Agricultural area monitoring information analysis method and system applied to digital agriculture

Abstract

The present invention relates to the field of smart agricultural technology, and provides a method and system for analyzing agricultural area monitoring information applied to digital agriculture, which can not only effectively integrate and utilize various agricultural area monitoring information, improve the efficiency and accuracy of data processing, but also can quickly determine the potential production risk labels of target agricultural areas under different monitoring tasks based on the global agricultural monitoring integrated knowledge, and provide timely and accurate decision-making basis for agricultural managers. This is of great significance for improving the intelligent level of agricultural production and realizing refined agricultural management.

Description

Technical Field

[0001] The present invention relates to the field of smart agriculture technology, and in particular to a method and system for analyzing crop area monitoring information applied to digital agriculture. Background Art

[0002] With the development of agricultural science and technology, refined agricultural management has become the key to improving agricultural production efficiency and quality. To achieve this goal, comprehensive monitoring of farming areas is an essential link. Existing agricultural monitoring technology has been able to obtain a large amount of farming area monitoring information through various sensors and remote sensing equipment, including IoT sensing farming environment monitoring logs, airborne remote sensing growth monitoring logs, and farming area positioning monitoring logs. These monitoring information covers soil, climate, crop growth status, pests and diseases, and other aspects, providing a rich data foundation for agricultural management.

[0003] However, existing agricultural monitoring technologies still have certain deficiencies in data processing and analysis. On the one hand, due to the diverse sources and formats of monitoring data, how to effectively extract and integrate the key elements of this information to form valuable monitoring knowledge is an urgent problem to be solved. On the other hand, traditional monitoring and analysis methods can often only process single or limited monitoring tasks and are difficult to adapt to the complex and changing agricultural production environment. Summary of the invention

[0004] In order to improve the above problems, the present invention provides a method and system for analyzing crop area monitoring information applied to digital agriculture.

[0005] A first aspect of an embodiment of the present invention provides a method for analyzing crop area monitoring information applied to digital agriculture, which is applied to a digital agriculture information analysis system. The method includes:

[0006] Collecting monitoring information of the target farming area; the monitoring information of the target farming area includes IoT sensing farming environment monitoring logs, airborne remote sensing growth monitoring logs and farming area positioning monitoring logs;

[0007] Performing monitoring element mining on each crop growth monitoring data in the multiple groups of crop growth monitoring data corresponding to the airborne remote sensing growth monitoring log to obtain crop growth monitoring element knowledge for each crop growth monitoring data; and performing regional monitoring element mining on the multiple regional blocks corresponding to the crop area positioning monitoring log to obtain regional block monitoring element knowledge for each regional block;

[0008] The agricultural environment monitoring element knowledge corresponding to the IoT sensor agricultural environment monitoring log, the crop growth monitoring element knowledge of all crop growth monitoring data, and the regional block monitoring element knowledge of all regional blocks are respectively transformed to obtain agricultural environment monitoring transformation knowledge, crop growth monitoring transformation knowledge, and agricultural area monitoring transformation knowledge;

[0009] Performing AI knowledge integration on the farming environment monitoring transformation knowledge, the crop growth monitoring transformation knowledge and the farming area monitoring transformation knowledge to obtain global agricultural monitoring integrated knowledge under the same knowledge relationship network;

[0010] Based on the global agricultural monitoring integrated knowledge, a potential production risk label of the target farming area monitoring information under at least one set monitoring task is determined.

[0011] Preferably, the agricultural environment monitoring element knowledge corresponding to the IoT sensor agricultural environment monitoring log, the crop growth monitoring element knowledge of all crop growth monitoring data, and the regional block monitoring element knowledge of all regional blocks are respectively transformed to obtain agricultural environment monitoring transformation knowledge, crop growth monitoring transformation knowledge, and agricultural area monitoring transformation knowledge, including:

[0012] Performing agricultural environment monitoring knowledge transformation on the agricultural environment monitoring element knowledge corresponding to the IoT sensor agricultural environment monitoring log to obtain agricultural environment monitoring transformation knowledge;

[0013] Performing growth monitoring knowledge transformation on crop growth monitoring element knowledge of all crop growth monitoring data to obtain crop growth monitoring transformation knowledge;

[0014] The regional monitoring knowledge is transformed into regional monitoring knowledge for all regional blocks to obtain the crop area monitoring transformation knowledge.

[0015] Preferably, the agricultural environment monitoring knowledge transformation is performed on the agricultural environment monitoring element knowledge corresponding to the IoT sensor agricultural environment monitoring log to obtain the agricultural environment monitoring transformation knowledge, including:

[0016] Setting mask features in the first log section and the second log section of the IoT sensor farming environment monitoring log respectively to form an IoT sensor farming environment monitoring mask log;

[0017] Embedding knowledge features into the mask log of the IoT sensor farming environment monitoring to obtain the farming environment monitoring element knowledge;

[0018] The agricultural environment monitoring element knowledge is transformed into agricultural environment monitoring knowledge through the first knowledge transformation subnet to obtain the agricultural environment monitoring transformation knowledge.

[0019] Preferably, the first knowledge transformation subnet includes a first residual branch and a value mapping branch; the agricultural environment monitoring knowledge transformation is performed on the agricultural environment monitoring element knowledge through the first knowledge transformation subnet to obtain the agricultural environment monitoring transformation knowledge, including:

[0020] Performing residual connection on the farming environment monitoring element knowledge through the first residual branch to obtain residual connection features;

[0021] By configuring the numerical mapping branch with fixed neural network variables, the residual connection features are processed by interval numerical mapping to obtain the agricultural environment monitoring transformation knowledge; the agricultural environment monitoring transformation knowledge is a knowledge vector in a multidimensional knowledge relationship network.

[0022] Preferably, the crop growth monitoring knowledge transformation is performed on the crop growth monitoring element knowledge of all crop growth monitoring data to obtain the crop growth monitoring transformation knowledge, including:

[0023] Combining the crop growth monitoring element knowledge of all the crop growth monitoring data according to the time sequence priority of each of the crop growth monitoring data in the monitoring information of the target farming area to obtain the crop growth monitoring element combination knowledge;

[0024] The knowledge of the combination of crop growth monitoring elements is refined through a deep learning subnet to obtain target knowledge features;

[0025] The target knowledge feature is transformed into growth monitoring knowledge through the second knowledge transformation subnet to obtain the crop growth monitoring transformation knowledge.

[0026] Preferably, the regional monitoring knowledge transformation is performed on the regional block monitoring element knowledge of all regional blocks to obtain the crop area monitoring transformation knowledge, including:

[0027] According to the time sequence priority of each of the regional blocks in the log block corresponding to the agricultural area positioning monitoring log, the regional block monitoring element knowledge of all the regional blocks is combined to obtain the regional block monitoring combination feature;

[0028] Performing regional feature embedding processing on the regional block monitoring combined features through a regional feature embedding layer to obtain regional feature embedding knowledge;

[0029] The regional feature embedding knowledge is transformed into regional monitoring knowledge through the third knowledge transformation subnet to obtain the crop area monitoring transformation knowledge.

[0030] Preferably, the AI ​​knowledge integration of the farming environment monitoring transformation knowledge, the crop growth monitoring transformation knowledge and the farming area monitoring transformation knowledge is performed to obtain the global agricultural monitoring integrated knowledge under the same knowledge relationship network, including:

[0031] The agricultural environment monitoring transformation knowledge, the crop growth monitoring transformation knowledge and the agricultural area monitoring transformation knowledge are processed by combining the linked feature dimensions through a second residual branch configured with fixed neural network variables to obtain global agricultural monitoring integrated knowledge under the same knowledge relationship network; the global agricultural monitoring integrated knowledge is a linked feature dimension combination vector under the same knowledge relationship network.

[0032] Preferably, determining the potential production risk label of the target farming area monitoring information under at least one set monitoring task based on the global agricultural monitoring integrated knowledge includes:

[0033] Determining a discrimination possibility score of the target farming area monitoring information under at least one set monitoring task based on the global agricultural monitoring integrated knowledge;

[0034] The potential production risk label of the monitoring information of the target farming area is determined according to the discrimination possibility score under the at least one set monitoring task.

[0035] Preferably, the determining of the discrimination possibility score of the target farming area monitoring information under at least one set monitoring task based on the global agricultural monitoring integrated knowledge includes:

[0036] Performing decision analysis on the global agricultural monitoring integrated knowledge through a decision subnet to obtain a decision analysis vector;

[0037] The discrimination subnet determines the discrimination possibility score of the target farming area monitoring information under at least one set monitoring task according to the decision analysis vector.

[0038] Preferably, the agricultural area monitoring information analysis method applied to digital agriculture is implemented by a long short-term memory network; the long short-term memory network is debugged by the following steps:

[0039] Acquire a crop area monitoring training set; the crop area monitoring training set includes past target crop area monitoring information and production risk label samples of the past target crop area monitoring information under a set monitoring task; the past target crop area monitoring information includes past IoT sensor crop environment monitoring logs, past airborne remote sensing growth monitoring logs, and past crop area positioning monitoring logs;

[0040] Loading the past target farming area monitoring information into the long short-term memory network;

[0041] By using the first monitoring analysis subnet of the long short-term memory network, each of the multiple groups of past crop growth monitoring data corresponding to the past airborne remote sensing growth monitoring log is subjected to monitoring element mining to obtain past crop growth monitoring element knowledge of each past crop growth monitoring data;

[0042] and performing regional monitoring element mining on a plurality of past regional blocks corresponding to the past agricultural area positioning monitoring logs to obtain past regional block monitoring element knowledge of each past regional block;

[0043] Through the second monitoring and analysis subnet of the long short-term memory network, the past agricultural environment monitoring element knowledge corresponding to the past IoT sensor agricultural environment monitoring log, the past crop growth monitoring element knowledge of all past crop growth monitoring data, and the past regional block monitoring element knowledge of all past regional blocks are respectively transformed to obtain past agricultural environment monitoring transformation knowledge, past crop growth monitoring transformation knowledge, and past agricultural area monitoring transformation knowledge;

[0044] Through the third monitoring analysis subnet of the long short-term memory network, AI knowledge integration is performed on the past agricultural environment monitoring transformation knowledge, the past crop growth monitoring transformation knowledge and the past agricultural area monitoring transformation knowledge to obtain past global agricultural monitoring integrated knowledge under the same knowledge relationship network;

[0045] And, based on the past global agricultural monitoring integrated knowledge, determining the past potential production risk label of the past target farming area monitoring information under the set monitoring task;

[0046] Loading the past potential production risk labels and the production risk label samples into the training error processing subnet to determine the training error and obtain training error information;

[0047] The network parameters in the second monitoring and analysis subnet and the third monitoring and analysis subnet are optimized according to the training error information to obtain a debugged long short-term memory network.

[0048] According to a second aspect of an embodiment of the present invention, a digital agriculture information analysis system is provided, comprising: a processor, a memory and a bus connected to the processor; the processor and the memory communicate with each other via the bus; the processor is used to call a computer program in the memory to execute the above-mentioned agricultural area monitoring information analysis method applied to digital agriculture.

[0049] According to a third aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which a program is stored, and when the program is executed by a processor, the above-mentioned method for analyzing agricultural area monitoring information applied to digital agriculture is implemented.

[0050] The method and system for analyzing agricultural area monitoring information applied to digital agriculture provided by the embodiment of the present invention first conducts in-depth monitoring element mining on each group of crop growth monitoring data in the airborne remote sensing growth monitoring log, and extracts the monitoring element knowledge closely related to crop growth; at the same time, regional monitoring element mining is performed on each regional block in the agricultural area positioning monitoring log to obtain the environment, soil and other element information in each block. Then, the present invention performs knowledge transformation processing on the knowledge corresponding to the IoT sensing agricultural environment monitoring log, crop growth monitoring data and regional block monitoring data, and converts this knowledge into a form that is more suitable for integration and analysis. Finally, through AI knowledge integration technology, the transformed various types of monitoring knowledge are integrated into a global agricultural monitoring integrated knowledge base, so as to more comprehensively and accurately reflect the actual situation of the agricultural area and provide strong data support for subsequent monitoring tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0052] Figure 1 A flowchart of a method for analyzing crop area monitoring information applied to digital agriculture provided by an embodiment of the present invention.

[0053] Figure 2 A schematic diagram of a product module of a digital agricultural information analysis system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0054] In response to the above problems, the present invention proposes a method for analyzing agricultural area monitoring information based on AI knowledge integration. The method first conducts in-depth monitoring element mining on each group of crop growth monitoring data in the airborne remote sensing growth monitoring log, and extracts the monitoring element knowledge closely related to crop growth; at the same time, regional monitoring element mining is performed on each regional block in the agricultural area positioning monitoring log to obtain the environment, soil and other element information in each block. Next, the present invention performs knowledge transformation processing on the knowledge corresponding to the IoT sensing agricultural environment monitoring log, crop growth monitoring data and regional block monitoring data, and converts this knowledge into a form that is more suitable for integration and analysis. Finally, through AI knowledge integration technology, the transformed various types of monitoring knowledge are integrated into a global agricultural monitoring integrated knowledge base, so that the actual situation of the agricultural area can be more comprehensively and accurately reflected, and strong data support can be provided for subsequent monitoring tasks.

[0055] Through this method, the present invention can not only effectively integrate and utilize various agricultural area monitoring information to improve the efficiency and accuracy of data processing, but also can quickly determine the potential production risk labels of the target agricultural area under different monitoring tasks based on the global agricultural monitoring integrated knowledge, providing agricultural managers with timely and accurate decision-making basis. This is of great significance for improving the intelligent level of agricultural production and realizing refined agricultural management.

[0056] The exemplary embodiments disclosed in the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0057] In order to better understand the above technical scheme, the technical scheme of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical scheme of the present invention, rather than limitations on the technical scheme of the present invention. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0058] See also Figure 1 , which is a flow chart of a method for analyzing crop area monitoring information applied to digital agriculture provided by an embodiment of the present invention. The method is applied to a digital agriculture information analysis system, and the specific content description of the method includes S110-S150.

[0059] S110, collecting monitoring information of the target farming area; the monitoring information of the target farming area includes IoT sensing farming environment monitoring log, airborne remote sensing growth monitoring log and farming area positioning monitoring log.

[0060] S120. Perform monitoring element mining on each crop growth monitoring data in the multiple groups of crop growth monitoring data corresponding to the airborne remote sensing growth monitoring log to obtain crop growth monitoring element knowledge for each crop growth monitoring data; and perform regional monitoring element mining on the multiple regional blocks corresponding to the crop area positioning monitoring log to obtain regional block monitoring element knowledge for each regional block.

[0061] S130, respectively perform knowledge transformation on the agricultural environment monitoring element knowledge corresponding to the IoT sensor agricultural environment monitoring log, the crop growth monitoring element knowledge of all crop growth monitoring data, and the regional block monitoring element knowledge of all regional blocks, and obtain agricultural environment monitoring transformation knowledge, crop growth monitoring transformation knowledge, and agricultural area monitoring transformation knowledge accordingly.

[0062] S140, performing AI knowledge integration on the farming environment monitoring transformation knowledge, the crop growth monitoring transformation knowledge and the farming area monitoring transformation knowledge to obtain global agricultural monitoring integrated knowledge under the same knowledge relationship network.

[0063] S150. Determine, based on the global agricultural monitoring integrated knowledge, a potential production risk label of the target farming area monitoring information under at least one set monitoring task.

[0064] For example, the digital agricultural information analysis system (also known as the intelligent agricultural monitoring system) is conducting detailed monitoring and management of a vast farmland. The farmland is divided into multiple areas, each with its own unique soil, climate and irrigation conditions. In order to ensure that crops can grow healthily and have a good harvest, the system needs to grasp various information about the farmland in real time and detect potential production risks in a timely manner.

[0065] First, the system continuously monitors environmental parameters in the farmland, such as temperature, humidity, light intensity, and soil nutrients, through a network of IoT sensors. These sensors are carefully placed in every corner of the farmland, constantly collecting data and transmitting the data in real time to the system's data center. The data center processes and analyzes the data, generating an IoT sensor farming environment monitoring log that records in detail the changes in the farmland environment.

[0066] At the same time, the system also uses remote sensing equipment carried by drones to regularly take aerial photos of farmland. Drones fly over farmland and capture the growth status of crops through high-definition cameras and multispectral sensors. These images and data are transmitted back to the data center. After image processing and data analysis, the system generates an airborne remote sensing growth monitoring log. This log describes the growth of crops in detail, including leaf color, plant height, growth rate, and whether there are diseases and pests.

[0067] In addition, the system uses GPS technology to accurately locate farmland in different areas. Each area block is assigned a unique location tag, and the system can accurately know the specific location and boundaries of each block. These positioning data are recorded in the farm area positioning monitoring log, providing a basis for subsequent monitoring and analysis.

[0068] After collecting this monitoring information, the system begins in-depth data analysis and mining. For each set of crop growth monitoring data in the airborne remote sensing growth monitoring log, the system extracts key information about the crop growth status, such as growth rate, leaf color, etc., and analyzes the relationship between this information and the health status of the crop. At the same time, the system also mines monitoring elements for each regional block in the agricultural area positioning monitoring log, determines the regional characteristics of each block, such as soil type and terrain height, and analyzes the impact of these characteristics on crop growth.

[0069] In order to effectively integrate these data from different sources and in different formats, the system performs knowledge transformation. It transforms and uniformly represents the knowledge in the IoT sensor farming environment monitoring log, airborne remote sensing growth monitoring log, and farming area positioning monitoring log, and obtains the farming environment monitoring transformation knowledge, crop growth monitoring transformation knowledge, and farming area monitoring transformation knowledge. This knowledge is stored in the system's knowledge base in the same format and semantics, which facilitates subsequent knowledge integration and analysis.

[0070] Next, the system uses artificial intelligence technology to integrate this transformed knowledge. It builds a global agricultural monitoring integrated knowledge network to comprehensively analyze the mutual influence and correlation between farmland environment, crop growth status and regional block conditions. Through this knowledge network, the system can fully understand the overall situation of the farmland and discover the potential connections and laws between different monitoring elements.

[0071] Finally, based on the integrated knowledge of global agricultural monitoring, the system conducts an in-depth analysis of the monitoring information of the target farming area to determine the potential production risks that may exist under the set monitoring tasks. For example, the system may find that the soil nutrients in a certain area are insufficient, resulting in slow growth of crops; or that a certain pest is spreading in a certain crop growing area, threatening the health of crops. In response to these potential risks, the system promptly generates corresponding production risk labels and reminds farm managers to take necessary intervention measures, such as increasing the amount of fertilizer and spraying pesticides, to ensure the normal growth and harvest of crops.

[0072] In S110, the IoT sensing farming environment monitoring log refers to a log that collects and records farmland environmental parameters through IoT sensor technology. IoT sensors can sense and measure various environmental factors in farmland in real time, such as temperature, humidity, light, wind speed, etc., and transmit these data to the data center for processing and analysis. For example, temperature sensors deployed in farmland can continuously monitor and record temperature changes in soil and air. These data are integrated into the IoT sensing farming environment monitoring log to analyze the environmental conditions for crop growth.

[0073] Airborne remote sensing growth monitoring logs refer to logs generated by remote sensing equipment carried by drones or other aircraft to monitor the growth status of farmland. Remote sensing equipment can capture high-resolution images and spectral information of farmland. By analyzing these data, the growth status, health status, and yield forecast of crops can be evaluated. For example, drones regularly fly over farmland and collect images and data of crops through their cameras and multispectral sensors. After processing, these images and data are recorded in airborne remote sensing growth monitoring logs to monitor the growth rate of crops and whether there are problems such as pests and diseases.

[0074] The crop area positioning monitoring log refers to the log generated by accurately locating and zoning farmland using GPS or other positioning technologies. Through positioning monitoring, the exact location and boundaries of each area block in the farmland can be determined, thereby achieving refined management of the farmland. For example, in the farmland, GPS equipment is used to accurately measure and record the coordinates and area of ​​each area block. This information is stored in the crop area positioning monitoring log and used for subsequent farmland management and operation guidance, such as precision fertilization, irrigation, and pest control.

[0075] These logs together constitute the "target farming area monitoring information", providing the smart agricultural monitoring system with comprehensive and accurate farmland data and crop growth status information, thereby supporting intelligent management and decision-making of farmland.

[0076] In S120, the crop growth monitoring data refers to specific data on the growth status of crops collected through remote sensing or other monitoring means. These data may include the growth rate, leaf color, plant height, leaf area index, etc. of crops, which are used to evaluate the growth and health of crops. For example, through remote sensing monitoring, a set of data on corn growth is obtained, including the height of each corn plant, the greenness of the leaves, etc. These data are crop growth monitoring data, which can be used to analyze the growth of corn and whether there are any growth abnormalities.

[0077] Monitoring factor mining refers to the process of extracting key information or features from a large amount of monitoring data. These key information or features can reflect the important status and change patterns of the monitored object, and provide support for subsequent decision-making and analysis. In the crop growth monitoring data, data mining technology can be used to analyze which factors are the key factors affecting corn growth, such as soil moisture, light intensity, etc. This process is monitoring factor mining, which helps find the important monitoring factors that affect crop growth.

[0078] Crop growth monitoring factor knowledge refers to the knowledge of key information or characteristics about crop growth obtained through monitoring factor mining. This knowledge describes the important states and changing patterns in the growth process of crops, and provides a basis for farmland management and decision-making. For example, through monitoring factor mining, it was found that soil moisture and light intensity are the key factors affecting corn growth. This discovery can be summarized as crop growth monitoring factor knowledge, which can be used to guide subsequent farmland management and regulatory measures.

[0079] Regional monitoring factor mining refers to the process of extracting key information or features about regional blocks from the agricultural area positioning monitoring log. These key information or features can reflect the uniqueness of each regional block and its impact on crop growth, providing support for subsequent farmland management and decision-making. For example, in the agricultural area positioning monitoring log, data mining technology can be used to analyze which regional blocks have significant effects on crop growth, such as soil type and terrain height. This process is regional monitoring factor mining, which helps find regional key factors that affect crop growth.

[0080] Regional block monitoring element knowledge refers to the knowledge about the key information or characteristics of each regional block obtained through regional monitoring element mining. This knowledge describes the uniqueness of each regional block and the law of its influence on crop growth, providing a basis for the precise management and regulation of farmland. For example, through regional monitoring element mining, it is found that the soil type of a certain regional block is particularly suitable for growing a certain crop. This discovery can be summarized as regional block monitoring element knowledge, which is used to guide subsequent precise planting and management measures for farmland.

[0081] In S130, the farming environment monitoring element knowledge refers to the key information and features about the farmland environment extracted from the IoT sensing farming environment monitoring log. This knowledge describes the state and change pattern of the farmland environment and has an important impact on the growth of crops. By analyzing the IoT sensing farming environment monitoring log, key information such as the temperature change trend and light intensity distribution in the farmland can be extracted. This knowledge constitutes the farming environment monitoring element knowledge, which is used to evaluate the impact of the farmland environment on crop growth.

[0082] Regional segmentation refers to dividing farmland into multiple smaller areas, each with similar environmental conditions and crop growth characteristics. This division helps to achieve refined management and regulation of farmland. In farmland, farmland can be divided into multiple regional segments based on factors such as soil type and terrain height. The crop growth environment in each regional segment is similar, which facilitates targeted management measures.

[0083] Knowledge transformation refers to the process of converting and uniformly representing knowledge from different sources and formats. In this process, the original knowledge is converted into a form that is easier to store, transmit and analyze for subsequent knowledge integration and application. In the farmland monitoring system, the raw data in the IoT sensor farming environment monitoring log can be converted into a standardized data format to facilitate integration and analysis with data from other sources. This process is an application of knowledge transformation.

[0084] The transformation knowledge of agricultural environment monitoring, the transformation knowledge of crop growth monitoring, and the transformation knowledge of agricultural area monitoring refer to the transformation knowledge of farmland environment, crop growth, and agricultural area blocks obtained after knowledge transformation. These knowledge are expressed in a unified form to facilitate subsequent knowledge integration and application. For example, by performing knowledge transformation on the agricultural environment monitoring log of IoT sensors, transformation knowledge such as temperature and humidity in the farmland environment can be obtained; by performing knowledge transformation on the crop growth monitoring data, transformation knowledge such as crop growth rate and leaf color can be obtained; by performing knowledge transformation on the agricultural area positioning monitoring log, transformation knowledge such as soil type and terrain height of each area block can be obtained. These knowledge together constitute the transformation knowledge base of the farmland monitoring system, providing support for subsequent farmland management and decision-making.

[0085] In other examples, the feature vector of agricultural environment monitoring factor knowledge is: [temperature, humidity, light intensity, soil pH, soil nutrient content]; this feature vector represents the key monitoring factors of the farmland environment, including temperature, humidity, light intensity, soil pH and soil nutrient content. These factors have an important impact on the growth of crops and need to be continuously monitored and recorded.

[0086] An example of a feature vector of crop growth monitoring element knowledge is: [plant height, leaf area index, leaf color (RGB value), growth rate, degree of pests and diseases]; this feature vector describes the key monitoring elements of crop growth, including plant height, leaf area index, leaf color (expressed by RGB value), growth rate, and degree of pests and diseases. These elements reflect the growth status and health of crops and are important indicators for evaluating crop growth.

[0087] An example of a feature vector of regional block monitoring element knowledge is: [soil type, terrain height, irrigation conditions, historical yields, pest and disease history]; this feature vector represents the key monitoring elements of farmland regional blocks, including soil type, terrain height, irrigation conditions, historical yields, and pest and disease history. These elements reflect the uniqueness of each regional block and its impact on crop growth, and are an important basis for precise farmland management.

[0088] An example of feature vector transformation of agricultural environment monitoring transformation knowledge is: standardize the original environmental monitoring data (such as temperature, humidity, etc.) to obtain standardized feature vectors, such as [standardized temperature, standardized humidity, standardized light intensity, standardized soil pH, standardized soil nutrient content]. This transformation helps to eliminate the dimensional differences between different monitoring elements, so that they can be compared and analyzed on the same scale.

[0089] An example of feature vector transformation of crop growth monitoring transformation knowledge is: normalize crop growth monitoring data (such as plant height, leaf area index, etc.) to obtain normalized feature vectors, such as [normalized plant height, normalized leaf area index, normalized leaf color, normalized growth rate, normalized degree of pests and diseases]. Normalization can map the value ranges of different monitoring elements to the same range (usually between 0 and 1), which helps to conduct unified analysis and processing.

[0090] An example of feature vector transformation of crop area monitoring transformation knowledge is: discretize the monitoring elements of regional blocks, convert continuous values ​​into discrete categories, such as dividing soil types into sandy soil, loam, clay and other categories, and obtain discretized feature vectors, such as [soil type category, terrain height category, irrigation condition category, historical yield level, historical level of pests and diseases]. Discretization helps to simplify the process of data analysis and model building, and can more intuitively show the uniqueness of each regional block and its impact on crop growth.

[0091] These transformation knowledge are all for better use and analysis of monitoring data, extracting useful information to support farmland management and decision-making. Through appropriate transformation processing, the original data can be converted into a form that is more convenient for analysis and application, thus improving the effectiveness and efficiency of the farmland monitoring system.

[0092] First, by collecting multi-dimensional monitoring information of the target farming area, including IoT sensing farming environment monitoring logs, airborne remote sensing growth monitoring logs, and farming area positioning monitoring logs, the present invention achieves a comprehensive perception of the farmland environment. This comprehensive data collection method not only improves the richness and accuracy of monitoring data, but also provides a solid foundation for subsequent data analysis and knowledge mining.

[0093] Secondly, the present invention conducts in-depth monitoring element mining on various types of collected monitoring data. By mining the crop growth monitoring data in the airborne remote sensing growth monitoring log, detailed knowledge about crop growth is obtained; at the same time, the regional blocks in the agricultural area positioning monitoring log are mined to reveal the differences and particularities between different regions. The mined knowledge provides strong support for the refined management of farmland.

[0094] Furthermore, the present invention performs knowledge transformation processing on the mined monitoring element knowledge. By transforming the IoT sensing farming environment monitoring knowledge, crop growth monitoring knowledge and regional block monitoring knowledge, various types of monitoring transformation knowledge under a unified knowledge framework are obtained. This knowledge transformation not only eliminates the knowledge heterogeneity between different data sources, but also improves the availability and integrability of knowledge.

[0095] On this basis, the present invention integrates the transformed monitoring knowledge into global agricultural monitoring integrated knowledge through AI knowledge integration technology. This integrated knowledge not only contains comprehensive information about the farmland environment, but also reflects the inherent connection and law between crop growth and regional division. This provides farmland managers with a comprehensive and accurate agricultural knowledge base, which helps guide scientific management and decision-making of farmland.

[0096] Finally, based on the global agricultural monitoring integrated knowledge, the present invention determines the potential production risk labels of the target farming area under different set monitoring tasks. These risk labels provide early warning information for farmland managers, helping to timely discover and respond to potential production risks, thereby ensuring the safety and efficient production of farmland.

[0097] In summary, the present invention realizes comprehensive perception and fine management of farmland environment through a series of innovative processes such as collecting multi-dimensional monitoring data, mining monitoring element knowledge, transforming and integrating knowledge, and determining potential production risk labels. This not only improves the scientific and intelligent level of farmland management, but also makes positive contributions to ensuring global food security and sustainable agricultural development.

[0098] In some possible embodiments, the agricultural environment monitoring element knowledge corresponding to the Internet of Things sensor agricultural environment monitoring log, the crop growth monitoring element knowledge of all crop growth monitoring data, and the regional block monitoring element knowledge of all regional blocks are respectively transformed to obtain agricultural environment monitoring transformation knowledge, crop growth monitoring transformation knowledge, and agricultural area monitoring transformation knowledge, including: performing agricultural environment monitoring knowledge transformation on the agricultural environment monitoring element knowledge corresponding to the Internet of Things sensor agricultural environment monitoring log to obtain agricultural environment monitoring transformation knowledge; performing growth monitoring knowledge transformation on the crop growth monitoring element knowledge of all crop growth monitoring data to obtain crop growth monitoring transformation knowledge; performing regional monitoring knowledge transformation on the regional block monitoring element knowledge of all regional blocks to obtain agricultural area monitoring transformation knowledge.

[0099] In some possible embodiments, the system performs a detailed knowledge transformation process on the IoT sensor farming environment monitoring log, crop growth monitoring data, and regional block monitoring element knowledge. This process is to transform the raw data into more meaningful information for subsequent analysis and application.

[0100] Specifically, the system first processes the knowledge of agricultural environment monitoring elements corresponding to the IoT sensor agricultural environment monitoring logs. These logs usually contain key data such as temperature, humidity, light intensity, soil nutrients, etc. in the farmland environment. The system transforms these raw data into a form that is easier to understand and analyze through agricultural environment monitoring knowledge transformation. For example, the system may convert temperature data into a temperature trend graph, or convert soil nutrient data into nutrient abundance and deficiency indicators. Such transformations help to quickly identify the changing patterns of the farmland environment and provide support for subsequent agricultural production decisions.

[0101] Next, the system transforms the crop growth monitoring element knowledge of all crop growth monitoring data into growth monitoring knowledge. These monitoring data usually include the growth rate, leaf color, plant height, leaf area index, etc. of crops. The system uses advanced algorithms and models to process and analyze these data to extract the key characteristics and laws of crop growth. For example, the system can predict the health of crops by analyzing changes in leaf color, or evaluate the growth rate of crops by comparing plant height data at different time points. Such transformation makes crop growth data more intuitive and useful, providing a powerful tool for precision agricultural management.

[0102] Finally, the system transforms the regional monitoring knowledge of all regional blocks into regional monitoring knowledge. In farmland, different areas may present different characteristics due to factors such as soil type, terrain height, and irrigation conditions. By processing and analyzing the monitoring data of these regional blocks, the system reveals the uniqueness of each area and its impact on crop growth. For example, the system can divide farmland into different management areas according to soil type and terrain height, and formulate corresponding management measures for each area. Such transformation helps to achieve precise management of farmland and improve agricultural production efficiency and benefits.

[0103] Through the above knowledge transformation processing, the system obtains agricultural environment monitoring transformation knowledge, crop growth monitoring transformation knowledge and agricultural area monitoring transformation knowledge. This knowledge is presented in a more intuitive, easier to understand and apply form, providing strong decision-making support for farmland managers. At the same time, this also reflects the wide application and far-reaching impact of modern technology in the field of agriculture.

[0104] Under some preferred design ideas, the agricultural environment monitoring element knowledge corresponding to the Internet of Things sensor agricultural environment monitoring log is transformed into agricultural environment monitoring knowledge to obtain the agricultural environment monitoring transformation knowledge, including: setting mask features in the first log paragraph and the second log paragraph of the Internet of Things sensor agricultural environment monitoring log respectively to form an Internet of Things sensor agricultural environment monitoring mask log; embedding knowledge features in the Internet of Things sensor agricultural environment monitoring mask log to obtain the agricultural environment monitoring element knowledge; transforming the agricultural environment monitoring element knowledge into agricultural environment monitoring knowledge through the first knowledge transformation subnet to obtain the agricultural environment monitoring transformation knowledge.

[0105] Under some preferred design ideas, the system uses a more sophisticated and efficient method to process the IoT sensor agricultural environment monitoring log to obtain agricultural environment monitoring transformation knowledge. This process involves multiple steps, each of which aims to extract and transform useful information and ultimately serve agricultural production decisions.

[0106] First, the system selects specific paragraphs in the IoT sensor farming environment monitoring log for processing. These paragraphs may contain data on key environmental parameters such as temperature, humidity, and light. In order to protect the integrity and security of the original data and enhance the processability of the data, the system sets mask features in the first log paragraph and the second log paragraph respectively. The mask features can be set by replacing some sensitive data, adding noise, or performing other forms of data perturbation to ensure effective analysis without exposing the original data. In this way, the system generates the IoT sensor farming environment monitoring mask log, which is ready for subsequent processing.

[0107] Next, the system embeds knowledge features into the masked logs of IoT sensor farming environment monitoring. This process aims to extract meaningful knowledge of environmental monitoring elements from the logs. The system may use various machine learning algorithms and natural language processing techniques to analyze and interpret the text and numerical data in the logs. For example, the system can identify patterns in the logs describing temperature changes, the correlation between humidity and light, etc. After these knowledge features are embedded, structured knowledge of farming environment monitoring elements is formed, which provides a basis for subsequent knowledge transformation.

[0108] Finally, the system transforms the knowledge of agricultural environment monitoring elements through the first knowledge transformation subnet. The first knowledge transformation subnet may be a specially designed neural network or other type of machine learning model, which can further transform and abstract the input environmental monitoring element knowledge. This transformation can include data normalization, feature extraction and combination, time series analysis and other operations. Through these transformations, the system obtains agricultural environment monitoring transformation knowledge, which is presented in a form that is more suitable for subsequent decision-making and analysis.

[0109] In summary, by setting mask features in the IoT sensor farming environment monitoring log, embedding knowledge features, and using the first knowledge transformation subnet for knowledge transformation, the system can effectively extract useful information from the original environmental monitoring data and convert it into a form that is easier to understand and apply. This not only improves the level of intelligence in farmland environmental monitoring, but also provides more accurate and comprehensive data support for agricultural production decisions.

[0110] In some other exemplary embodiments, the first knowledge transformation subnet includes a first residual branch and a numerical mapping branch; the agricultural environment monitoring element knowledge is transformed into agricultural environment monitoring knowledge through the first knowledge transformation subnet to obtain the agricultural environment monitoring transformation knowledge, including: performing residual connection on the agricultural environment monitoring element knowledge through the first residual branch to obtain residual connection features; performing interval numerical mapping processing on the residual connection features through the numerical mapping branch configured with fixed neural network variables to obtain the agricultural environment monitoring transformation knowledge; the agricultural environment monitoring transformation knowledge is a knowledge vector in a multidimensional knowledge relationship network.

[0111] In other exemplary embodiments, the system designs a more sophisticated and complex first knowledge transformation subnet to achieve effective transformation of agricultural environment monitoring element knowledge. This subnet consists of two main components, the first residual branch and the value mapping branch, which work together to complete the transformation and refinement of knowledge.

[0112] First, the system performs residual connection on the knowledge of agricultural environment monitoring elements through the first residual branch. Residual connection is a common technique in deep learning. It directly passes the input to the output through cross-layer connection and adds it to the output after processing by the network layer, thereby effectively alleviating the gradient vanishing problem and improving the training effect of the model. In this scenario, residual connection can retain the original information in the knowledge of agricultural environment monitoring elements, ensuring that important features are not lost in subsequent processing.

[0113] After the residual connection, the system obtains the residual connection features containing the original information and the transformed information. Next, these features are passed to the numerical mapping branch configured with a fixed neural network variable. The function of the numerical mapping branch is to perform interval numerical mapping on the residual connection features and convert them into a form that is more suitable for subsequent analysis. This process may involve operations such as linear transformation and nonlinear activation, which aims to map the features to a specific numerical interval to highlight the differences and correlations between different features.

[0114] After completing the interval numerical mapping process, the system obtains the knowledge of agricultural environment monitoring transformation. This knowledge exists in the form of knowledge vectors in a multidimensional knowledge relationship network. The multidimensional knowledge relationship network is a complex knowledge representation structure that can capture the intrinsic connections and hierarchical relationships between different knowledge. By embedding the knowledge of agricultural environment monitoring transformation into this network, the system can understand and apply this knowledge more comprehensively, providing strong support for subsequent decision-making and analysis.

[0115] In summary, through the collaborative processing of the first residual branch and the numerical mapping branch in the first knowledge transformation subnet, the system can effectively transform the knowledge of agricultural environment monitoring elements into more expressive and applicable agricultural environment monitoring transformation knowledge. This knowledge is presented in the form of knowledge vectors in the multidimensional knowledge relationship network, providing farmland managers with richer decision-making basis and analysis perspective.

[0116] Under some alternative technical schemes, the crop growth monitoring element knowledge of all crop growth monitoring data is transformed into growth monitoring knowledge to obtain the crop growth monitoring transformation knowledge, including: combining the crop growth monitoring element knowledge of all crop growth monitoring data according to the time priority of each crop growth monitoring data in the monitoring information of the target farming area to obtain the crop growth monitoring element combination knowledge; performing knowledge extraction on the crop growth monitoring element combination knowledge through a deep learning subnet to obtain the target knowledge features; performing growth monitoring knowledge transformation on the target knowledge features through a second knowledge transformation subnet to obtain the crop growth monitoring transformation knowledge.

[0117] Under some alternative technical solutions, the system uses a more sophisticated and dynamic method to process crop growth monitoring data to obtain crop growth monitoring transformation knowledge. This process involves multiple steps, aiming to extract valuable information from a large amount of growth monitoring data and convert it into knowledge that can be used for decision-making and analysis.

[0118] First, the system combines the crop growth monitoring element knowledge of all crop growth monitoring data according to the temporal priority of each crop growth monitoring data in the monitoring information of the target farming area. The temporal priority may be determined based on the data collection time, the importance of the growth stage or other relevant factors. Through combined processing, the system can integrate the monitoring data scattered at different time points to form a more comprehensive and representative crop growth monitoring element combination knowledge.

[0119] Next, the system uses the deep learning subnet to extract the combined knowledge of crop growth monitoring factors. The deep learning subnet may be a neural network with multiple hidden layers, which can learn and extract deep features from a large amount of input data. In this step, the system further abstracts and refines the combined knowledge through the training and learning of the deep learning subnet to obtain the target knowledge features. These target knowledge features can include key information such as the growth rate of crops, leaf color changes, plant height changes, etc., which can more accurately reflect the growth status of crops.

[0120] Finally, the system transforms the target knowledge features into growth monitoring knowledge through the second knowledge transformation subnet. The second knowledge transformation subnet may be a specially designed neural network or other type of machine learning model, which can transform the target knowledge features into a form more suitable for subsequent analysis and application. In this step, the system uses the transformation capability of the second knowledge transformation subnet to further transform and process the target knowledge features to obtain crop growth monitoring transformation knowledge. This transformation knowledge can include information such as crop growth trend prediction and health status assessment. They are presented in a form more suitable for decision-making and analysis, providing strong support for farmland managers.

[0121] In summary, by combining and processing crop growth monitoring data according to time priority, using the deep learning subnet for knowledge extraction, and transforming growth monitoring knowledge through the second knowledge transformation subnet, the system can effectively extract valuable information from a large amount of growth monitoring data and convert it into a form that is easier to understand and apply. This not only improves the intelligence level of crop growth monitoring, but also provides more accurate and comprehensive data support for agricultural production decisions.

[0122] Under some other alternative technical schemes, the regional block monitoring element knowledge of all regional blocks is transformed into regional monitoring knowledge to obtain the agricultural area monitoring transformation knowledge, including: combining the regional block monitoring element knowledge of all regional blocks according to the time priority of each regional block in the log block corresponding to the agricultural area positioning monitoring log to obtain regional block monitoring combination features; performing regional feature embedding processing on the regional block monitoring combination features through a regional feature embedding layer to obtain regional feature embedded knowledge; performing regional monitoring knowledge transformation on the regional feature embedded knowledge through a third knowledge transformation subnet to obtain the agricultural area monitoring transformation knowledge.

[0123] In other alternative technical solutions, the system processes the block monitoring data of the farming area more finely and comprehensively, aiming to obtain more instructive farming area monitoring transformation knowledge. The specific implementation process is as follows:

[0124] First, the system combines the regional block monitoring element knowledge of all regional blocks according to the temporal priority in the log block corresponding to the agricultural area positioning monitoring log of each regional block. The temporal priority here is determined based on the order of appearance or importance of the regional block in the log block. Through combined processing, the system can effectively integrate the monitoring element knowledge of different regional blocks to form a comprehensive regional block monitoring combination feature.

[0125] Next, the system uses the regional feature embedding layer to embed the regional block monitoring combination features. The regional feature embedding layer is a specially designed neural network layer that can capture and strengthen the key information in the regional block monitoring combination features while suppressing irrelevant information. Through the regional feature embedding process, the system obtains regional feature embedding knowledge containing rich regional features.

[0126] Finally, the system transforms the regional feature embedding knowledge into regional monitoring knowledge through the third knowledge transformation subnet. The third knowledge transformation subnet is a complex and efficient neural network structure that can perform deep transformation and abstraction on the input regional feature embedding knowledge. Through this transformation, the system obtains the crop area monitoring transformation knowledge, which is presented in a form more suitable for farmland management and decision-making, providing strong data support for agricultural production.

[0127] In summary, by combining and processing the regional block monitoring element knowledge according to the time priority, using the regional feature embedding layer to embed regional features, and transforming regional monitoring knowledge through the third knowledge transformation subnet, the system can effectively extract valuable information from a large amount of regional block monitoring data and convert it into more instructive agricultural area monitoring transformation knowledge. This not only improves the intelligent level of farmland monitoring, but also provides more accurate and comprehensive data support for agricultural production decision-making.

[0128] In some exemplary embodiments, the AI ​​knowledge integration of the agricultural environment monitoring transformation knowledge, the crop growth monitoring transformation knowledge and the agricultural area monitoring transformation knowledge is performed to obtain the global agricultural monitoring integrated knowledge under the same knowledge relationship network, including: performing linkage feature dimension combination processing on the agricultural environment monitoring transformation knowledge, the crop growth monitoring transformation knowledge and the agricultural area monitoring transformation knowledge through a second residual branch configured with fixed neural network variables to obtain the global agricultural monitoring integrated knowledge under the same knowledge relationship network; the global agricultural monitoring integrated knowledge is a linkage feature dimension combination vector under the same knowledge relationship network.

[0129] In some exemplary embodiments, the system performs AI knowledge integration on agricultural environment monitoring transformation knowledge, crop growth monitoring transformation knowledge, and agricultural area monitoring transformation knowledge in a very sophisticated and efficient process. This process aims to integrate knowledge from different sources and types to form a more comprehensive and instructive global agricultural monitoring integrated knowledge.

[0130] Specifically, the system performs linkage feature dimension combination processing on the above three transformation knowledge through the second residual branch configured with fixed neural network variables. Here, "fixed neural network variables" means that the neural network parameters in this branch remain unchanged during the training process, which helps to maintain the stability of the model; the "second residual branch" is a specially designed neural network structure that can effectively handle the correlation between different feature dimensions.

[0131] In the process of combining the linkage feature dimensions, the system first represents the agricultural environment monitoring transformation knowledge, crop growth monitoring transformation knowledge, and agricultural area monitoring transformation knowledge as feature vectors. Then, these feature vectors are combined and transformed in the second residual branch to capture the linkage relationship between them. This linkage relationship may be manifested as the changing laws of crop growth under different environmental conditions, the growth differences of crops in different regions, etc.

[0132] Through such processing, the system obtains the global agricultural monitoring integrated knowledge under the same knowledge relationship network. The "same knowledge relationship network" here refers to a unified knowledge representation space, which contains various agricultural monitoring knowledge and their interrelationships. The global agricultural monitoring integrated knowledge exists in the form of a combination vector of linkage feature dimensions, which can more comprehensively reflect the overall situation and development trend of agricultural production.

[0133] In summary, by configuring the second residual branch with fixed neural network variables to carry out linkage feature dimension combination processing on the agricultural environment monitoring transformation knowledge, crop growth monitoring transformation knowledge and agricultural area monitoring transformation knowledge, the system can effectively integrate these knowledge together to form more instructive global agricultural monitoring integrated knowledge. This not only improves the intelligent level of agricultural monitoring, but also provides more comprehensive and accurate data support for agricultural production decision-making.

[0134] In other exemplary embodiments, determining the potential production risk label of the target farming area monitoring information under at least one set monitoring task based on the global agricultural monitoring integrated knowledge includes: determining the discrimination possibility score of the target farming area monitoring information under at least one set monitoring task based on the global agricultural monitoring integrated knowledge; determining the potential production risk label of the target farming area monitoring information based on the discrimination possibility score under the at least one set monitoring task.

[0135] In other exemplary embodiments, the system uses the global agricultural monitoring integrated knowledge to determine the potential production risk label of the target farming area monitoring information under at least one set monitoring task. This process involves multiple steps, aiming to accurately identify the production risks that may exist in the target farming area by comprehensively analyzing the global agricultural monitoring integrated knowledge.

[0136] First, the system determines the discrimination possibility score of the target farming area monitoring information under at least one set monitoring task based on the global agricultural monitoring integrated knowledge. Here, the "discrimination possibility score" is a quantitative indicator used to measure the correlation or degree of compliance between the target farming area monitoring information and the set monitoring task. The system compares the linkage feature dimension combination vector in the global agricultural monitoring integrated knowledge with the characteristics of the target farming area monitoring information, calculates the similarity or difference between them, and thus obtains the discrimination possibility score. This score reflects the degree of production risk that may exist in the target farming area under a specific monitoring task.

[0137] Next, the system determines the potential production risk label of the monitoring information of the target farming area based on the discrimination possibility score under at least one set monitoring task. In this step, the system classifies or labels the target farming area according to the level of the discrimination possibility score or the preset threshold conditions. For example, if the discrimination possibility score exceeds a preset high-risk threshold, the system can mark the target farming area as a "high-risk area"; if the score is lower than a preset low-risk threshold, it can be marked as a "low-risk area"; and those between the two can be marked as "medium-risk areas." These labels provide intuitive risk prompts for farmland managers, helping them to take appropriate management measures in a timely manner.

[0138] In summary, by determining the discrimination possibility score based on the global agricultural monitoring integrated knowledge and determining the potential production risk label based on the score, the system can effectively identify the production risks that may exist in the target farming area under the set monitoring task. This provides strong decision-making support for farmland managers, helping them to take targeted management measures in a timely manner to ensure the smooth progress of agricultural production.

[0139] Under some optional design ideas, the discrimination possibility score of the target agricultural area monitoring information under at least one set monitoring task is determined based on the global agricultural monitoring integrated knowledge, including: performing decision analysis on the global agricultural monitoring integrated knowledge through a decision subnet to obtain a decision analysis vector; and determining the discrimination possibility score of the target agricultural area monitoring information under at least one set monitoring task based on the decision analysis vector through a discriminant subnet.

[0140] Under some optional design ideas, the system uses global agricultural monitoring integrated knowledge to determine the discrimination possibility score of the target farming area monitoring information under at least one set monitoring task, which is further refined into two main steps. These two steps are implemented through different neural network subnets.

[0141] First, the system conducts decision analysis on the global agricultural monitoring integrated knowledge through the decision subnet. The decision subnet is a specially designed neural network structure that is responsible for extracting information that is critical for decision-making from the global agricultural monitoring integrated knowledge. This information can include the growth trend of crops, the impact of environmental factors on crops, and the production differences between different farming areas. Through the processing of the decision subnet, these key information are integrated into a decision analysis vector, which represents the comprehensive performance of the global agricultural monitoring integrated knowledge at the decision-making level in the form of a numerical value.

[0142] Next, the system uses the discriminant subnet to determine the discrimination possibility score of the target crop area monitoring information under at least one set monitoring task based on the decision analysis vector. The discriminant subnet is another specially designed neural network structure. Its function is to evaluate the correlation between the target crop area monitoring information and the set monitoring task based on the decision analysis vector. Specifically, the discriminant subnet calculates the similarity or degree of match between the decision analysis vector and each set monitoring task, and gives a quantitative score, namely the discrimination possibility score. This score reflects the severity of the problems or risks that may exist in the target crop area under a specific monitoring task.

[0143] Through this design, the system can more accurately determine the discrimination possibility score of the target farming area monitoring information under the set monitoring task, thereby providing more reliable data support for subsequent farmland management and decision-making. At the same time, this neural network-based processing method also improves the intelligence level of the system, enabling it to automatically extract useful information from a large amount of data and make reasonable decisions and judgments.

[0144] In some independently implementable technical solutions, the agricultural area monitoring information analysis method applied to digital agriculture is implemented through a long short-term memory network; the long short-term memory network is debugged through the following steps: obtaining an agricultural area monitoring training set; the agricultural area monitoring training set includes past target agricultural area monitoring information and production risk label samples of the past target agricultural area monitoring information under a set monitoring task; the past target agricultural area monitoring information includes past IoT sensor agricultural environment monitoring logs, past airborne remote sensing growth monitoring logs and past agricultural area positioning monitoring logs; the past target agricultural area monitoring information is stored in the ... is stored in the agricultural area monitoring training set; the agricultural area monitoring training set includes past target agricultural area monitoring information and production risk label samples of the past target agricultural area monitoring information under a set monitoring task; the past target agricultural area monitoring information is stored in the agricultural area monitoring training set; the agricultural area monitoring training set includes past target agricultural area monitoring information and production risk label samples of the past target agricultural area monitoring information under a set monitoring task; the agricultural area monitoring information is stored in the agricultural area monitoring training set; the agricultural area monitoring information is stored in the agricultural area monitoring training set; the agricultural area monitoring information is stored in the agricultural area monitoring training set; the agricultural area monitoring information is stored in the agricultural area monitoring training set; the agricultural area monitoring information is stored in the agricultural area monitoring training set; the agricultural area monitoring information is stored in the agricultural area monitoring training set; the agricultural area monitoring information is stored in the agricultural area monitoring training set; the agricultural area monitoring information is stored in the agricultural area monitoring training set; the agricultural area monitoring information is stored in the agricultural area monitoring training set; the agricultural area monitoring information is stored in the agricultural area monitoring training set; the agricultural area monitoring information is stored in the agricultural The crop area monitoring information is loaded into the long short-term memory network; the monitoring element mining is performed on each of the multiple groups of past crop growth monitoring data corresponding to the past airborne remote sensing growth monitoring logs through the first monitoring and analysis subnet of the long short-term memory network to obtain the past crop growth monitoring element knowledge of each past crop growth monitoring data; and the regional monitoring element mining is performed on the multiple past regional blocks corresponding to the past crop area positioning monitoring logs to obtain the past regional block monitoring element knowledge of each past regional block; the monitoring element mining is performed on the multiple past regional blocks corresponding to the past crop area positioning monitoring logs to obtain the past regional block monitoring element knowledge of each past regional block; The monitoring and analysis subnet transforms the past agricultural environment monitoring element knowledge corresponding to the past IoT sensor agricultural environment monitoring log, the past crop growth monitoring element knowledge of all past crop growth monitoring data, and the past regional block monitoring element knowledge of all past regional blocks, and obtains the past agricultural environment monitoring transformation knowledge, the past crop growth monitoring transformation knowledge, and the past agricultural area monitoring transformation knowledge accordingly; the third monitoring and analysis subnet of the long short-term memory network transforms the past agricultural environment monitoring transformation knowledge, the past crop growth monitoring transformation knowledge, and the past agricultural area monitoring transformation knowledge accordingly. The AI ​​knowledge integration is performed by exchanging knowledge to obtain past global agricultural monitoring integrated knowledge under the same knowledge relationship network; and, based on the past global agricultural monitoring integrated knowledge, the past potential production risk labels of the past target crop area monitoring information under the set monitoring task are determined; the past potential production risk labels and the production risk label samples are loaded into the training error processing subnet to determine the training error and obtain the training error information; the network parameters in the second monitoring and analysis subnet and the third monitoring and analysis subnet are optimized based on the training error information to obtain the debugged long short-term memory network.

[0145] Among some independently implementable technical solutions, the method for analyzing crop area monitoring information applied to digital agriculture is implemented through a long short-term memory network (LSTM). The LSTM is a special type of recurrent neural network (RNN) that can learn long-term dependencies and therefore performs well in processing time series data and problems with sequence characteristics. The following will explain in detail how to debug this LSTM network through a series of steps to make it suitable for crop area monitoring information analysis.

[0146] First, the system obtains the agricultural area monitoring training set. This training set contains the past monitoring information of the target agricultural area and the production risk label samples of this monitoring information under the set monitoring task. Specifically, the past target agricultural area monitoring information includes the past IoT sensor agricultural environment monitoring logs, the past airborne remote sensing growth monitoring logs and the past agricultural area positioning monitoring logs. These logs record the environment, growth and location status information of the agricultural area at different time points.

[0147] Next, the system loads the past monitoring information of the target crop area into the long short-term memory network. Then, through the first monitoring and analysis subnet of the long short-term memory network, the monitoring element mining is performed on each of the multiple groups of past crop growth monitoring data corresponding to the past airborne remote sensing growth monitoring logs. This process aims to extract the key elements in each group of past crop growth monitoring data, such as growth rate, health status, etc., and obtain the past crop growth monitoring element knowledge of each past crop growth monitoring data. At the same time, the first monitoring and analysis subnet also mines regional monitoring elements for multiple past regional blocks corresponding to the past crop area positioning monitoring logs, such as soil quality, climate conditions, etc., to obtain the past regional block monitoring element knowledge of each past regional block.

[0148] Subsequently, the second monitoring and analysis subnet of the long short-term memory network performs knowledge transformation on the past agricultural environment monitoring element knowledge corresponding to the past IoT sensor agricultural environment monitoring log, the past crop growth monitoring element knowledge of all past crop growth monitoring data, and the past regional block monitoring element knowledge of all past regional blocks. This transformation may include data standardization, feature extraction, or dimensionality reduction, with the aim of making this knowledge easier for subsequent analysis and integration. After the transformation, the past agricultural environment monitoring transformation knowledge, the past crop growth monitoring transformation knowledge, and the past agricultural area monitoring transformation knowledge are obtained.

[0149] Next, the three transformation knowledge are integrated by AI knowledge through the third monitoring analysis subnet of the long short-term memory network. This process may involve fusing knowledge from different sources and types to form a more comprehensive and instructive global agricultural monitoring integrated knowledge. This integrated knowledge can more accurately reflect the overall status and development trend of agricultural production.

[0150] Then, based on the integrated knowledge of past global agricultural monitoring, the system determines the past potential production risk labels of the past target crop area monitoring information under the set monitoring task. This is achieved by comparing the integrated knowledge with the preset risk assessment criteria or applying a certain risk calculation model.

[0151] Finally, the system loads past potential production risk labels and production risk label samples into the training error processing subnet to determine the training error. This process aims to compare the difference between the risk label predicted by the system and the actual risk label to obtain a quantitative training error information. Based on this training error information, the system can optimize the network parameters in the second monitoring and analysis subnet and the third monitoring and analysis subnet, such as adjusting weights, updating learning rates, etc., to reduce the prediction error and improve the accuracy of the model. After such optimization, a debugged long short-term memory network is obtained, which will be more suitable for agricultural area monitoring information analysis tasks.

[0152] Long Short-Term Memory (LSTM) is a special type of recurrent neural network (RNN) that is designed to solve the gradient vanishing or gradient exploding problems encountered by traditional RNNs when processing long sequence data, thereby enabling learning long-term dependencies. LSTM performs well in many sequence modeling tasks, such as natural language processing, speech recognition, time series prediction, etc.

[0153] The core idea of ​​LSTM is to control the flow of information by introducing gates and cell states. The gates include forget gate, input gate and output gate, which are responsible for deciding which information should be retained, which new information should be added and which information should be output.

[0154] Forget gate: decides what information to discard from the cell state. It outputs a value between 0 and 1 by looking at the current input and the hidden state of the previous moment, indicating the degree of retention or discarding of information.

[0155] Input gate: determines what new information should be added to the cell state. It also generates a candidate value by looking at the current input and the hidden state of the previous moment, and uses another gating signal to determine how much of this candidate value should be added to the cell state.

[0156] Output gate: Determines what information to output based on the current cell state. It first passes through a sigmoid layer to determine which parts of the cell state should be output, then passes the cell state through a tanh layer to obtain a value between -1 and 1, and then multiplies it with the output of the sigmoid layer to obtain the final output.

[0157] The cell state is a key part in LSTM, which acts like a conveyor belt to pass information throughout the sequence. There are only a small number of linear interactions, allowing information to flow easily without change. Through the gating mechanism, LSTM is able to selectively update the information in the cell state, so that long-term dependencies can be captured and preserved.

[0158] In the context of crop area monitoring information analysis, LSTM can predict future production risks by learning from past monitoring data. For example, it can use past IoT sensing crop environment monitoring logs, airborne remote sensing growth monitoring logs, and crop area positioning monitoring logs to learn the impact of environmental factors, growth trends, and geographic locations on crop growth risks. By performing sequence modeling on these data, LSTM can predict the growth status and production risks of crops in the future, thereby providing decision support for farm managers.

[0159] Please refer to Figure 2 The embodiment of the present invention further provides a digital agricultural information analysis system 100, including a processor 111, and a memory 112 and a bus 113 connected to the processor 111. The processor 111 and the memory 112 communicate with each other through the bus 113. The processor 111 is used to call the program instructions in the memory 112 to execute the above-mentioned agricultural area monitoring information analysis method applied to digital agriculture.

[0160] In summary, this method first conducts in-depth monitoring factor mining on each group of crop growth monitoring data in the airborne remote sensing growth monitoring log, and extracts the monitoring factor knowledge closely related to crop growth; at the same time, regional monitoring factor mining is performed on each regional block in the agricultural area positioning monitoring log to obtain the environment, soil and other element information in each block. Then, the present invention performs knowledge transformation processing on the knowledge corresponding to the IoT sensing agricultural environment monitoring log, crop growth monitoring data and regional block monitoring data, and converts this knowledge into a form that is more suitable for integration and analysis. Finally, through AI knowledge integration technology, the transformed various types of monitoring knowledge are integrated into a global agricultural monitoring integrated knowledge base, so that the actual situation of the agricultural area can be more comprehensively and accurately reflected, and strong data support can be provided for subsequent monitoring tasks.

[0161] Through this method, the present invention can not only effectively integrate and utilize various agricultural area monitoring information to improve the efficiency and accuracy of data processing, but also can quickly determine the potential production risk labels of the target agricultural area under different monitoring tasks based on the global agricultural monitoring integrated knowledge, providing agricultural managers with timely and accurate decision-making basis. This is of great significance for improving the intelligent level of agricultural production and realizing refined agricultural management.

[0162] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity or system including the element.

[0163] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0164] The above are only embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A method for analyzing crop area monitoring information applied to digital agriculture, characterized in that: Applied to a digital agricultural information analysis system, the method comprises: Collecting monitoring information of the target farming area; the monitoring information of the target farming area includes IoT sensing farming environment monitoring logs, airborne remote sensing growth monitoring logs and farming area positioning monitoring logs; Performing monitoring element mining on each crop growth monitoring data in the multiple groups of crop growth monitoring data corresponding to the airborne remote sensing growth monitoring log to obtain crop growth monitoring element knowledge for each crop growth monitoring data; and performing regional monitoring element mining on the multiple regional blocks corresponding to the crop area positioning monitoring log to obtain regional block monitoring element knowledge for each regional block; The agricultural environment monitoring element knowledge corresponding to the IoT sensing agricultural environment monitoring log, the crop growth monitoring element knowledge of all crop growth monitoring data, and the regional block monitoring element knowledge of all regional blocks are respectively transformed to obtain agricultural environment monitoring transformation knowledge, crop growth monitoring transformation knowledge, and agricultural area monitoring transformation knowledge; wherein, the agricultural environment monitoring element knowledge refers to the key information and features about the farmland environment extracted from the IoT sensing agricultural environment monitoring log, and the agricultural environment monitoring element knowledge describes the state and change law of the farmland environment. The temperature change trend and light intensity distribution in the farmland are extracted through the analysis of the IoT sensing agricultural environment monitoring log, which constitute the agricultural environment monitoring element knowledge, which is used to evaluate the impact of the farmland environment on crop growth; knowledge transformation refers to the process of converting and uniformly representing knowledge from different sources and in different formats; Performing AI knowledge integration on the farming environment monitoring transformation knowledge, the crop growth monitoring transformation knowledge and the farming area monitoring transformation knowledge to obtain global agricultural monitoring integrated knowledge under the same knowledge relationship network; wherein the AI ​​knowledge integration is fusion processing; Determine, based on the global agricultural monitoring integrated knowledge, a potential production risk label of the target farming area monitoring information under at least one set monitoring task; The agricultural environment monitoring element knowledge corresponding to the IoT sensor agricultural environment monitoring log, the crop growth monitoring element knowledge of all crop growth monitoring data, and the regional block monitoring element knowledge of all regional blocks are respectively transformed to obtain agricultural environment monitoring transformation knowledge, crop growth monitoring transformation knowledge, and agricultural area monitoring transformation knowledge, including: Performing agricultural environment monitoring knowledge transformation on the agricultural environment monitoring element knowledge corresponding to the IoT sensor agricultural environment monitoring log to obtain agricultural environment monitoring transformation knowledge; Performing growth monitoring knowledge transformation on crop growth monitoring element knowledge of all crop growth monitoring data to obtain crop growth monitoring transformation knowledge; Performing regional monitoring knowledge transformation on the regional block monitoring element knowledge of all regional blocks to obtain the crop area monitoring transformation knowledge; The agricultural environment monitoring knowledge transformation is performed on the agricultural environment monitoring element knowledge corresponding to the IoT sensor agricultural environment monitoring log to obtain the agricultural environment monitoring transformation knowledge, including: Setting mask features in the first log section and the second log section of the IoT sensor farming environment monitoring log respectively to form an IoT sensor farming environment monitoring mask log; Embedding knowledge features into the mask log of the IoT sensor farming environment monitoring to obtain the farming environment monitoring element knowledge; Performing agricultural environment monitoring knowledge transformation on the agricultural environment monitoring element knowledge through the first knowledge transformation subnet to obtain the agricultural environment monitoring transformation knowledge; The first knowledge transformation subnet includes a first residual branch and a value mapping branch; the agricultural environment monitoring knowledge transformation is performed on the agricultural environment monitoring element knowledge through the first knowledge transformation subnet to obtain the agricultural environment monitoring transformation knowledge, including: Performing residual connection on the farming environment monitoring element knowledge through the first residual branch to obtain residual connection features; By configuring the numerical mapping branch with fixed neural network variables, the residual connection features are processed by interval numerical mapping to obtain the agricultural environment monitoring transformation knowledge; the agricultural environment monitoring transformation knowledge is a knowledge vector in a multidimensional knowledge relationship network.

2. The method according to claim 1, characterized in that The crop growth monitoring knowledge transformation is performed on the crop growth monitoring element knowledge of all crop growth monitoring data to obtain crop growth monitoring transformation knowledge, including: Combining the crop growth monitoring element knowledge of all the crop growth monitoring data according to the time sequence priority of each of the crop growth monitoring data in the monitoring information of the target farming area to obtain the crop growth monitoring element combination knowledge; The knowledge of the combination of crop growth monitoring elements is refined through a deep learning subnet to obtain target knowledge features; The target knowledge feature is transformed into growth monitoring knowledge through the second knowledge transformation subnet to obtain the crop growth monitoring transformation knowledge.

3. The method according to claim 1, characterized in that The regional monitoring knowledge transformation is performed on the regional block monitoring element knowledge of all regional blocks to obtain the crop area monitoring transformation knowledge, including: According to the time sequence priority of each of the regional blocks in the log block corresponding to the agricultural area positioning monitoring log, the regional block monitoring element knowledge of all the regional blocks is combined to obtain the regional block monitoring combination feature; Performing regional feature embedding processing on the regional block monitoring combined features through a regional feature embedding layer to obtain regional feature embedding knowledge; The regional feature embedding knowledge is transformed into regional monitoring knowledge through the third knowledge transformation subnet to obtain the crop area monitoring transformation knowledge.

4. The method according to claim 1, characterized in that: The AI ​​knowledge integration of the farming environment monitoring transformation knowledge, the crop growth monitoring transformation knowledge and the farming area monitoring transformation knowledge is performed to obtain global agricultural monitoring integrated knowledge under the same knowledge relationship network, including: The agricultural environment monitoring transformation knowledge, the crop growth monitoring transformation knowledge and the agricultural area monitoring transformation knowledge are processed by combining the linked feature dimensions through a second residual branch configured with fixed neural network variables to obtain global agricultural monitoring integrated knowledge under the same knowledge relationship network; the global agricultural monitoring integrated knowledge is a linked feature dimension combination vector under the same knowledge relationship network.

5. The method according to claim 1, characterized in that: The step of determining the potential production risk label of the target farming area monitoring information under at least one set monitoring task based on the global agricultural monitoring integrated knowledge includes: Determining a discrimination possibility score of the target farming area monitoring information under at least one set monitoring task based on the global agricultural monitoring integrated knowledge; Determining a potential production risk label of the monitoring information of the target farming area according to the discrimination possibility score under the at least one set monitoring task; Wherein, determining the discrimination possibility score of the target farming area monitoring information under at least one set monitoring task based on the global agricultural monitoring integrated knowledge includes: Performing decision analysis on the global agricultural monitoring integrated knowledge through a decision subnet to obtain a decision analysis vector; The discrimination subnet determines the discrimination possibility score of the target farming area monitoring information under at least one set monitoring task according to the decision analysis vector.

6. The method according to claim 1, characterized in that The agricultural area monitoring information analysis method applied to digital agriculture is implemented by a long short-term memory network; the long short-term memory network is debugged by the following steps: Acquire a crop area monitoring training set; the crop area monitoring training set includes past target crop area monitoring information and production risk label samples of the past target crop area monitoring information under a set monitoring task; the past target crop area monitoring information includes past IoT sensor crop environment monitoring logs, past airborne remote sensing growth monitoring logs, and past crop area positioning monitoring logs; Loading the past target farming area monitoring information into the long short-term memory network; By using the first monitoring analysis subnet of the long short-term memory network, each of the multiple groups of past crop growth monitoring data corresponding to the past airborne remote sensing growth monitoring log is subjected to monitoring element mining to obtain past crop growth monitoring element knowledge of each past crop growth monitoring data; and performing regional monitoring element mining on a plurality of past regional blocks corresponding to the past agricultural area positioning monitoring logs to obtain past regional block monitoring element knowledge of each past regional block; Through the second monitoring and analysis subnet of the long short-term memory network, the past agricultural environment monitoring element knowledge corresponding to the past IoT sensor agricultural environment monitoring log, the past crop growth monitoring element knowledge of all past crop growth monitoring data, and the past regional block monitoring element knowledge of all past regional blocks are respectively transformed to obtain past agricultural environment monitoring transformation knowledge, past crop growth monitoring transformation knowledge, and past agricultural area monitoring transformation knowledge; Through the third monitoring analysis subnet of the long short-term memory network, AI knowledge integration is performed on the past agricultural environment monitoring transformation knowledge, the past crop growth monitoring transformation knowledge and the past agricultural area monitoring transformation knowledge to obtain past global agricultural monitoring integrated knowledge under the same knowledge relationship network; And, based on the past global agricultural monitoring integrated knowledge, determining the past potential production risk label of the past target farming area monitoring information under the set monitoring task; Loading the past potential production risk labels and the production risk label samples into a training error processing subnet to determine the training error and obtain training error information; The network parameters in the second monitoring and analysis subnet and the third monitoring and analysis subnet are optimized according to the training error information to obtain a debugged long short-term memory network.

7. A digital agricultural information analysis system, characterized in that: It comprises a processor, a memory and a bus connected to the processor; the processor and the memory communicate with each other via the bus; The processor is used to call the computer program in the memory to execute the agricultural area monitoring information analysis method applied to digital agriculture as described in any one of claims 1-6.

8. A computer-readable storage medium, characterized in that: A program is stored thereon, and when the program is executed by a processor, the method for analyzing agricultural area monitoring information applied to digital agriculture as described in any one of claims 1-6 is implemented.

Images