Intelligent irrigation and fertilization method and system based on feedback control

Abstract

The present invention provides an intelligent irrigation and fertilization method and system based on feedback control, which relates to the field of data processing technology. By constructing an edge water and fertilizer analysis node and an integrated control cloud fertilization and irrigation cloud edge architecture, nutrient demand deviation analysis is realized through the edge water and fertilizer analysis node, and edge water and fertilizer deviation parameters are obtained. The edge water and fertilizer deviation parameters transmitted back by the edge water and fertilizer analysis node are received by the integrated control cloud to generate a water and fertilizer feedback control strategy to feedback and adjust irrigation and fertilization in the crop planting area. The technical problem that the fertilization and irrigation control of large-scale crop planting in the prior art relies on manual experience, resulting in low fertilization and irrigation control accuracy and mismatch between resource supply and growth demand in the crop growth process is solved. The technical effect of targeted and precise fertilization and irrigation according to the crop growth situation is achieved, while ensuring that the resource supply in the crop growth process is adapted to the growth demand, effectively improving nutrient utilization and reducing resource waste.

Description

Technical Field

[0001] The invention relates to the technical field of data processing, and in particular to an intelligent irrigation and fertilization method and system based on feedback control. Background Art

[0002] With the development of modernization, the invention and large-scale use of irrigation and fertilization equipment have improved the efficiency of irrigation and fertilization of cultivated land. However, in the existing technology, the parameter setting of irrigation and fertilization equipment is based on a single dimension with reference to the type of crops planted. In large-scale, full-coverage irrigation and fertilization applications at multiple growth stages of crops, there is a lack of adaptability between actual fertilization and irrigation and plant growth conditions, resulting in the risk of restricted crop growth and even irreversible damage to the soil.

[0003] In summary, in the prior art, there are technical problems that fertilization and irrigation control for large-scale crop cultivation relies on manual experience, resulting in low precision of fertilization and irrigation control and mismatch between resource supply and growth demand during the crop growth process. Summary of the invention

[0004] The present application provides an intelligent irrigation and fertilization method and system based on feedback control, which is used to solve the technical problems in the prior art that the fertilization and irrigation control of large-scale crop planting relies on manual experience, resulting in low fertilization and irrigation control accuracy and mismatch between resource supply and growth demand in the crop growth process.

[0005] In view of the above problems, the present application provides an intelligent irrigation and fertilization method and system based on feedback control.

[0006] The first aspect of the present application provides an intelligent irrigation and fertilization method based on feedback control, the method comprising: interactively obtaining an effective adjustment threshold of a water-fertilizer integrated device, and effectively adjusting the irrigation and fertilization area according to the effective adjustment threshold to obtain a crop planting zoning; interactively obtaining a target planted crop, and synchronizing the target planted crop and the crop planting zoning to a pre-built crop monitoring and analysis network for sensor deployment analysis to generate a feedback control sensor array, wherein the feedback control sensor array comprises K sensor deployment nodes and K sensor device sets; deploying and positioning the K sensor deployment nodes in the crop planting zoning, and calling and deploying sensors with reference to the K sensor device sets to generate K edge sensor platforms; pre-building a standard cloud-edge architecture, wherein the standard cloud-edge architecture comprises an integrated control The cloud and K edge water and fertilizer analysis nodes complete the non-standard processing of the standard cloud-edge architecture by building a communication connection between the K edge sensing platforms and the K edge water and fertilizer analysis nodes, and obtain a fertilization and irrigation cloud-edge architecture; the K edge water and fertilizer analysis nodes receive and perform nutrient demand deviation analysis based on the K real-time soil nutrient information and K real-time crop images sent by the K edge sensing platforms, and obtain K edge water and fertilizer deviation parameters in combination with the target crops; the integrated control cloud of the fertilization and irrigation cloud-edge architecture receives the K edge water and fertilizer deviation parameters transmitted back by the K edge water and fertilizer analysis nodes for data fusion, and obtains a water and fertilizer feedback control strategy; the integrated control cloud controls the water and fertilizer integration device to perform feedback adjustment of irrigation and fertilization on the crop planting area with the water and fertilizer feedback control strategy.

[0007] The second aspect of the present application provides an intelligent irrigation and fertilization system based on feedback control, the system comprising: a planting zoning generation unit, used to interactively obtain an effective adjustment threshold of a water-fertilizer integrated device, and effectively adjust the irrigation and fertilization area according to the effective adjustment threshold to obtain a crop planting zoning; a sensor array generation unit, used to interactively obtain a target planted crop, and synchronize the target planted crop and the crop planting zoning to a pre-built crop monitoring and analysis network for sensor deployment analysis, and generate a feedback control sensor array, wherein the feedback control sensor array comprises K sensor deployment nodes and K sensor device sets; a sensor platform construction unit, used to deploy and position the K sensor deployment nodes in the crop planting zoning, and refer to the K sensor device sets for sensor call deployment, and generate K edge sensor platforms; a cloud-edge architecture generation unit, used to pre-build a standard cloud-edge architecture, wherein the standard The quasi-cloud-edge architecture includes an integrated control cloud and K edge water and fertilizer analysis nodes. By building a communication connection between the K edge sensing platforms and the K edge water and fertilizer analysis nodes, the non-standard processing of the standard cloud-edge architecture is completed to obtain a fertilization and irrigation cloud-edge architecture; a water and fertilizer deviation analysis unit is used for the K edge water and fertilizer analysis nodes to receive and perform nutrient demand deviation analysis based on the K real-time soil nutrient information and K real-time crop images sent by the K edge sensing platforms, and in combination with the target crops, to obtain K edge water and fertilizer deviation parameters; a feedback adjustment execution unit is used for the integrated control cloud of the fertilization and irrigation cloud-edge architecture to receive the K edge water and fertilizer deviation parameters transmitted back by the K edge water and fertilizer analysis nodes for data fusion to obtain a water and fertilizer feedback control strategy. The integrated control cloud controls the water and fertilizer integration device to perform feedback adjustment of irrigation and fertilization on the crop planting area according to the water and fertilizer feedback control strategy.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] The method provided in the embodiment of the present application obtains the effective adjustment threshold of the water-fertilizer integrated device through interaction, and effectively adjusts the irrigation and fertilization area according to the effective adjustment threshold to obtain the crop planting zoning; interactively obtains the target planted crops, and synchronizes the target planted crops and the crop planting zoning to a pre-built crop monitoring and analysis network for sensor deployment analysis to generate a feedback control sensor array, wherein the feedback control sensor array includes K sensor deployment nodes and K sensor device sets; deployment and positioning are performed in the crop planting zoning according to the K sensor deployment nodes, and sensor call deployment is performed with reference to the K sensor device sets to generate K edge sensor platforms; a standard cloud-edge architecture is pre-built, wherein the standard cloud-edge architecture includes an integrated control cloud and K edge water and fertilizer analysis nodes. point, by building the communication connection between the K edge sensing platforms and the K edge water and fertilizer analysis nodes, the non-standard processing of the standard cloud edge architecture is completed to obtain the fertilization and irrigation cloud edge architecture; the K edge water and fertilizer analysis nodes receive and perform nutrient demand deviation analysis based on the K real-time soil nutrient information and K real-time crop images sent by the K edge sensing platforms, and combine the target planted crops to obtain K edge water and fertilizer deviation parameters; the integrated control cloud of the fertilization and irrigation cloud edge architecture receives the K edge water and fertilizer deviation parameters communicated back by the K edge water and fertilizer analysis nodes for data fusion to obtain a water and fertilizer feedback control strategy, and the integrated control cloud controls the water and fertilizer integration device to perform feedback adjustment of irrigation and fertilization on the crop planting area with the water and fertilizer feedback control strategy. The technical effect of targeted and precise fertilization and irrigation according to the crop growth situation is achieved, while ensuring that the resource supply in the crop growth process adapts to the growth needs, effectively improving nutrient utilization and reducing resource waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A schematic diagram of the process flow of the intelligent irrigation and fertilization method based on feedback control provided in this application;

[0011] Figure 2 A flow chart of a water-fertilizer feedback control strategy in the intelligent fertigation method based on feedback control provided in this application;

[0012] Figure 3 A schematic diagram of the structure of the intelligent irrigation and fertilization system based on feedback control provided in this application.

[0013] Explanation of the accompanying drawings: planting zoning generation unit 1, sensor array generation unit 2, sensor platform construction unit 3, cloud-edge architecture generation unit 4, water and fertilizer deviation analysis unit 5, feedback regulation execution unit 6. DETAILED DESCRIPTION

[0014] The present application provides an intelligent fertilization method and system based on feedback control, which is used to solve the technical problem that the fertilization and irrigation control of large-scale crop planting in the prior art relies on manual experience, resulting in low fertilization and irrigation control accuracy and the mismatch between resource supply and growth demand during the crop growth process. The method achieves the technical effect of targeted and precise fertilization and irrigation according to the crop growth conditions, effectively improving nutrient utilization and reducing resource waste while ensuring that the resource supply during the crop growth process is adapted to the growth demand.

[0015] The acquisition, storage, use, and processing of data in the technical solution of the present invention are in compliance with relevant regulations.

[0016] Below, the technical solutions in the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments described herein. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. It should also be noted that, for the convenience of description, only the parts related to the present invention are shown in the accompanying drawings, rather than all of them.

[0017] Embodiment 1

[0018] like Figure 1 As shown, the present application provides an intelligent fertigation method based on feedback control, the method comprising:

[0019] A100: interactively obtaining an effective adjustment threshold of the water-fertilizer integrated device, and performing effective adjustment frame selection on the irrigation and fertilization area according to the effective adjustment threshold to obtain the crop planting zoning;

[0020] Specifically, in this embodiment, the water-fertilizer integrated device combines an irrigation system and a fertilizer application system, and can dissolve or mix an appropriate amount of fertilizer in the irrigation water during irrigation, and apply the water and fertilizer to the soil around the crop roots in an integrated manner through the irrigation system to improve the nutrient utilization efficiency and yield of the crop.

[0021] The effective adjustment threshold is the irrigable radiation range of the water-fertilizer integrated device. This embodiment effectively adjusts the irrigation and fertilization area according to the effective adjustment threshold to obtain the crop planting zoning. The crop planting zoning is the specific area and range for planting the target crops. This embodiment sets the crop planting zoning to ensure that crops are planted in areas that can be irrigated and fertilized, thereby ensuring crop growth efficiency and yield.

[0022] A200: interactively obtaining target crops, and synchronizing the target crops and the crop planting zones to a pre-built crop monitoring and analysis network for sensor deployment analysis, and generating a feedback control sensor array, wherein the feedback control sensor array includes K sensor deployment nodes and K sensor device sets;

[0023] In one embodiment, the target crop is obtained interactively, and the target crop and the crop planting area are synchronized to a pre-built crop monitoring and analysis network for sensor deployment analysis to generate a feedback control sensor array. The method step A200 provided in the present application also includes:

[0024] A210: interactively obtaining a plurality of crop planting interval thresholds and a plurality of crop monitoring interval thresholds, and performing associated storage of the plurality of crop planting interval thresholds and the plurality of crop monitoring interval thresholds based on a knowledge graph to generate a monitoring constraint analysis layer;

[0025] A220: interactively obtaining a plurality of sample sensor device sets of a plurality of sample crops, and performing associated storage of the plurality of sample crops and the plurality of sample sensor device sets based on a knowledge graph to generate a sensor constraint analysis layer;

[0026] A230: The monitoring constraint analysis layer and the sensing constraint analysis layer are arranged in parallel and the output ends of the monitoring constraint analysis layer and the sensing constraint analysis layer are connected to the input end of the pre-constructed monitoring array generation layer to complete the construction of the crop monitoring and analysis network;

[0027] A240: interactively obtain the target planting interval of the target planted crop, and synchronously input the target planting interval and the target planted crop into the monitoring constraint analysis layer and the sensing constraint analysis layer of the crop monitoring and analysis network, respectively, to obtain a target monitoring interval threshold and a target sensing device set;

[0028] A250: The monitoring array generation layer performs grid processing on the crop planting area according to the target monitoring interval threshold to obtain K grid nodes, and synchronizes the target sensor device set to the K grid nodes to obtain the K sensor deployment nodes and K sensor device sets.

[0029] Specifically, it should be understood that due to the differences in leaf size and growth height of different crops, the planting spacing of different crops is different, and therefore there are differences in the layout intervals of distributed sensors for achieving effective crop monitoring.

[0030] Based on this, this embodiment interactively obtains the planting intervals of multiple crops and the layout intervals of the sensor arrays for crop monitoring corresponding to the planting intervals to form multiple crop planting interval thresholds and multiple crop monitoring interval thresholds.

[0031] Multiple crop planting interval thresholds of multiple crops are used as multiple first attribute values ​​of the first attribute in the knowledge graph, and multiple crop monitoring interval thresholds are used as multiple second attribute values ​​of the second attribute in the knowledge graph. The multiple crop planting interval thresholds and the multiple crop monitoring interval thresholds are associated and stored based on the first attribute, the first attribute value, the second attribute, and the second attribute value to generate the monitoring constraint analysis layer. After obtaining the planting interval of the crop, the monitoring constraint analysis layer can quickly analyze and give the sensor layout interval distance for effectively monitoring the growth of the crop.

[0032] At the same time, different crops have different soil nutrient requirements for healthy growth in the soil. Accordingly, different crops also have different sensor types for monitoring the nutrient content of the soil during their growth. Based on this, this embodiment interactively obtains multiple sample sensor device sets for multiple sample crops, and the soil nutrient content monitoring sensor combination of each sample sensor device set can effectively monitor the soil nutrients required by the corresponding sample crops.

[0033] This embodiment adopts the same method of constructing the monitoring constraint analysis layer, and performs associated storage of the multiple sample crops and multiple sample sensor device sets based on the knowledge graph to generate a sensor constraint analysis layer. The sensor constraint analysis can quickly analyze and effectively monitor the nutrient content required for crop growth in the soil planted during the crop growth process under the condition of obtaining the crop type.

[0034] The monitoring constraint analysis layer and the sensing constraint analysis layer are arranged in parallel and the output ends of the monitoring constraint analysis layer and the sensing constraint analysis layer are connected to the input end of the pre-constructed monitoring array generation layer to complete the construction of the crop monitoring and analysis network.

[0035] A target planting interval of the target crop is interactively obtained, where the target planting interval is the interval distance between plants of the target crop when it is planted in the crop planting area.

[0036] The target planting interval and the target planted crops are synchronously input into the monitoring constraint analysis layer and the sensing constraint analysis layer of the crop monitoring and analysis network respectively, so as to determine the sensor layout interval threshold for soil monitoring of the target planted crops and the sensor composition of each sensor layout position based on the synchronous analysis of the monitoring constraint analysis layer and the sensing constraint analysis layer, and obtain the target monitoring interval threshold and the target sensing device set. It should be understood that the target sensing device set includes an unspecified variety of soil nutrient content monitoring sensors and a fixed combination of image acquisition sensors and soil moisture sensors.

[0037] The monitoring array generation layer performs grid processing on the crop planting area according to the target monitoring interval threshold to obtain K grid nodes, and synchronizes the target sensor device set to the K grid nodes to obtain the K sensor deployment nodes and K sensor device sets.

[0038] This embodiment achieves the technical effect of quickly analyzing and determining the sensor layout interval and co-position sensor layout for effective monitoring of soil nutrients and effective collection of plant growth status images during the crop growth process when the crop planting interval and crop type are obtained by constructing a dual-channel crop monitoring and analysis network.

[0039] A300: Deploy and position the K sensor deployment nodes in the crop planting area, and call and deploy sensors with reference to the K sensor equipment sets to generate K edge sensor platforms;

[0040] Specifically, in this embodiment, the sensor deployment positions are located on the actual land of the crop planting area according to the K sensor deployment nodes, and the sensors are called with reference to the K sensor equipment sets and the sensors are deployed at the located K points. Specifically, the soil nutrient content monitoring sensor and the soil moisture sensor are deployed in the soil, and the image acquisition sensor is deployed at a fixed position above the soil to generate K edge sensing platforms. Each edge sensing platform can collect soil nutrient data at a fixed location and collect crop images during the growth of the target crops.

[0041] A400: pre-build a standard cloud-edge architecture, wherein the standard cloud-edge architecture includes an integrated control cloud and K edge water and fertilizer analysis nodes, and completes non-standard processing of the standard cloud-edge architecture by building a communication connection between the K edge sensing platforms and the K edge water and fertilizer analysis nodes to obtain a fertilization and irrigation cloud-edge architecture;

[0042] In one embodiment, a standard cloud-edge architecture is pre-built, wherein the standard cloud-edge architecture includes an integrated control cloud and K edge water and fertilizer analysis nodes, and the method step A400 provided in the present application further includes:

[0043] A410: interactively obtaining a plurality of standard leaf growth image sets of the target planted crop, and constructing and generating a first growth deviation determiner based on the plurality of standard leaf growth image sets, wherein the plurality of standard leaf growth image sets are mapped to a plurality of standard growth stages;

[0044] A420: interactively obtaining a plurality of defective leaf growth image sets, and using the plurality of defective leaf growth image sets to optimize the discrimination function of a pre-constructed second growth deviation judgement device, wherein the second growth deviation judgement device includes a plurality of growth defect discrimination channels mapped to the plurality of standard growth stages;

[0045] A430: interactively obtaining a plurality of standard soil nutrient sets of the target planted crop, wherein the plurality of standard soil nutrient sets are mapped to the plurality of standard growth stages, and the plurality of standard growth stages and the plurality of standard soil nutrient sets are associated and stored based on a knowledge graph to generate a third growth deviation determiner;

[0046] A440: connecting the outputs of the first growth deviation determiner, the second growth deviation determiner and the third growth deviation determiner in series to obtain a standard water and fertilizer analysis network;

[0047] A450: Synchronize the standard water and fertilizer analysis network to the K edge water and fertilizer analysis nodes.

[0048] In one embodiment, a plurality of standard leaf growth image sets of the target planted crop are interactively obtained, and a first growth deviation determiner is constructed based on the plurality of standard leaf growth atlases. Step A410 of the method provided in the present application further includes:

[0049] A411: Preset a pixel extraction interval, and perform pixel point extraction on the plurality of standard leaf growth image sets based on the pixel extraction interval to obtain a plurality of groups of pixel point sets;

[0050] A412: calling RGB values ​​of the plurality of sets of pixel points, and serializing the RGB value calling results to obtain a plurality of standard RGB value change thresholds;

[0051] A413: Use the knowledge graph to associate and store the multiple standard growth stages and the multiple standard RGB value change thresholds to obtain the first growth deviation judgement.

[0052] In one embodiment, a plurality of defective leaf growth image sets are interactively obtained, and the plurality of defective leaf growth image sets are used to optimize the discrimination function of the pre-constructed second growth deviation determiner. The method step A420 provided in the present application further includes:

[0053] A421: The plurality of defective leaf growth image sets correspond to a plurality of soil nutrient defect types, and each of the plurality of defective leaf growth image sets includes a plurality of groups of defective leaf growth images corresponding to the plurality of standard growth stages;

[0054] A422: obtaining a first standard growth stage from the plurality of standard growth stages, and obtaining a plurality of groups of first stage growth images from the plurality of defective leaf growth image sets based on the first standard growth stage, wherein the plurality of groups of first stage growth images correspond to the plurality of soil nutrient deficiency types;

[0055] A423: constructing a plurality of nutrient defect recognition sub-channels based on a BP neural network, and using the plurality of groups of first-stage growth images to train the plurality of nutrient defect recognition sub-channels, and setting the trained plurality of nutrient defect recognition sub-channels in parallel to generate a first growth defect discrimination channel;

[0056] A424: Similarly, multiple growth defect discrimination channels are constructed using the multiple defective leaf growth image sets, and the construction of the second growth deviation judgement device is completed by connecting the multiple growth defect discrimination channels in parallel.

[0057] Specifically, in this embodiment, multiple standard leaf growth image sets of the target crop are interactively obtained, and the multiple standard leaf growth image sets are mapped to multiple standard growth stages. The pictures contained in each standard leaf growth image set can represent the normal and healthy leaf growth of the crop at different growth stages.

[0058] Before extracting the RGB values ​​of the pixels of the standard leaf growth image, in order to reduce the amount of data extraction, the present embodiment presets a pixel extraction interval, for example, the extraction interval is 100, which means that one pixel is extracted every 100 pixels. Based on the pixel extraction interval, pixel extraction is performed on the multiple standard leaf growth image sets to obtain multiple groups of pixel sets, each group of pixel sets corresponds to a standard leaf growth image set, and each pixel set of each group of pixel sets corresponds to a standard leaf growth image.

[0059] This embodiment performs RGB value calls on the multiple groups of pixel point sets to obtain multiple pixel point RGB value sets, and then classifies the data of each pixel point RGB value set according to red, green and blue to obtain multiple RGB value call results, wherein each RGB value call result is composed of a green value set, a red value set and a blue value set. The multiple RGB value call results are internally serialized to obtain multiple standard RGB value change thresholds, and each standard RGB value change threshold is composed of a red value change threshold, a green value change threshold and a blue value change threshold.

[0060] The knowledge graph is used to associate and store the multiple standard growth stages and the multiple standard RGB value change thresholds to obtain the first growth deviation judgement. This embodiment describes in detail the method of using the first growth deviation judgement in the subsequent description.

[0061] A plurality of defective leaf growth image sets are interactively obtained, wherein the plurality of defective leaf growth image sets correspond to a plurality of soil nutrient deficiency types, and each of the plurality of defective leaf growth image sets includes a plurality of groups of defective leaf growth images corresponding to the plurality of standard growth stages, and the images contained in each defective leaf growth image set can represent the abnormal leaf growth conditions of crops at different growth stages due to different nutrient deficiencies.

[0062] The first standard growth stage of the first growth cycle of the standard target planting crop is obtained by calling from the multiple standard growth stages, and multiple groups of first stage growth images are obtained from the multiple defective leaf growth image sets based on the first standard growth stage, and the multiple groups of first stage growth images correspond to the multiple soil nutrient deficiency types.

[0063] Based on the BP neural network, multiple nutrient deficiency identification sub-channels are obtained, and the multiple groups of first-stage growth images and the corresponding multiple soil nutrient deficiency types are used as training data. The conventional training method of the back propagation neural network is adopted to train the multiple nutrient deficiency identification sub-channels one by one, and the multiple nutrient deficiency identification sub-channels that have completed training (the leaf nutrient deficiency type identification accuracy is stably higher than 98%) are set in parallel to generate a first growth defect discrimination channel. The first growth defect discrimination channel can discriminate the soil nutrient type missing of the current crop based on the collected leaves of the crop in the first standard growth stage.

[0064] By analogy, multiple growth defect discrimination channels are constructed using the multiple defective leaf growth image sets, and the construction of the second growth deviation judger is completed by setting the multiple growth defect discrimination channels in parallel. This embodiment describes in detail the operation method of the second growth deviation judger in the subsequent description.

[0065] Specifically, it should be understood that each growth stage has a corresponding soil nutrient standard. Based on this, the present embodiment interactively obtains multiple standard soil nutrient sets of the target crop in multiple standard growth stages, and associates and stores the multiple standard growth stages and the multiple standard soil nutrient sets based on the knowledge graph to generate a third growth deviation discriminator. The third growth deviation discriminator can quickly calculate the soil nutrient deviation based on the real-time obtained soil nutrient content and the standard soil nutrient content. The present embodiment elaborates on the operation method of the third growth deviation discriminator in the subsequent specification.

[0066] The output end of the first growth deviation judger is connected to the input end of the second growth deviation judger, and the output end of the second growth deviation judger is connected to the output end of the third growth deviation judger to generate the standard water and fertilizer analysis network with a cascade structure.

[0067] In this embodiment, the standard water and fertilizer analysis network is copied to the K edge water and fertilizer analysis nodes, so that the nutrient deviation analysis of the soil planted with the target crop at an unspecified growth stage can be performed based on the K edge water and fertilizer analysis nodes.

[0068] Build communication connections between the K edge sensing platforms and the K edge water and fertilizer analysis nodes, complete non-standard processing of the standard cloud-edge architecture, and obtain a fertilization and irrigation cloud-edge architecture.

[0069] This embodiment constructs the fertilization and irrigation cloud-edge architecture, which can realize effective nutrient monitoring of the planting environment of the target crops and effective collection of growth images of the target crops, providing reference data for subsequent fertilization and irrigation feedback control based on the growth images of the target crops.

[0070] A500: The K edge water and fertilizer analysis nodes receive and perform nutrient demand deviation analysis based on the K real-time soil nutrient information and K real-time crop images sent by the K edge sensing platforms and in combination with the target planted crops, to obtain K edge water and fertilizer deviation parameters;

[0071] In one embodiment, the K edge water and fertilizer analysis nodes receive and perform nutrient demand deviation analysis based on the K real-time soil nutrient information and K real-time crop images sent by the K edge sensing platforms and the target planted crops to obtain K edge water and fertilizer deviation parameters. The method step A500 provided in the present application also includes:

[0072] A510: A first edge water and fertilizer analysis node receives a first real-time crop image sent by a first edge sensing platform, wherein the first edge water and fertilizer analysis node is pre-deployed with the standard water and fertilizer analysis network, and the first real-time crop image has a real-time growth stage identifier;

[0073] A520: Execute RGB value extraction on the first real-time crop image to obtain a real-time RGB value change threshold, and synchronize the first real-time crop image, the real-time growth stage identifier, and the real-time RGB value change threshold to the standard water and fertilizer analysis network of the first edge water and fertilizer analysis node;

[0074] A530: In the first growth deviation determiner, using the real-time growth stage identifier to traverse the multiple standard growth stages and the multiple standard RGB value change thresholds, calling to obtain the real-time standard change threshold, and determining whether the real-time RGB value change threshold satisfies the real-time standard change threshold;

[0075] A540: If the real-time RGB value change threshold does not meet the real-time standard change threshold, activating the second growth deviation determiner;

[0076] A550: In the second growth deviation determiner, the real-time growth stage identifier is used to traverse the multiple growth defect discrimination channels, the real-time growth defect discrimination channel is activated, and the first real-time crop image is synchronized to the multiple nutrient defect identification sub-channels of the real-time growth defect discrimination channel to perform nutrient defect identification, and a first nutrient defect identification result is obtained;

[0077] A560: The first edge water and fertilizer analysis node activates the first edge sensing platform to perform directional soil data collection on the crop planting area according to the first nutrient deficiency identification result, and obtains first real-time soil nutrient information;

[0078] A570: In the third growth deviation determiner, the real-time growth stage identifier is used to traverse the multiple standard soil nutrient sets to obtain a real-time standard soil nutrient set, and the first real-time soil nutrient information and the real-time standard soil nutrient set are compared to obtain a first marginal water and fertilizer deviation parameter.

[0079] Specifically, in this embodiment, since the method for the K edge water and fertilizer analysis nodes to analyze and determine the K edge water and fertilizer deviation parameters is consistent, this embodiment takes any first edge water and fertilizer analysis node to analyze and determine the first edge water and fertilizer deviation parameter as an example to provide a detailed explanation of the implementation method of K edge water and fertilizer analysis nodes to perform soil water and fertilizer deviation analysis.

[0080] It should be understood that in this embodiment, only the image acquisition sensor in the edge sensing platform is in operation for a long time, and the soil nutrient content monitoring sensor and the soil moisture sensor are controlled by the edge water and fertilizer analysis node, and only run the corresponding sensor type data collection when the edge water and fertilizer analysis node is activated.

[0081] Based on step A400, it can be known that the first edge water and fertilizer analysis node is pre-deployed with the standard water and fertilizer analysis network, and the first edge water and fertilizer analysis node receives the first real-time crop image sent by the image acquisition sensor of the first edge sensing platform, and the first real-time crop image has a real-time growth stage identification.

[0082] The same method as that used in step A412 to obtain multiple standard RGB value change thresholds is used to perform RGB value extraction on the first real-time crop image to obtain the real-time RGB value change threshold, and the first real-time crop image, the real-time growth stage identifier and the real-time RGB value change threshold are synchronized to the standard water and fertilizer analysis network of the first edge water and fertilizer analysis node.

[0083] In the first growth deviation determiner of the standard water and fertilizer analysis network, the real-time growth stage identifier is used to traverse the multiple standard growth stages and the multiple standard RGB value change thresholds, and the real-time standard change threshold that matches the current real-time growth stage identifier is called to obtain, and it is determined whether the real-time RGB value change threshold meets the real-time standard change threshold.

[0084] If the real-time RGB value change threshold does not meet the real-time standard change threshold, it indicates that the leaf color characteristics of the current target crop do not meet the leaf color characteristics of healthy plants at the current growth stage, and there are nutritional deficiencies. Conversely, if the real-time RGB value change threshold meets the real-time standard change threshold, it indicates that the leaf color characteristics of the current target crop meet the leaf color characteristics of healthy plants at the current growth stage, and there are no nutritional deficiencies.

[0085] If the real-time RGB value change threshold does not meet the real-time standard change threshold, the second growth deviation determiner is activated, and a specific judgment of the deviation nutrient type is performed based on the second growth deviation determiner.

[0086] In the second growth deviation determiner, the real-time growth stage identifier is first used to traverse the multiple growth defect discrimination channels, the real-time growth defect discrimination channels are activated, and the first real-time crop image is synchronized to the multiple nutrient defect identification sub-channels of the real-time growth defect discrimination channels, and nutrient defect identification is performed synchronously to obtain a first nutrient defect identification result, which includes one or more nutrient defects.

[0087] The first edge water and fertilizer analysis node activates one or more corresponding soil nutrient content monitoring sensors in the first edge sensing platform according to the first nutrient deficiency identification result, and performs targeted collection of soil nutrient data for the crop planting area based on the activated specific soil nutrient content monitoring sensors to obtain first real-time soil nutrient information, wherein the first real-time soil nutrient information includes measured values ​​of soil nutrient content of one or more nutrient deficiencies corresponding to the first nutrient deficiency identification result.

[0088] In the third growth deviation determiner, the real-time growth stage identifier is used to traverse the multiple standard soil nutrient sets to obtain a real-time standard soil nutrient set, and the first real-time soil nutrient information and the real-time standard soil nutrient set are compared to calculate the nutrient content deviation to obtain a first marginal water and fertilizer deviation parameter, wherein the first marginal water and fertilizer deviation parameter includes soil nutrient content deviation values ​​corresponding to one or more nutrient deficiencies of the first nutrient deficiency identification result.

[0089] By analogy, the K edge water and fertilizer analysis nodes receive and perform nutrient demand deviation analysis based on the K real-time soil nutrient information and K real-time crop images sent by the K edge sensing platforms and the target crops to obtain K edge water and fertilizer deviation parameters.

[0090] This embodiment uses a pre-built standard water and fertilizer analysis network to perform sensor data acquisition control and water and fertilizer deviation analysis. Compared with real-time monitoring of soil water and fertilizer deviation based solely on multiple types of soil nutrient content monitoring sensors, it achieves the technical effect of improving the accuracy of water and fertilizer deviation monitoring while reducing the cost of water and fertilizer deviation monitoring.

[0091] A600: The integrated control cloud of the fertilization and irrigation cloud-edge architecture receives the K edge water and fertilizer deviation parameters communicated back by the K edge water and fertilizer analysis nodes for data fusion to obtain a water and fertilizer feedback control strategy. The integrated control cloud controls the water and fertilizer integration device to perform feedback adjustment of irrigation and fertilization on the crop planting area using the water and fertilizer feedback control strategy.

[0092] In one embodiment, Figure 2 As shown, the integrated control cloud of the fertilization and irrigation cloud-edge architecture receives the K edge water and fertilizer deviation parameters communicated back by the K edge water and fertilizer analysis nodes for data fusion to obtain a water and fertilizer feedback control strategy. The method step A600 provided in this application also includes:

[0093] A610: Perform weight configuration according to the distance relationship between the K edge sensing platforms and the center of the crop planting area to obtain K deviation weights;

[0094] A620: Using the K deviation weights to perform weighted calculation on the K edge water-fertilizer deviation parameters to complete data fusion, and obtain optimized water-fertilizer deviation parameters;

[0095] A630: Perform feedback control conversion based on the optimized water-fertilizer deviation parameter to generate the water-fertilizer feedback control strategy.

[0096] Specifically, in this embodiment, weights are configured according to the distance relationship between the K edge sensing platforms and the center of the crop planting area to obtain K deviation weights, and the K deviation weights are pre-stored in the integrated control cloud of the fertilization and irrigation cloud-edge architecture.

[0097] The integrated control cloud receives the K edge water and fertilizer deviation parameters communicated back by the K edge water and fertilizer analysis nodes, uses the K deviation weights to perform weighted calculation on the K edge water and fertilizer deviation parameters to complete data fusion, and obtains optimized water and fertilizer deviation parameters, which are the overall nutrient deviation status of the crop planting area.

[0098] Then, the integrated control cloud uses the K edge water and fertilizer analysis nodes as a medium to activate K soil moisture sensors of K edge sensing platforms, obtain K real-time soil moisture feedback data, calculate the mean of the K real-time soil moisture feedback data as the overall soil moisture status of the crop planting area, interactively obtain multiple soil standard moisture levels of multiple standard growth stages of the target crop planting in history, compare the soil moisture status, and obtain the soil moisture deviation.

[0099] Then, multiple groups of historical soil moisture deviations - historical water-fertilizer deviation parameters - historical water-fertilizer feedback control strategies are interactively obtained, and then the soil moisture deviations and the optimized water-fertilizer deviation parameters are used to traverse the multiple groups of historical soil moisture deviations - historical water-fertilizer deviation parameters - historical water-fertilizer feedback control strategies to match the water-fertilizer feedback control strategies, and the water-fertilizer feedback control strategies are obtained. The integrated control cloud controls the water-fertilizer integrated device to perform feedback adjustment of irrigation and fertilization on the crop planting areas using the water-fertilizer feedback control strategy.

[0100] This embodiment achieves the technical effect of carrying out targeted and precise fertilization and irrigation according to the crop growth conditions, while ensuring that the resource supply in the crop growth process adapts to the growth needs, effectively improving nutrient utilization and reducing resource waste.

[0101] Embodiment 2

[0102] Based on the same inventive concept as the intelligent fertigation method based on feedback control in the aforementioned embodiment, Figure 3 As shown, the present application provides an intelligent fertigation system based on feedback control, wherein the system comprises:

[0103] The planting zoning generation unit 1 is used to interactively obtain the effective adjustment threshold of the water-fertilizer integrated device, and to effectively adjust the fertilization area according to the effective adjustment threshold to obtain the crop planting zoning;

[0104] A sensor array generation unit 2 is used to interactively obtain target crops, and synchronize the target crops and the crop planting zones to a pre-built crop monitoring and analysis network for sensor deployment analysis, and generate a feedback control sensor array, wherein the feedback control sensor array includes K sensor deployment nodes and K sensor device sets;

[0105] The sensor platform construction unit 3 is used to deploy and position the K sensor deployment nodes in the crop planting area, and to call and deploy sensors with reference to the K sensor equipment sets to generate K edge sensor platforms;

[0106] The cloud-edge architecture generation unit 4 is used to pre-build a standard cloud-edge architecture, wherein the standard cloud-edge architecture includes an integrated control cloud and K edge water and fertilizer analysis nodes, and the non-standard processing of the standard cloud-edge architecture is completed by building a communication connection between the K edge sensing platforms and the K edge water and fertilizer analysis nodes to obtain a fertilization and irrigation cloud-edge architecture;

[0107] The water-fertilizer deviation analysis unit 5 is used for the K edge water-fertilizer analysis nodes to receive and perform nutrient demand deviation analysis based on the K real-time soil nutrient information and K real-time crop images sent by the K edge sensing platforms and in combination with the target planted crops, to obtain K edge water-fertilizer deviation parameters;

[0108] The feedback regulation execution unit 6 is used for the integrated control cloud of the fertilization and irrigation cloud-edge architecture to receive the K edge water and fertilizer deviation parameters communicated back by the K edge water and fertilizer analysis nodes for data fusion to obtain a water and fertilizer feedback control strategy. The integrated control cloud controls the water and fertilizer integration device to perform feedback regulation of irrigation and fertilization on the crop planting area with the water and fertilizer feedback control strategy.

[0109] In one embodiment, the sensor array generation unit 2 further includes:

[0110] Interactively obtain multiple crop planting interval thresholds and multiple crop monitoring interval thresholds, and associate and store the multiple crop planting interval thresholds and the multiple crop monitoring interval thresholds based on the knowledge graph to generate a monitoring constraint analysis layer;

[0111] Interactively obtain multiple sample sensor device sets of multiple sample crops, and associate and store the multiple sample crops and the multiple sample sensor device sets based on the knowledge graph to generate a sensor constraint analysis layer;

[0112] The monitoring constraint analysis layer and the sensing constraint analysis layer are arranged in parallel and the output ends of the monitoring constraint analysis layer and the sensing constraint analysis layer are connected to the input end of the pre-constructed monitoring array generation layer to complete the construction of the crop monitoring and analysis network;

[0113] Interactively obtain the target planting interval of the target planted crop, and synchronously input the target planting interval and the target planted crop into the monitoring constraint analysis layer and the sensing constraint analysis layer of the crop monitoring and analysis network, respectively, to obtain a target monitoring interval threshold and a target sensing device set;

[0114] The monitoring array generation layer performs grid processing on the crop planting area according to the target monitoring interval threshold to obtain K grid nodes, and synchronizes the target sensor device set to the K grid nodes to obtain the K sensor deployment nodes and K sensor device sets.

[0115] In one embodiment, the cloud edge architecture generation unit 4 further includes:

[0116] Interactively obtaining a plurality of standard leaf growth image sets of the target planted crop, and constructing and generating a first growth deviation determiner based on the plurality of standard leaf growth image sets, wherein the plurality of standard leaf growth image sets are mapped to a plurality of standard growth stages;

[0117] Interactively obtain a plurality of defective leaf growth image sets, and use the plurality of defective leaf growth image sets to optimize the discrimination function of a pre-constructed second growth deviation judge, wherein the second growth deviation judge includes a plurality of growth defect discrimination channels mapped to the plurality of standard growth stages;

[0118] Interactively obtaining a plurality of standard soil nutrient sets of the target planted crop, wherein the plurality of standard soil nutrient sets are mapped to the plurality of standard growth stages, and the plurality of standard growth stages and the plurality of standard soil nutrient sets are associated and stored based on a knowledge graph to generate a third growth deviation determiner;

[0119] Connecting the outputs of the first growth deviation judgement device, the second growth deviation judgement device and the third growth deviation judgement device in series to obtain a standard water and fertilizer analysis network;

[0120] The standard water-fertilizer analysis network is synchronized to the K edge water-fertilizer analysis nodes.

[0121] In one embodiment, the cloud edge architecture generation unit 4 further includes:

[0122] Preset a pixel extraction interval, and perform pixel point extraction on the plurality of standard leaf growth image sets based on the pixel extraction interval to obtain a plurality of groups of pixel point sets;

[0123] Calling RGB values ​​for the multiple sets of pixel points, and serializing the RGB value calling results to obtain multiple standard RGB value change thresholds;

[0124] The knowledge graph is used to associate and store the multiple standard growth stages and the multiple standard RGB value change thresholds to obtain the first growth deviation determiner.

[0125] In one embodiment, the cloud edge architecture generation unit 4 further includes:

[0126] The plurality of defective leaf growth image sets correspond to a plurality of soil nutrient defect types, and each of the plurality of defective leaf growth image sets includes a plurality of groups of defective leaf growth images corresponding to the plurality of standard growth stages;

[0127] Obtaining a first standard growth stage from the plurality of standard growth stages, and obtaining a plurality of groups of first stage growth images from the plurality of defective leaf growth image sets based on the first standard growth stage, wherein the plurality of groups of first stage growth images correspond to the plurality of soil nutrient deficiency types;

[0128] Based on the BP neural network, a plurality of nutrient defect recognition sub-channels are obtained, and the plurality of groups of first-stage growth images are used to train the plurality of nutrient defect recognition sub-channels, and the plurality of trained nutrient defect recognition sub-channels are arranged in parallel to generate a first growth defect discrimination channel;

[0129] By analogy, the plurality of defective leaf growth image sets are used to construct a plurality of growth defect discrimination channels, and the second growth deviation judgement device is constructed by arranging the plurality of growth defect discrimination channels in parallel.

[0130] In one embodiment, the water-fertilizer deviation analysis unit 5 further includes:

[0131] The first edge water and fertilizer analysis node receives a first real-time crop image sent by the first edge sensing platform, wherein the first edge water and fertilizer analysis node is pre-deployed with the standard water and fertilizer analysis network, and the first real-time crop image has a real-time growth stage identifier;

[0132] Performing RGB value extraction on the first real-time crop image to obtain a real-time RGB value change threshold, and synchronizing the first real-time crop image, the real-time growth stage identifier, and the real-time RGB value change threshold to the standard water and fertilizer analysis network of the first edge water and fertilizer analysis node;

[0133] In the first growth deviation determiner, the real-time growth stage identifier is used to traverse the multiple standard growth stages and the multiple standard RGB value change thresholds, and the real-time standard change threshold is obtained by calling, and it is determined whether the real-time RGB value change threshold meets the real-time standard change threshold;

[0134] If the real-time RGB value change threshold does not meet the real-time standard change threshold, activating the second growth deviation determiner;

[0135] In the second growth deviation determiner, the real-time growth stage identifier is used to traverse the multiple growth defect discrimination channels, the real-time growth defect discrimination channel is activated, and the first real-time crop image is synchronized to the multiple nutrient defect identification sub-channels of the real-time growth defect discrimination channel to perform nutrient defect identification, and a first nutrient defect identification result is obtained;

[0136] The first edge water and fertilizer analysis node activates the first edge sensing platform to perform directional soil data collection on the crop planting area according to the first nutrient deficiency identification result, and obtains first real-time soil nutrient information;

[0137] In the third growth deviation determiner, the real-time growth stage identifier is used to traverse the multiple standard soil nutrient sets to obtain a real-time standard soil nutrient set, and the first real-time soil nutrient information and the real-time standard soil nutrient set are compared to obtain a first marginal water and fertilizer deviation parameter.

[0138] In one embodiment, the feedback adjustment execution unit 6 further includes:

[0139] Performing weight configuration according to the distance relationship between the K edge sensing platforms and the center of the crop planting area to obtain K deviation weights;

[0140] The K deviation weights are used to perform weighted calculation on the K edge water-fertilizer deviation parameters to complete data fusion, and obtain optimized water-fertilizer deviation parameters;

[0141] Feedback control conversion is performed based on the optimized water-fertilizer deviation parameter to generate the water-fertilizer feedback control strategy.

[0142] Any of the methods or steps described above may be stored as computer instructions or programs in various types of computer memories, and the computer instructions or programs may be recognized by various types of computer processors to implement any of the methods or steps described above.

[0143] Based on the above specific embodiments of the present invention, any improvements and modifications made to the present invention by those skilled in the art without departing from the principles of the present invention shall fall within the patent protection scope of the present invention.

Claims

1. Intelligent fertigation method based on feedback control, characterized in that: The method comprises: Interactively obtain the effective adjustment threshold of the water-fertilizer integrated device, and select the effective adjustment frame of the irrigation and fertilization area according to the effective adjustment threshold to obtain the crop planting zoning; Interactively obtain target planting crops, and synchronize the target planting crops and the crop planting zones to a pre-built crop monitoring and analysis network for sensor deployment analysis, and generate a feedback control sensor array, wherein the feedback control sensor array includes K sensor deployment nodes and K sensor device sets; Deploy and position the K sensor deployment nodes in the crop planting area, and call and deploy sensors with reference to the K sensor equipment sets to generate K edge sensor platforms; Pre-constructing a standard cloud-edge architecture, wherein the standard cloud-edge architecture includes an integrated control cloud and K edge water and fertilizer analysis nodes, and completing non-standard processing of the standard cloud-edge architecture by building a communication connection between the K edge sensing platforms and the K edge water and fertilizer analysis nodes to obtain a fertilization and irrigation cloud-edge architecture; The K edge water and fertilizer analysis nodes receive and perform nutrient demand deviation analysis based on the K real-time soil nutrient information and K real-time crop images sent by the K edge sensing platforms and in combination with the target planted crops to obtain K edge water and fertilizer deviation parameters; The integrated control cloud of the fertilization and irrigation cloud-edge architecture receives the K edge water and fertilizer deviation parameters communicated back by the K edge water and fertilizer analysis nodes for data fusion to obtain a water and fertilizer feedback control strategy, and the integrated control cloud controls the water and fertilizer integration device to perform feedback regulation of irrigation and fertilization on the crop planting area with the water and fertilizer feedback control strategy; Interactively obtaining target planting crops, and synchronizing the target planting crops and the crop planting zones to a pre-built crop monitoring and analysis network for sensor deployment analysis, and generating a feedback control sensor array, the method further comprising: Interactively obtain multiple crop planting interval thresholds and multiple crop monitoring interval thresholds, and associate and store the multiple crop planting interval thresholds and the multiple crop monitoring interval thresholds based on the knowledge graph to generate a monitoring constraint analysis layer; Interactively obtain multiple sample sensor device sets of multiple sample crops, and associate and store the multiple sample crops and the multiple sample sensor device sets based on the knowledge graph to generate a sensor constraint analysis layer; The monitoring constraint analysis layer and the sensing constraint analysis layer are arranged in parallel and the output ends of the monitoring constraint analysis layer and the sensing constraint analysis layer are connected to the input end of the pre-constructed monitoring array generation layer to complete the construction of the crop monitoring and analysis network; Interactively obtain the target planting interval of the target planted crop, and synchronously input the target planting interval and the target planted crop into the monitoring constraint analysis layer and the sensing constraint analysis layer of the crop monitoring and analysis network, respectively, to obtain a target monitoring interval threshold and a target sensing device set; The monitoring array generation layer performs grid processing on the crop planting area according to the target monitoring interval threshold to obtain K grid nodes, and synchronizes the target sensor device set to the K grid nodes to obtain the K sensor deployment nodes and K sensor device sets; The standard cloud-edge architecture includes an integrated control cloud and K edge water and fertilizer analysis nodes, and the method further includes: Interactively obtaining a plurality of standard leaf growth image sets of the target planted crop, and constructing and generating a first growth deviation determiner based on the plurality of standard leaf growth image sets, wherein the plurality of standard leaf growth image sets are mapped to a plurality of standard growth stages; Interactively obtain a plurality of defective leaf growth image sets, and use the plurality of defective leaf growth image sets to optimize the discrimination function of a pre-constructed second growth deviation judge, wherein the second growth deviation judge includes a plurality of growth defect discrimination channels mapped to the plurality of standard growth stages; Interactively obtaining a plurality of standard soil nutrient sets of the target planted crop, wherein the plurality of standard soil nutrient sets are mapped to the plurality of standard growth stages, and the plurality of standard growth stages and the plurality of standard soil nutrient sets are associated and stored based on a knowledge graph to generate a third growth deviation determiner; Connecting the outputs of the first growth deviation judgement device, the second growth deviation judgement device and the third growth deviation judgement device in series to obtain a standard water and fertilizer analysis network; Synchronizing the standard water and fertilizer analysis network to the K edge water and fertilizer analysis nodes; The integrated control cloud of the fertilization and irrigation cloud-edge architecture receives the K edge water-fertilizer deviation parameters communicated back by the K edge water-fertilizer analysis nodes for data fusion to obtain a water-fertilizer feedback control strategy, and the method further includes: Performing weight configuration according to the distance relationship between the K edge sensing platforms and the center of the crop planting area to obtain K deviation weights; The K deviation weights are used to perform weighted calculation on the K edge water-fertilizer deviation parameters to complete data fusion, and obtain optimized water-fertilizer deviation parameters; Feedback control conversion is performed based on the optimized water-fertilizer deviation parameter to generate the water-fertilizer feedback control strategy.

2. The method according to claim 1, characterized in that Interactively obtaining a plurality of standard leaf growth image sets of the target planted crop, and constructing and generating a first growth deviation determiner based on the plurality of standard leaf growth atlases, the method further comprising: Preset a pixel extraction interval, and perform pixel point extraction on the plurality of standard leaf growth image sets based on the pixel extraction interval to obtain a plurality of groups of pixel point sets; Calling RGB values ​​for the multiple sets of pixel points, and serializing the RGB value calling results to obtain multiple standard RGB value change thresholds; The plurality of standard growth stages and the plurality of standard RGB value change thresholds are associated and stored using a knowledge graph to obtain the first growth deviation determiner.

3. The method according to claim 2, characterized in that Interactively obtaining a plurality of defective leaf growth image sets, and using the plurality of defective leaf growth image sets to optimize the discrimination function of a pre-constructed second growth deviation determiner, the method further comprising: The plurality of defective leaf growth image sets correspond to a plurality of soil nutrient defect types, and each of the plurality of defective leaf growth image sets includes a plurality of groups of defective leaf growth images corresponding to the plurality of standard growth stages; Obtaining a first standard growth stage from the plurality of standard growth stages, and obtaining a plurality of groups of first stage growth images from the plurality of defective leaf growth image sets based on the first standard growth stage, wherein the plurality of groups of first stage growth images correspond to the plurality of soil nutrient deficiency types; Based on the BP neural network, a plurality of nutrient defect recognition sub-channels are obtained, and the plurality of groups of first-stage growth images are used to train the plurality of nutrient defect recognition sub-channels, and the plurality of trained nutrient defect recognition sub-channels are arranged in parallel to generate a first growth defect discrimination channel; By analogy, the plurality of defective leaf growth image sets are used to construct a plurality of growth defect discrimination channels, and the second growth deviation judgement device is constructed by arranging the plurality of growth defect discrimination channels in parallel.

4. The method according to claim 3, characterized in that The K edge water and fertilizer analysis nodes receive and perform nutrient demand deviation analysis based on the K real-time soil nutrient information and K real-time crop images sent by the K edge sensing platforms and in combination with the target planted crops to obtain K edge water and fertilizer deviation parameters. The method further includes: The first edge water and fertilizer analysis node receives a first real-time crop image sent by the first edge sensing platform, wherein the first edge water and fertilizer analysis node is pre-deployed with the standard water and fertilizer analysis network, and the first real-time crop image has a real-time growth stage identifier; Performing RGB value extraction on the first real-time crop image to obtain a real-time RGB value change threshold, and synchronizing the first real-time crop image, the real-time growth stage identifier, and the real-time RGB value change threshold to the standard water and fertilizer analysis network of the first edge water and fertilizer analysis node; In the first growth deviation determiner, the real-time growth stage identifier is used to traverse the multiple standard growth stages and the multiple standard RGB value change thresholds, and the real-time standard change threshold is obtained by calling, and it is determined whether the real-time RGB value change threshold meets the real-time standard change threshold; If the real-time RGB value change threshold does not meet the real-time standard change threshold, activating the second growth deviation determiner; In the second growth deviation determiner, the real-time growth stage identifier is used to traverse the multiple growth defect discrimination channels, the real-time growth defect discrimination channel is activated, and the first real-time crop image is synchronized to the multiple nutrient defect identification sub-channels of the real-time growth defect discrimination channel to perform nutrient defect identification, and a first nutrient defect identification result is obtained; The first edge water and fertilizer analysis node activates the first edge sensing platform to perform directional soil data collection on the crop planting area according to the first nutrient deficiency identification result, and obtains first real-time soil nutrient information; In the third growth deviation determiner, the real-time growth stage identifier is used to traverse the multiple standard soil nutrient sets to obtain a real-time standard soil nutrient set, and the first real-time soil nutrient information and the real-time standard soil nutrient set are compared to obtain a first marginal water and fertilizer deviation parameter.

5. Intelligent fertigation system based on feedback control, characterized in that: The steps for implementing the method according to any one of claims 1 to 4 include: A planting zoning generation unit, used to interactively obtain an effective adjustment threshold of the water-fertilizer integrated device, and to effectively adjust the fertilization area according to the effective adjustment threshold to obtain the crop planting zoning; A sensor array generation unit, used for interactively obtaining target planting crops, and synchronizing the target planting crops and the crop planting zones to a pre-built crop monitoring and analysis network for sensor deployment analysis, and generating a feedback control sensor array, wherein the feedback control sensor array includes K sensor deployment nodes and K sensor device sets; A sensor platform construction unit, configured to deploy and position the K sensor deployment nodes in the crop planting area, and call and deploy sensors with reference to the K sensor device sets to generate K edge sensor platforms; A cloud-edge architecture generation unit is used to pre-build a standard cloud-edge architecture, wherein the standard cloud-edge architecture includes an integrated control cloud and K edge water and fertilizer analysis nodes, and non-standard processing of the standard cloud-edge architecture is completed by building a communication connection between the K edge sensing platforms and the K edge water and fertilizer analysis nodes to obtain a fertilization and irrigation cloud-edge architecture; A water-fertilizer deviation analysis unit is used for the K edge water-fertilizer analysis nodes to receive and perform nutrient demand deviation analysis based on the K real-time soil nutrient information and K real-time crop images sent by the K edge sensing platforms and in combination with the target planted crops, to obtain K edge water-fertilizer deviation parameters; A feedback regulation execution unit is used for the integrated control cloud of the fertilization and irrigation cloud-edge architecture to receive the K edge water and fertilizer deviation parameters communicated back by the K edge water and fertilizer analysis nodes for data fusion to obtain a water and fertilizer feedback control strategy. The integrated control cloud controls the water and fertilizer integration device to perform feedback regulation of irrigation and fertilization on the crop planting area using the water and fertilizer feedback control strategy.

Images