtain standardized processed data, which is convenient for subsequent fusion analysis.
[0028] The edge node has a built-in data fusion algorithm to perform fusion analysis on standardized data. It combines preset crop growth period parameters, soil basic parameters and environmental parameters to extract soil water and fertilizer demand characteristics, crop growth demand characteristics and environmental impact characteristics to obtain fused feature data. A weighted fusion algorithm is used to perform fusion analysis on standardized data, assigning weights according to the importance of parameters and fusing multi-dimensional data into a unified feature vector; First, perform mean fusion on each type of parameter, as shown in the following formula.
[0029] in It is the fusion value of the i-th type of parameter (e.g., soil type, crop type); It is the number of parameters under the i-th type of parameter; It is the i-th type of parameter. Sub-weights for each parameter (e.g., moisture weight of 0.3 in the soil class); It is the i-th type of parameter. Normalized values of the parameters; Based on the fusion of each type of parameter, and taking into account the weights of each parameter, the initial fusion value is calculated by weighted summation, as shown in the following formula:
[0030] in This is the initial fusion value, ranging from [0,1]. It is the weight of the i-th type of parameter; It is the fusion value of the i-th type of parameter.
[0031] It should be noted that the core of the fusion analysis is to "fit the actual farmland". The differences between "different crops, different growth stages and different soil types" cannot be reflected by the fusion of multi-source data alone. Therefore, it is necessary to combine the "crop growth stage parameters and soil basic parameters" preset locally at the edge nodes to correct the initial fusion values and provide a scenario-based basis for subsequent feature extraction. Crop growth period parameters include crop growth period growth coefficient; soil basic parameters include soil type target fusion value; environmental parameters include evaporation correction coefficient and wind speed correction coefficient. Based on the "corrected single-dimensional fusion value", and combined with the "target feature value" in the preset parameters, three core features are extracted: "soil water and fertilizer demand characteristics, crop growth demand characteristics, and environmental impact characteristics". These three features correspond to "supply capacity, actual demand, and interference factors" in water and fertilizer synergistic decision-making, and are the core inputs of the decision-making model. The extraction logic is as follows to ensure the relevance and interpretability of the features: The edge nodes automatically retrieve the corresponding crop growth period parameters, soil basic parameters, and environmental parameters based on the user-preset crop type and current growth period; Soil water and fertilizer demand characteristics reflect the gap between the current soil water and fertilizer supply capacity and the target supply capacity. The larger the gap, the stronger the demand, and the closer the characteristic value is to 0; the smaller the gap, the weaker the demand, and the closer the characteristic value is to 1. The extraction formula is as follows:
[0032] in These are soil water and fertilizer requirement characteristic values, ranging from [0,1]. It is the fusion value of soil type; It is the preset fusion value for soil-related targets; Crop growth requirements characteristics reflect the gap between the current crop growth state and the target growth state. Combined with the crop growth coefficient during its growth period, this quantifies the crop's actual water and fertilizer requirements. The closer the characteristic value is to 1, the stronger the demand; the closer it is to 0, the weaker the demand. The extraction formula is as follows:
[0033] in These are characteristic values representing crop growth requirements, ranging from [0,1]. It is the fusion value of the crop category; It is the crop growth coefficient during the growing season; Environmental impact characteristics reflect the degree of interference of current meteorological conditions on water and fertilizer supply. Combined with evaporation correction coefficients and wind speed correction coefficients, the impact of environmental factors on water and fertilizer consumption is quantified. The closer the characteristic value is to 1, the stronger the interference (the faster the water and fertilizer consumption); the closer it is to 0, the weaker the interference. The extraction formula is as follows:
[0034] in These are environmental impact characteristic values, ranging from [0,1]. It is a fusion value related to meteorology; It is the evaporation correction factor. It is the wind speed correction factor; After extracting the three core features, the feature values need to be validated first, and then organized into structured fusion feature data to ensure that the feature values are reasonable and in a uniform format, so that they can be directly input into the subsequent lightweight water and fertilizer synergistic decision-making model.
[0035] The edge node has a built-in lightweight water and fertilizer synergistic decision-making model. Based on the fused feature data combined with the preset water and fertilizer control threshold and crop water and fertilizer demand model, the optimal water and fertilizer synergistic control parameters are dynamically calculated. The edge node reads the pre-deployed lightweight water and fertilizer collaborative decision-making model from the local Flash / storage area, verifies the model integrity, and confirms that the model weights and activation function parameters are not missing. After the verification is passed, the pre-trained model weights that match the crop type and growth period of the current plot are loaded, and the model inference parameters are initialized. Edge nodes read the fusion feature data output from the preceding data fusion step, perform validity checks on the fusion feature data, and correct any feature value that exceeds the [0,1] interval by using a pruning method; if there are missing values, they are supplemented by the mean of features in the same dimension; the verified fusion feature data is normalized and corrected twice, and the corrected data is used to obtain the model input vector. Edge nodes read preset water and fertilizer control thresholds from the local parameter library, convert the water and fertilizer control thresholds into model inference constraints, and clarify the upper and lower bounds of the parameters; Load a crop water and fertilizer requirement model that matches the current crop and growth stage. This model includes the basic water and fertilizer requirements for different growth stages; the optimal ratio of nitrogen, phosphorus, and potassium; correction coefficients for temperature, light, and evaporation; and crop water and fertilizer sensitivity coefficients (seedling stage / growth stage / flowering stage / fruiting stage). Transform crop water and fertilizer demand models into reasoning rules or mathematical expressions and embed them into the decision-making process; The edge nodes employ a lightweight water and fertilizer co-processing decision-making model. Using fused feature data as input, and constrained by water and fertilizer control thresholds, the model combines crop water and fertilizer demand to obtain preliminary water and fertilizer demand predictions. The preliminary water and fertilizer demand predictions are then pruned and constrained by the water and fertilizer control thresholds. If the preliminary water and fertilizer demand prediction is greater than the upper threshold, the upper threshold is used as the preliminary water and fertilizer demand prediction; if the preliminary water and fertilizer demand prediction is less than the lower threshold, the lower threshold is used as the preliminary water and fertilizer demand prediction; if the preliminary water and fertilizer demand prediction exceeds the safe range, it is marked, and the constrained parameters are corrected based on the crop water and fertilizer demand model. The model outputs candidate water and fertilizer co-processing control parameters that satisfy the constraints, align with crop requirements, and are suitable for the current farmland conditions. The optimization objectives are set as minimizing the deviation between soil water and fertilizer supply and demand, achieving optimal water and fertilizer conservation, and minimizing energy consumption. The weighted scoring method is used to integrate the optimization objectives with the preset weights to obtain a comprehensive score. The group with the highest comprehensive score is selected as the optimal water and fertilizer synergistic control parameters, and finally the optimal water and fertilizer synergistic control parameters are output.
[0036] The optimal water and fertilizer synergistic control parameters are converted into control commands and sent to the water and fertilizer execution equipment in real time to control the operation of the water and fertilizer execution equipment and realize the synchronous supply of water and nutrients to farmland. The edge node reads the optimal water and fertilizer coordination control parameters output by the previous decision-making step from the local cache, performs a validity check on the read optimal parameters, and if the check fails, it immediately calls the candidate optimal parameters output by the previous step and re-executes the check; if the check fails twice in a row, the edge node triggers a local audible and visual alarm, pauses the instruction conversion process, records the fault log and pushes it to the cloud backend. The edge node has a built-in instruction conversion module that converts the optimal parameters into digital control instructions that the device can recognize, according to the communication protocol of the water and fertilizer execution device. After the instruction conversion is completed, the edge node performs CRC check on all control instructions to ensure that the instructions have no format errors and no data loss. After the check passes, the instructions are grouped and sorted according to the supply sequence to avoid instruction conflicts. After receiving control commands, the water and fertilizer execution equipment parses the command content and executes them step by step according to the grouped commands to achieve synchronous supply of water and nutrients to the farmland.
[0037] Real-time data on farmland status after water and fertilizer application is collected and transmitted to edge nodes. The edge nodes compare and analyze the farmland status data with preset target parameters and make corrections to the water and fertilizer application equipment based on the comparison and analysis results.
[0038] After receiving the "execution completed" confirmation frame from the water and fertilizer execution device, the edge node delays for 5 minutes (to avoid data fluctuations caused by insufficient water and fertilizer penetration) and sends a data collection command to the sensing terminal deployed in the farmland, triggering the sensing terminal to collect the current farmland status data. The edge node compares and analyzes the received farmland status data with the preset target parameters, calculates the deviation value of each parameter, and if the deviation value exceeds the preset allowable range, the parameters of the water and fertilizer synergy decision model are adjusted, the optimal water and fertilizer synergy control parameters are regenerated, and sent to the water and fertilizer execution equipment for correction; if the deviation value is within the preset allowable range, the current water and fertilizer synergy control parameters are maintained, and the next round of data acquisition and synergy control cycle begins.
[0039] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.
[0040] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.
[0041] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.
[0042] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.
[0043] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
[0044] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for coordinated water and fertilizer management in farmland based on edge computing, characterized in that, Includes the following steps: Deploy sensing terminals in farmland areas to collect farmland soil parameters, meteorological parameters, and crop growth parameters in real time, forming multi-source raw sensing data; The multi-source raw sensing data is transmitted to the edge nodes, and the edge nodes preprocess the multi-source raw sensing data to obtain standardized processed data. The edge node has a built-in data fusion algorithm to perform fusion analysis on standardized data. It combines preset crop growth period parameters, soil basic parameters and environmental parameters to extract soil water and fertilizer demand characteristics, crop growth demand characteristics and environmental impact characteristics to obtain fused feature data. The edge node has a built-in lightweight water and fertilizer synergistic decision-making model. Based on the fused feature data combined with the preset water and fertilizer control threshold and crop water and fertilizer demand model, the optimal water and fertilizer synergistic control parameters are dynamically calculated. The optimal water and fertilizer synergistic control parameters are converted into control commands and sent to the water and fertilizer execution equipment in real time, so that the water and fertilizer execution equipment can operate according to the preset strategy. Real-time data on farmland status after water and fertilizer application is collected and transmitted to edge nodes. The edge nodes compare and analyze the data after application with preset target parameters and make corrections to the water and fertilizer application equipment based on the comparison and analysis results.
2. The method for coordinated water and fertilizer management in farmland based on edge computing according to claim 1, characterized in that, The sensing terminal includes a soil sensor, a meteorological sensor, and a crop growth sensor.
3. The method for coordinated water and fertilizer management in farmland based on edge computing according to claim 2, characterized in that, The preprocessing process of edge nodes for multi-source raw sensing data is as follows: By removing invalid data, eliminating abnormal fluctuations caused by random interference from sensors, removing abnormal data that exceeds physical meaning, and mapping parameters of different dimensions to the [0,1] interval, standardized data is obtained.
4. The method for coordinated water and fertilizer management in farmland based on edge computing according to claim 3, characterized in that, The process of fusing and analyzing standardized data is as follows: The mean values of farmland soil parameters, meteorological parameters, and crop growth parameters were fused using the following formula: ； in It is the fusion value of the i-th type of parameter; It is the number of parameters under the i-th type of parameter; It is the i-th type of parameter. Sub-weights of each parameter; It is the i-th type of parameter. Normalized values of the parameters; Based on the fusion of each type of parameter, and taking into account the weights of each parameter, the initial fusion value is calculated by weighted summation, as shown in the following formula: ； in This is the initial fusion value, ranging from [0,1]. It is the weight of the i-th type of parameter; It is the fusion value of the i-th type of parameter.
5. The method for coordinated water and fertilizer management in farmland based on edge computing according to claim 4, characterized in that, The crop growth period parameters include the crop growth period growth coefficient, the soil basic parameters include the soil type target fusion value, and the environmental parameters include the evaporation correction coefficient and the wind speed correction coefficient.
6. The method for coordinated water and fertilizer management in farmland based on edge computing according to claim 5, characterized in that, The process of extracting soil water and fertilizer requirements, crop growth requirements, and environmental impact characteristics to obtain fused feature data is as follows: The edge nodes automatically retrieve the corresponding crop growth period parameters, soil basic parameters, and environmental parameters based on the crop type and current growth period preset by the user, and extract soil water and fertilizer demand characteristics, crop growth demand characteristics, and environmental impact characteristics based on the crop growth period parameters, soil basic parameters, and environmental parameters. The formula for extracting soil water and fertilizer demand characteristics is as follows: ； in These are the characteristic values of soil water and fertilizer requirements; It is the fusion value for soil types; It is the preset fusion value for soil-related targets; The formula for extracting crop growth requirement characteristics is as follows: ； in These are characteristic values representing crop growth requirements; It is the fusion value of the crop category; It is the crop growth coefficient during the growing season; The formula for extracting environmental impact characteristics is as follows: ； in These are environmental impact characteristic values; It is a fusion value related to meteorology; It is the evaporation correction factor. It is the wind speed correction factor; After extracting the three core features, the feature values are first validated for validity, and then organized into structured fused feature data.
7. The method for coordinated water and fertilizer management in farmland based on edge computing according to claim 6, characterized in that, The process of dynamically calculating the optimal water and fertilizer synergistic control parameters is as follows: The edge node reads the pre-deployed lightweight water and fertilizer collaborative decision-making model, verifies the model's integrity, and confirms that the model weights and activation function parameters are not missing. After the verification is passed, the pre-trained model weights that match the current plot are loaded, and the model inference parameters are initialized. Edge nodes read the fusion feature data output from the preceding data fusion step, and perform a validity check on the fusion feature data. If any feature value exceeds the [0,1] interval, a pruning method is used to correct it; if there are missing values, the mean of the features in the same dimension is used to supplement them. The validated fused feature data is normalized and corrected twice to obtain the model input vector. Edge nodes read preset water and fertilizer control thresholds from the local parameter library, convert the water and fertilizer control thresholds into model inference constraints, and clarify the upper and lower bounds of the parameters; Load a crop water and fertilizer requirement model that matches the current crop and growth stage. This model includes the basic water and fertilizer requirements for different growth stages; the optimal ratio of nitrogen, phosphorus, and potassium; correction coefficients for temperature, light, and evaporation; and crop water and fertilizer sensitivity coefficients. Transform crop water and fertilizer demand models into reasoning rules or mathematical expressions and embed them into the decision-making process; The edge nodes use a lightweight water and fertilizer synergy decision-making model, which takes the model input vector as input and outputs candidate water and fertilizer synergy control parameters. The optimization objectives are set as minimizing the deviation between soil water and fertilizer supply and demand, achieving optimal water and fertilizer conservation, and minimizing energy consumption. The weighted scoring method is used to integrate the optimization objectives with the preset weights to obtain a comprehensive score. The group with the highest comprehensive score among the candidate water and fertilizer synergistic control parameters is selected as the optimal water and fertilizer synergistic control parameters, and finally the optimal water and fertilizer synergistic control parameters are output.
8. The method for coordinated water and fertilizer management in farmland based on edge computing according to claim 7, characterized in that, The edge nodes use a lightweight water and fertilizer co-processing decision-making model. The process of outputting candidate water and fertilizer co-processing control parameters with the model input vector as input is as follows: The edge nodes use a lightweight water and fertilizer collaborative decision-making model, which takes fused feature data as input and combines the crop water and fertilizer demand model to obtain preliminary water and fertilizer demand predictions under the constraint of water and fertilizer control threshold. The water and fertilizer control threshold is used to perform interval cropping and safety constraints on the preliminary water and fertilizer demand forecast. If the preliminary water and fertilizer demand forecast is greater than the upper limit of the threshold, the upper limit is taken as the preliminary water and fertilizer demand forecast. If the preliminary water and fertilizer demand forecast is less than the lower limit of the threshold, then the lower limit shall be taken as the preliminary water and fertilizer demand forecast. If the preliminary water and fertilizer demand forecast exceeds the safe range, it is marked and the constrained parameters are corrected based on the crop water and fertilizer demand model, and candidate water and fertilizer synergistic control parameters are output.
9. The method for coordinated water and fertilizer management in farmland based on edge computing according to claim 8, characterized in that, The optimal water and fertilizer synergistic control parameters are converted into control commands and sent to the water and fertilizer execution equipment in real time. The operation of the water and fertilizer execution equipment is controlled as follows: The edge node reads the optimal water and fertilizer synergistic control parameters, performs a validity check on the read optimal parameters, and if the check fails, it immediately calls the candidate optimal parameters output in the previous step and re-executes the check. If two consecutive verifications fail, a local audible and visual alarm is triggered at the edge node, the command conversion process is paused, a fault log is recorded and pushed to the cloud backend. The edge node has a built-in instruction conversion module that converts the optimal parameters into digital control instructions one by one according to the communication protocol of the water and fertilizer execution equipment. After the instruction conversion is completed, the edge node performs CRC check on all control instructions. After the check passes, the instructions are grouped and sorted according to the supply sequence. After receiving control commands, the water and fertilizer execution equipment parses the command content and executes them step by step according to the grouped commands to achieve synchronous supply of water and nutrients to the farmland.
10. The method for coordinated water and fertilizer management in farmland based on edge computing according to claim 9, characterized in that, The process of modifying the water and fertilizer application equipment based on the comparative analysis results is as follows: After receiving the execution completion confirmation frame returned by the water and fertilizer execution device, the edge node sends a data collection command to the sensing terminal deployed in the farmland, triggering the sensing terminal to collect the current farmland status data; The edge nodes compare and analyze the received farmland status data with the preset target parameters, and calculate the deviation value of each parameter; If the deviation exceeds the preset allowable range, the parameters of the water and fertilizer synergy decision model will be adjusted, the optimal water and fertilizer synergy control parameters will be regenerated, and the parameters will be sent to the water and fertilizer execution equipment for correction. If the deviation is within the preset allowable range, the current water and fertilizer synergistic control parameters will be maintained, and the next round of data acquisition and synergistic control cycle will begin.