Agricultural data interaction method and platform based on multi-source information

Abstract

The invention relates to an agricultural data interaction method and platform based on multi-source information, and belongs to the technical field of agricultural data processing. Comprising the following steps: processing collected data through a corresponding edge calculation module to generate a dynamic label; analyzing and processing the agricultural information data to obtain a supervision label and corresponding supervision data; all the tags and the corresponding data are imported into a geographic information system, matching is carried out according to the corresponding areas and the respective tags, and meanwhile, the monitoring tags are recognized and updated through a cloud platform computing center. The real-time data is fused and sent to the cloud computing platform through the edge computing module, the interaction prompt is generated through association with the supervision data, so that the real-time performance and the high efficiency of data processing are improved, meanwhile, the monitoring data and the supervision data are associated through the interaction prompt, and the interactivity of visual display is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of agricultural data processing, and in particular relates to an agricultural data interaction method and platform based on multi-source information. Background Art

[0002] With the development of agricultural modernization and information technology, a large amount of multi-source data has been generated in agricultural production, including meteorological data, soil data, crop growth data, pest and disease monitoring data, agricultural production management data, etc. These data come from a wide range of sources and in different formats, which poses a great challenge to the integration and utilization of agricultural data. Traditional agricultural data management systems often cannot effectively collect information from different sources, and lack real-time and intelligent data analysis and early warning capabilities, and cannot provide sufficient data support for agricultural production decisions.

[0003] At the same time, the visualization of agricultural data is relatively simple, only considering the direct data of agricultural production, and not visualizing the decision-making data of regulatory departments, socio-economic data, science and technology and innovation data, etc., and it is unable to highlight the interactivity of data between different functional departments. Summary of the invention

[0004] In order to solve the above-mentioned problems existing in the prior art, the present invention provides an agricultural data interaction method and platform based on multi-source information, which integrates real-time data through an edge computing module and sends it to a cloud computing platform, and generates interactive prompts by associating it with regulatory data, thereby improving the real-time and efficiency of data processing. At the same time, through interactive prompts, the monitoring data and regulatory data are associated, thereby improving the interactivity of the visual display.

[0005] The purpose of the present invention can be achieved through the following technical solutions: A first aspect of the present disclosure provides an agricultural data interaction method based on multi-source information, comprising the following steps: Regional agricultural data processing: Divide the target areas for agricultural supervision, set up real-time data collection equipment for each area, process the collected data through the corresponding edge computing module, and generate dynamic labels; Processing of agriculture-related information: Collect relevant information about smart agriculture through access to third-party platforms, analyze and process the information data, and obtain regulatory labels and corresponding regulatory data; Global data visualization: import all tags and corresponding data into the geographic information system, match them according to the corresponding areas and their respective tags, and identify and update the monitoring tags through the cloud platform computing center; The global data visualization includes the following steps: Data matching display: Dynamic tags are displayed in a mobile manner at corresponding locations on the map according to the coordinate system of the data collection device, and interactive prompts are provided by calculating the correlation between dynamic tags and regulatory tags; Deepening of interactive prompts: Call the trend prediction model based on the dynamic tags of interactive prompts, perform feature engineering on the monitoring data corresponding to the dynamic tags, and input them into the trend prediction model to obtain the trend changes of the dynamic tags. At the same time, the monitoring data will be used for early warning prompts based on regulatory data.

[0006] Furthermore, the regional agricultural data processing includes the following steps: Agricultural area division: set up data collection equipment according to data application requirements, set up edge computing modules according to the expansion of data collection, and divide agricultural supervision areas based on edge computing modules; Collect monitoring data: Through the set data collection equipment, each edge computing module obtains real-time data by setting the collection frequency; Dynamic label generation: After the edge computing module pre-processes the real-time data, it generates dynamic labels for the pre-processed data through the association rule engine; Monitoring data upload: The data with dynamic tags assigned and the data without dynamic tags are integrated into corresponding data packets and encrypted before uploading to the cloud computing center.

[0007] Furthermore, the dynamic tag generation further includes the following steps: Align and normalize the state data and dynamic tags of agricultural historical data based on time periods, merge the state data based on dynamic tags, and establish a dynamic tag transaction database; The association rule mining algorithm is used to mine the frequent item sets corresponding to each dynamic tag from the dynamic tag transaction database, generate association rules from the frequent item sets, and calculate the support and confidence of the rules; The association rules are screened by setting the minimum support threshold and the confidence threshold to generate a series of association rules for dynamic tags.

[0008] Furthermore, the processing of the agriculture-related information comprises the following steps: Collect relevant information: obtain decision-making data, social and economic data, and science and technology and innovation data from regulatory authorities through third-party platforms; Generate regulatory labels: Use a multimodal large model to process relevant information data, fuse data from different modalities, and generate corresponding regulatory labels; Extract regulatory data: Obtain the range threshold of the corresponding data type based on the generated regulatory labels. For regulatory labels that cannot be directly obtained from relevant information, use a multimodal large model to infer and supplement reference values; Monitoring data labeling: For monitoring data uploaded to the cloud computing center, labels are annotated by data type and by calculating the similarity between the range threshold or reference value of the monitoring data and the regulatory data.

[0009] Furthermore, the global data visualization further includes the following steps: Import multi-source data: All relevant data of the target agricultural area are uniformly converted into a format supported by the geographic information system, and descriptive metadata is added to the data and stored in the cloud computing center. The pre-processed data is imported into the GIS for visualization; Association rule update: The cloud computing center collects monitoring data, and adds it to the dynamic label transaction database after establishing a complete data format by labeling the monitoring data. It updates the association rules through association mining and sends them to the edge computing module.

[0010] Furthermore, the interaction prompt deepening includes the following steps: Integrate the monitoring data used for dynamic label generation into a dataset, and perform feature engineering on the integrated dataset, including feature extraction, feature selection, and feature construction; Select a machine learning model as the basis of the trend prediction model, use the existing historical data to train the selected model, use the fusion results of multi-source data as input features, and use the trend changes of dynamic labels as output; The data corresponding to the dynamic label and the regulatory label are compared based on the interactive prompts, and an early warning prompt is issued when the monitoring data exceeds the range threshold or reference value of the regulatory data.

[0011] A second aspect of the present disclosure provides an agricultural data interaction platform based on multi-source information, which is used to implement an agricultural data interaction method based on multi-source information as described above, including a data collection and fusion module, a data storage and management module, a data analysis and mining module, an intelligent decision-making and early warning module, and a data sharing and interaction module; The data collection and fusion module, the data storage and management module, the data analysis and mining module, the intelligent decision-making and early warning module and the data sharing and interaction module are sequentially connected in communication; The data collection and fusion module is used to collect and pre-process multi-source agricultural data, and then fuse and associate heterogeneous data from different data sources; The fusion and association include the following steps: Data collection: Using intelligent sensing technology through the set data collection equipment, each edge computing module obtains real-time data by setting the collection frequency, including meteorological sensors, soil sensors, cameras, drones and satellite remote sensing; Data preprocessing: Thoroughly clean the collected raw data, remove invalid, erroneous or incomplete records, perform data standardization, and convert data of different dimensions and scales into a unified data format that can be used for association rule engine analysis; Data fusion: The pre-processed data is input into the association rule engine to generate dynamic labels, complete the fusion of multi-source data, and perform subsequent prediction and visualization.

[0012] As a preferred technical solution of the present invention, the data storage and management module is used to store the collected and collected data in a database, and select a cloud computing center to realize parallel storage and access of data.

[0013] As a preferred technical solution of the present invention, the data analysis and mining module is used to analyze and process all data through the cloud computing center, including: A rule engine that mines agricultural historical data and corresponding status through association rule mining algorithms to obtain a series of dynamic labels; Analyze the decision-making data, socio-economic data, and science and innovation data of regulatory authorities through multimodal big models to generate regulatory labels and regulatory data; The machine learning model is trained with existing historical data to establish a trend prediction model, and the monitoring data of dynamic tags is input into the model to predict trend changes.

[0014] As a preferred technical solution of the present invention, the intelligent decision-making and early warning module is used to calculate the correlation based on the generated dynamic tags and regulatory tags, and interactively prompt the dynamic tags when the designed threshold is reached, and compare the data corresponding to the dynamic tags and the regulatory tags according to the interactive prompts, and issue an early warning prompt when the monitoring data exceeds the range threshold or reference value of the regulatory data; The data sharing and interaction module is used to integrate the target agricultural areas, dynamic tags and real-time data into the geographic information system for visual display, wherein the dynamic tags also include interactive prompts, trend forecasts and early warning prompts.

[0015] The beneficial effects of the present invention are: The present invention firstly generates dynamic labels for direct agricultural production data through an association rule engine through edge computing modules in different areas, and completes data collection and fusion through distributed processing, which reduces the difficulty of integrating massive data and improves the processing efficiency of real-time data; at the same time, other smart agriculture-related information data are integrated through a multimodal large model to extract regulatory labels and regulatory data, guide the data labeling of monitoring data, and improve the consistency and relevance of data labeling; finally, the dynamic label and the regulatory label correlation calculation are interactively prompted, thereby linking the monitoring data and the regulatory data, and visualizing them based on GIS, thereby improving the interactivity of agricultural data management.

[0016] The present invention establishes a dynamic map of the target agricultural area based on GIS in a visual display, which includes an agricultural area map, dynamic labels and real-time data. The dynamic labels include interactive prompts, trend forecasts and early warning prompts. The interactive prompts provide users with the association status between the current monitoring data and the regulatory data. At the same time, trend change forecasts and early warning prompts are performed for the dynamic labels of the interactive prompts, thereby improving the visualization effect, helping users to understand complex data and analysis results in an intuitive and concise manner, and enhancing real-time response capabilities through early warning prompts, thereby improving decision-making efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0018] Figure 1 A flowchart of an agricultural data interaction method based on multi-source information provided by an embodiment of the present invention; Figure 2 A schematic diagram of a process flow for regional agricultural data processing provided by an embodiment of the present invention; Figure 3 A schematic diagram of a process for processing agriculture-related information provided by an embodiment of the present invention; Figure 4 A schematic diagram of a process for global data visualization provided by an embodiment of the present invention; Figure 5 A schematic diagram of the structure of an agricultural data interaction platform based on multi-source information provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0020] The present invention provides an agricultural data interaction method and platform based on multi-source information, which is used to realize the integration and visualization of multi-source data of agricultural production. By using a geographic information system to visualize and monitor all data, early warning information is provided to agricultural producers and decision makers in a timely manner, thereby improving the interaction capabilities of different functional departments with agricultural data and providing a more efficient processing method for smart agriculture. Specifically, the agricultural data interaction method and platform based on multi-source information of the present invention are described in detail through the following content: This embodiment provides an agricultural data interaction method based on multi-source information. Figure 1 As shown, the following steps are included: S1. Regional agricultural data processing: Divide the target area for agricultural supervision, set up real-time data collection equipment for each area, process the collected data through the corresponding edge computing module, and generate dynamic labels, such as Figure 2 As shown, the following steps are included: S11. Agricultural area division: Data collection equipment is set up according to data application requirements, edge computing modules are set up according to the expansion status of data collection, and agricultural supervision area division is carried out based on edge computing modules; It is understandable that data collection equipment includes meteorological sensors, soil sensors, cameras, drones, and satellite remote sensing, covering all aspects from environmental monitoring to crop growth, soil quality, irrigation management, etc., providing more comprehensive basic data for smart agriculture.

[0021] It should be noted that agricultural regulatory data requires the setting of different types and quantities of collection equipment according to different regulatory objectives. However, due to the rapid expansion of agricultural data, it is impossible to complete the processing and analysis of real-time data within a certain period of time, resulting in poor data management and analysis decision-making effects. Therefore, an edge computing module is set up to perform real-time processing of collected data in a cloud-edge collaborative manner to improve the real-time and efficiency of data processing.

[0022] S12, collecting monitoring data: through the set data collection equipment, each edge computing module obtains real-time data by setting the collection frequency; S13, dynamic label generation: After the edge computing module pre-processes the real-time data, it generates dynamic labels for the pre-processed data through the association rule engine; It should be noted that the association rule engine includes a series of association rules, wherein the association rules are mined and obtained from agricultural historical data and corresponding status through the cloud computing center, including the following steps: S131, aligning and normalizing the state data and dynamic tags of the agricultural historical data based on the time period, merging the state data based on the dynamic tags, and establishing a dynamic tag transaction database; S132, using an association rule mining algorithm (Apriori, FP-Growth, etc.) to mine frequent item sets corresponding to each dynamic tag from the dynamic tag transaction database, generate association rules from the frequent item sets, and calculate the support and confidence of the rules; S133. Filter the association rules by setting a minimum support threshold and a confidence threshold to generate a series of association rules for dynamic tags.

[0023] It should be noted that dynamic tags are used to reflect some conditions associated with the current status data, such as insect infestation, spores, weather, soil moisture, irrigation, seedling conditions, disaster conditions, video surveillance, production, traceability, etc. in the agricultural production area. The generated association rules include "If the soil moisture is high and the temperature is moderate, then the insect infestation is high".

[0024] S14, monitoring data upload: the data with assigned dynamic tags and the data without assigned dynamic tags are integrated into corresponding data packets, encrypted and uploaded to the cloud computing center.

[0025] It can be understood that the data assigned with dynamic labels are integrated into data packets based on the dynamic labels, and the data not assigned with dynamic labels are integrated into data packets with unknown labels, wherein the data assigned with dynamic labels can be understood as the conditional data to which the association rule engine generates the dynamic labels, and the other remaining data is regarded as the data not assigned with dynamic labels.

[0026] It should be noted that in this step, the data is divided and uploaded to facilitate subsequent trend prediction through the data corresponding to the dynamic tags. At the same time, all data is uploaded for the cloud computing center to update the association rules. The update here includes enriching the dynamic tags and adjusting the association rule screening thresholds, thereby improving the comprehensiveness of the dynamic tags and the flexibility of the association rules.

[0027] S2. Processing of agriculture-related information: Collect relevant information about smart agriculture through accessing third-party platforms, analyze and process the information data, and obtain regulatory labels and corresponding regulatory data, such as Figure 3 As shown, the following steps are included: S21. Collect relevant information: obtain decision-making data, socio-economic data, and science and technology and innovation data from regulatory authorities through third-party platforms; It is understandable that in the process of realizing smart agriculture, it not only includes agricultural production data, but also includes decision-making data of regulatory departments, social and economic data, and science and technology and innovation data, etc. Among them, agricultural production data involves direct data in the agricultural production process, including crop planting, farmland management, climate conditions, etc.; decision-making data is relevant data formulated by regulatory departments based on national or regional policies, regulations and plans, and these data help to make decisions and supervise agricultural activities; social and economic data reflects the impact of social and economic factors on agricultural production and helps regulatory departments evaluate the agricultural economic situation and risks; science and technology and innovation data involves data in the fields of agricultural technology, innovation, education, etc., providing technical support for agricultural production and supervision.

[0028] It should be noted that in the above process, it is necessary to encrypt and license the acquired data to prevent data leakage problems and improve data security.

[0029] S22. Generate supervision labels: Use a multi-modal large model to process relevant information data, fuse data from different modalities (such as text, images, etc.), and generate corresponding supervision labels, which specifically include the steps: S221. Data preprocessing: Include preprocessing of text data and image data: Text data preprocessing: Use regular expressions to clean HTML tags, special characters and extra spaces in the text; use jieba to segment the Chinese text, disassembling each sentence into meaningful words; according to the task requirements, remove stop words (such as "de", "he") or perform part-of-speech tagging.

[0030] Image data preprocessing: Use OpenCV to scale, rotate and crop the image to ensure that the input image has the same size (such as 224x224); perform normalization processing on the image, scaling the pixel value range from [0, 255] to [0, 1]; use image enhancement methods (such as flipping, rotating, color change) to increase data diversity.

[0031] S222. Modal feature extraction: Include text modal feature extraction and image modal feature extraction: Text modal feature extraction: Load a pre-trained BERT model (such as the model provided by Hugging Face), input the text data into the BERT model for encoding, obtain the context-related word vectors of the text, and extract the hidden state of the last layer as the feature representation of the text.

[0032] Image modal feature extraction: Load a ResNet model (usually ResNet-50 or ResNet-101), input the preprocessed image into ResNet, and use its residual connection to extract image features; extract the features of the second-to-last layer (the layer before the fully connected layer) as the image representation.

[0033] S223, Multimodal Fusion: Extract feature vectors from text and image modalities respectively, such as BERT text features and ResNet image features, concatenate the text features and image features to obtain a larger feature vector, and standardize the concatenated feature vector to ensure consistent feature scale.

[0034] S224, Model training: Build a multimodal Transformer model and input text and image features into the model through multiple encoders (processing text and image modalities respectively); Establish connections between multiple modalities through the self-attention mechanism and learn the associations between different modalities; Use the Cross-Entropy Loss function for classification task training, or use other suitable loss functions.

[0035] S225, Generate supervision labels: Input new multimodal data, use the trained multimodal Transformer model for inference, and output a feature vector; The feature vector is passed to the adaptive Softmax classifier, which uses a hierarchical calculation method and performs calculations according to the frequency of the category. The model outputs a probability distribution representing the probabilities of different regulatory labels and selects the label with the highest probability as the final regulatory label.

[0036] S23. Extracting regulatory data: Obtaining the range threshold of the corresponding data type based on the generated regulatory labels. For regulatory labels that cannot be directly obtained from relevant information, inferring and supplementing reference values ​​through a multimodal large model; It is understandable that the regulatory data refers to the limited range or reference value of each data type corresponding to the regulatory tag.

[0037] S24. Monitoring data labeling: For the monitoring data uploaded to the cloud computing center, labels are provided for the monitoring data by calculating the data type and the similarity between the range threshold or reference value of the monitoring data and the regulatory data.

[0038] It is understandable that the monitoring data annotation label is the dynamic label of the data used for subsequent association rule updating.

[0039] It should be noted that the multimodal big model is a deep learning model that can simultaneously process and understand multiple types of data (modalities), usually including multiple data forms such as text, images, sounds, time series, etc. In this embodiment, the multimodal big model is used to complete the regulatory label generation and regulatory data range threshold extraction and reference value generation of the regulatory department's decision-making data, socio-economic data, and science and technology and innovation data, so that the regulatory labels are more accurate and the data thresholds are more complete. At the same time, the above-mentioned generated results are used to annotate the latest monitoring data, which facilitates the subsequent updating of association rules and improves the consistency and relevance of data label annotation.

[0040] S3, global data visualization: import all tags and corresponding data into the geographic information system, and match them according to the corresponding areas and their respective tags. At the same time, the monitoring tags are identified and updated through the cloud platform computing center, such as Figure 4 As shown, the following steps are included: S31. Import multi-source data: All relevant data of the target agricultural area are uniformly converted into a format supported by the Geographic Information System (GIS), and descriptive metadata is added to the data and stored in the cloud computing center. The pre-processed data is imported into GIS for visualization.

[0041] Among them, the associated data includes remote sensing data, monitoring data, regulatory data, and public data, as well as descriptive metadata such as acquisition time, data source, and coordinate system. It is understandable that data import uses a GIS platform (such as ArcGIS or QGIS) to load raster and vector data; directly imports data from the database to the WebGIS platform through the API; superimposes the farmland boundary vector layer with the NDVI remote sensing image layer, and displays the real-time data of the weather station in the form of a heat map; distinguishes different crop types through color, transparency, etc., and dynamically renders real-time meteorological changes.

[0042] It should be noted that, in this embodiment, a geographic information system is established for the target agricultural area, and then the monitoring data and regulatory data are loaded into the GIS in the form of dynamic tags and regulatory tags, respectively, so as to efficiently display a large amount of data and improve the interactivity of the data.

[0043] S32. Data matching display: Dynamic tags are displayed in a mobile manner at corresponding positions on the map according to the coordinate system of the data acquisition device, and interactive prompts are provided by calculating the correlation between dynamic tags and regulatory tags.

[0044] It should be noted that the correlation calculation between dynamic labels and regulatory labels is based on the data types and data values ​​of each label, and the correlation between the two is obtained by using correlation calculation methods such as the Pearson correlation coefficient method. The correlation is defined by setting a threshold to complete the association between dynamic labels and regulatory labels for interactive prompts, which can be rendered in color on the map.

[0045] S33, interactive prompt deepening: according to the dynamic tag of the interactive prompt, the trend prediction model is called, the monitoring data corresponding to the dynamic tag is processed by feature engineering and then input into the trend prediction model to obtain the trend change of the dynamic tag, and the monitoring data is used for early warning prompts based on the supervision data, which specifically includes the following steps: S331, integrating the monitoring data used for dynamic label generation into a data set, and performing feature engineering on the integrated data set, including feature extraction, feature selection, and feature construction; S332, select a machine learning model as the basis of the trend prediction model, use the existing historical data to train the selected model, use the fusion results of multi-source data as input features, and use the trend changes of dynamic labels as output; S333. Compare the data corresponding to the dynamic label and the regulatory label based on the interactive prompt, and issue an early warning prompt when the monitoring data exceeds the range threshold or reference value of the regulatory data.

[0046] It should be noted that in the process of trend prediction, there is no limitation on the machine learning model. The monitoring data corresponding to the dynamic label is used as input, which is equivalent to the conditional data of the association rule. That is, the input features of the trend prediction model have been screened, which improves the prediction accuracy of the dynamic label. At the same time, the monitoring data is compared with the supervision data on the basis of interactive prompts. When it exceeds the range, an early warning prompt is issued to make timely processing of the current status. In this process, the matching of the compared data types is completed through interactive prompts. Therefore, through the above trend prediction and early warning prompts, the comprehensive status of the dynamic label is grasped at the same time for visual display. The above processing processes are all based on interactive prompts. Therefore, the prediction and early warning are completed by screening data, which improves the relevance and efficiency of visualization.

[0047] In this embodiment, the display granularity of the interactive prompt is displayed according to the degree of data visualization. The number of dynamic labels generated is adjusted by adjusting the minimum support threshold and confidence threshold in the association rule mining process, thereby adjusting the number of visual interactive prompts. The dynamic labels for interactive prompts have corresponding relevance to regulatory decisions. The platform completes the interaction between the collected direct data and regulatory decision data through the interactive method mentioned in this embodiment. After obtaining the dynamic labels for interactive prompts, trend prediction is performed, which provides association display and trend prediction for the visualization of agricultural data, improves data integration efficiency and display pertinence, and provides a reference for decision support.

[0048] S34, association rule update: The cloud computing center collects monitoring data, and adds it to the dynamic label transaction database after establishing a complete data format by labeling the monitoring data. It updates the association rules through association mining and sends them to the edge computing module.

[0049] This embodiment also provides an agricultural data interaction platform based on multi-source information, such as Figure 5 As shown, it includes a data acquisition and fusion module, a data storage and management module, a data analysis and mining module, an intelligent decision and early warning module and a data sharing and interaction module; the data acquisition and fusion module, the data storage and management module, the data analysis and mining module, the intelligent decision and early warning module and the data sharing and interaction module are communicated and connected in sequence.

[0050] The data collection and fusion module is used to collect and pre-process multi-source agricultural data, and then fuse and associate heterogeneous data from different data sources.

[0051] The fusion and association include the following steps: A11. Data collection: By using intelligent sensing technology through the set data collection equipment, each edge computing module obtains real-time data by setting the collection frequency, including meteorological sensors, soil sensors, cameras, drones and satellite remote sensing.

[0052] A12. Data preprocessing: Thoroughly clean the collected raw data and remove invalid, erroneous or incomplete records to eliminate noise and outliers in the data; Perform data standardization processing to convert data of different dimensions and scales into a unified data format that can be used for association rule engine analysis.

[0053] A13. Data fusion: Input the pre-processed data into the association rule engine to generate dynamic labels, complete the fusion of multi-source data, and perform subsequent prediction and visualization.

[0054] The data storage and management module is used to store the collected and collected data in a database and select a cloud computing center to realize parallel storage and access of data.

[0055] It should be noted that the data storage and management module, when realizing its functions, includes: formatting, deduplication and error correction of all data; designing the database model according to data characteristics and query requirements; selecting a suitable distributed storage system (such as HDFS, HBase); designing data partitioning strategies (such as time partitioning, logical division by region, business line, etc.) to optimize parallel storage and access; loading data through ETL tools or distributed data import tools (such as Sqoop, Kafka); introducing Redis (Remote Dictionary Server) to accelerate hot data queries; setting permissions and security policies to protect data.

[0056] The data analysis and mining module is used to analyze and process all data through the cloud computing center, including: A rule engine that mines agricultural historical data and corresponding status through association rule mining algorithms to obtain a series of dynamic labels; Analyze the decision-making data, socio-economic data, and science and innovation data of regulatory authorities through multimodal big models to generate regulatory labels and regulatory data; The machine learning model is trained with existing historical data to establish a trend prediction model, and the monitoring data of dynamic tags is input into the model to predict trend changes.

[0057] The intelligent decision-making and early warning module is used to calculate the correlation based on the generated dynamic tags and regulatory tags, and to provide interactive prompts for the dynamic tags when the designed threshold is reached. At the same time, the data corresponding to the dynamic tags and the regulatory tags are compared based on the interactive prompts, and an early warning prompt is issued when the monitoring data exceeds the range threshold or reference value of the regulatory data.

[0058] The data sharing and interaction module is used to integrate the target agricultural areas, dynamic tags and real-time data into the geographic information system for visual display, wherein the dynamic tags also include interactive prompts, trend forecasts and early warning prompts.

[0059] It should be noted that interactive prompts refer to the data interaction between direct data collected from agricultural production and the decision-making data, socio-economic data, and science and innovation data of regulatory departments by matching dynamic tags and regulatory tags, thereby realizing the visualization of the correlation between regulatory data of different functional departments, and displaying the data on a shared platform in an efficient, real-time and accurate state, providing agricultural producers and decision makers with a convenient data query method.

[0060] The present invention firstly generates dynamic labels for direct agricultural production data through an association rule engine through edge computing modules in different areas, and completes data collection and fusion through distributed processing, which reduces the difficulty of integrating massive data and improves the processing efficiency of real-time data; at the same time, other smart agriculture-related information data are integrated through a multimodal large model to extract regulatory labels and regulatory data, guide the data labeling of monitoring data, and improve the consistency and relevance of data labeling; finally, the dynamic label and the regulatory label correlation calculation are interactively prompted, thereby linking the monitoring data and the regulatory data, and visualizing them based on GIS, thereby improving the interactivity of agricultural data management.

[0061] The present invention establishes a dynamic map of the target agricultural area based on GIS in a visual display, which includes an agricultural area map, dynamic labels and real-time data. The dynamic labels include interactive prompts, trend forecasts and early warning prompts. The interactive prompts provide users with the association status between the current monitoring data and the regulatory data. At the same time, trend change forecasts and early warning prompts are performed for the dynamic labels of the interactive prompts, thereby improving the visualization effect, helping users to understand complex data and analysis results in an intuitive and concise manner, and enhancing real-time response capabilities through early warning prompts, thereby improving decision-making efficiency.

[0062] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. An agricultural data interaction method based on multi-source information, characterized in that: The following steps are involved: Regional agricultural data processing: Divide the target areas for agricultural supervision, set up real-time data collection equipment for each area, process the collected data through the corresponding edge computing module, and generate dynamic labels; Processing of agriculture-related information: Collect relevant information about smart agriculture through access to third-party platforms, analyze and process the information data, and obtain regulatory labels and corresponding regulatory data; Global data visualization: import all tags and corresponding data into the geographic information system, match them according to the corresponding areas and their respective tags, and identify and update the monitoring tags through the cloud platform computing center; The global data visualization includes the following steps: Data matching display: Dynamic tags are displayed in a mobile manner at corresponding locations on the map according to the coordinate system of the data collection device, and interactive prompts are provided by calculating the correlation between dynamic tags and regulatory tags; Deepening of interactive prompts: Call the trend prediction model based on the dynamic tags of interactive prompts, perform feature engineering on the monitoring data corresponding to the dynamic tags, and input them into the trend prediction model to obtain the trend changes of the dynamic tags. At the same time, the monitoring data will be used for early warning prompts based on regulatory data.

2. The agricultural data interaction method based on multi-source information according to claim 1 is characterized in that: The regional agricultural data processing comprises the following steps: Agricultural area division: set up data collection equipment according to data application requirements, set up edge computing modules according to the expansion of data collection, and divide agricultural supervision areas based on edge computing modules; Collect monitoring data: Through the set data collection equipment, each edge computing module obtains real-time data by setting the collection frequency; Dynamic label generation: After the edge computing module pre-processes the real-time data, it generates dynamic labels for the pre-processed data through the association rule engine; Monitoring data upload: The data with dynamic tags assigned and the data without dynamic tags are integrated into corresponding data packets and encrypted before uploading to the cloud computing center.

3. The agricultural data interaction method based on multi-source information according to claim 2 is characterized in that: The dynamic tag generation further includes the following steps: Align and normalize the state data and dynamic tags of agricultural historical data based on time periods, merge the state data based on dynamic tags, and establish a dynamic tag transaction database; The association rule mining algorithm is used to mine the frequent item sets corresponding to each dynamic tag from the dynamic tag transaction database, generate association rules from the frequent item sets, and calculate the support and confidence of the rules; The association rules are screened by setting the minimum support threshold and the confidence threshold to generate a series of association rules for dynamic tags.

4. The agricultural data interaction method based on multi-source information according to claim 1 is characterized in that: The processing of the agricultural related information comprises the following steps: Collect relevant information: obtain decision-making data, socio-economic data, and science and technology and innovation data from regulatory authorities through third-party platforms; Generate regulatory labels: Use a multimodal large model to process relevant information data, fuse data from different modalities, and generate corresponding regulatory labels; Extract regulatory data: Obtain the range threshold of the corresponding data type based on the generated regulatory labels. For regulatory labels that cannot be directly obtained from relevant information, use a multimodal large model to infer and supplement reference values; Monitoring data labeling: For monitoring data uploaded to the cloud computing center, labels are annotated by data type and by calculating the similarity between the range threshold or reference value of the monitoring data and the regulatory data.

5. The agricultural data interaction method based on multi-source information according to claim 1 is characterized in that: The global data visualization further includes the following steps: Import multi-source data: All relevant data of the target agricultural area are uniformly converted into a format supported by the geographic information system, and descriptive metadata is added to the data and stored in the cloud computing center. The pre-processed data is imported into the GIS for visualization; Association rule update: The cloud computing center collects monitoring data, and adds it to the dynamic label transaction database after establishing a complete data format by labeling the monitoring data. It updates the association rules through association mining and sends them to the edge computing module.

6. The agricultural data interaction method based on multi-source information according to claim 1 is characterized in that: The interactive prompt deepening includes the following steps: Integrate the monitoring data used for dynamic label generation into a dataset, and perform feature engineering on the integrated dataset, including feature extraction, feature selection, and feature construction; Select a machine learning model as the basis of the trend prediction model, use the existing historical data to train the selected model, use the fusion results of multi-source data as input features, and use the trend changes of dynamic labels as output; The data corresponding to the dynamic label and the regulatory label are compared based on the interactive prompts, and an early warning prompt is issued when the monitoring data exceeds the range threshold or reference value of the regulatory data.

7. An agricultural data interaction platform based on multi-source information, used to implement an agricultural data interaction method based on multi-source information as claimed in any one of claims 1 to 6, characterized in that: It includes data collection and fusion module, data storage and management module, data analysis and mining module, intelligent decision-making and early warning module and data sharing and interaction module; The data collection and fusion module, the data storage and management module, the data analysis and mining module, the intelligent decision-making and early warning module and the data sharing and interaction module are sequentially connected in communication; The data collection and fusion module is used to collect and pre-process multi-source agricultural data, and then fuse and associate heterogeneous data from different data sources; The fusion and association include the following steps: Data collection: Using intelligent sensing technology through the set data collection equipment, each edge computing module obtains real-time data by setting the collection frequency, including meteorological sensors, soil sensors, cameras, drones and satellite remote sensing; Data preprocessing: Thoroughly clean the collected raw data, remove invalid, erroneous or incomplete records, perform data standardization, and convert data of different dimensions and scales into a unified data format that can be used for association rule engine analysis; Data fusion: The pre-processed data is input into the association rule engine to generate dynamic labels, complete the fusion of multi-source data, and perform subsequent prediction and visualization.

8. The agricultural data interactive platform based on multi-source information according to claim 7, characterized in that: The data storage and management module is used to store the collected and collected data in a database and select a cloud computing center to realize parallel storage and access of data.

9. The agricultural data interactive platform based on multi-source information according to claim 7, characterized in that: The data analysis and mining module is used to analyze and process all data through the cloud computing center, including: A rule engine that mines agricultural historical data and corresponding status through association rule mining algorithms to obtain a series of dynamic labels; Analyze the decision-making data, socio-economic data, and science and innovation data of regulatory authorities through multimodal big models to generate regulatory labels and regulatory data; The machine learning model is trained with existing historical data to establish a trend prediction model, and the monitoring data of dynamic tags is input into the model to predict trend changes.

10. The agricultural data interactive platform based on multi-source information according to claim 7, characterized in that: The intelligent decision-making and early warning module is used to calculate the correlation between the generated dynamic tags and regulatory tags, and to interactively prompt the dynamic tags when the designed threshold is reached. At the same time, the data corresponding to the dynamic tags and the regulatory tags are compared according to the interactive prompts, and an early warning prompt is issued when the monitoring data exceeds the range threshold or reference value of the regulatory data; The data sharing and interaction module is used to integrate the target agricultural areas, dynamic tags and real-time data into the geographic information system for visual display, wherein the dynamic tags also include interactive prompts, trend forecasts and early warning prompts.

Images