A risk early warning method, device, equipment and medium

Abstract

The application belongs to the field of artificial intelligence and relates to a risk early warning method, device, equipment and medium, comprising the following steps: obtaining underwriting data, plot attribute information and multi-source remote sensing image data of a target plot; performing remote sensing crop identification on the multi-source remote sensing image data to obtain a crop identification result of the target plot; establishing a space-time topology relationship model of the target plot based on the crop identification result and the plot attribute information; identifying a target virtual increase risk based on the underwriting data and the space-time topology relationship model to obtain a multi-dimensional virtual increase risk identification result; obtaining multi-dimensional risk factors of the target plot from a preset risk factor library and determining a risk threshold set of the target plot according to the multi-dimensional risk factors; performing risk verification on the multi-dimensional virtual increase risk identification result based on the risk threshold set to generate a differentiated early warning. The application can be applied to the business field of insurance and the like and can realize accurate and efficient agricultural insurance target virtual increase risk identification and early warning.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology and is applied to online business scenarios such as insurance, particularly to a risk warning method, device, equipment, and medium. Background Technology

[0002] In the field of agricultural insurance, risk control regarding inflated insured assets is of great significance. Such inflated insured assets can lead to systemic losses of government subsidies, information asymmetry and moral hazard, and undermine the fairness of insurance operations.

[0003] However, current traditional methods for identifying inflated claims have significant limitations. First, they rely on a single data dimension, resulting in numerous blind spots and making it difficult to comprehensively and accurately detect inflated claims. Second, manual verification is disconnected from remote sensing data, lacking an effective matching and verification mechanism. Relying on visual comparison is inefficient and prone to missed detections, hindering timely detection of inflated claims. Third, the system's early warning capabilities are weak, employing static threshold warnings based on fixed area difference ratios, which cannot address the differences in insurance characteristics across different crops and regions, leading to unsatisfactory early warning results. Summary of the Invention

[0004] The purpose of this application is to provide a risk warning method, device, computer equipment, and storage medium to solve the problem that the existing technology cannot accurately identify and warn of the risk of inflated agricultural insurance targets.

[0005] Firstly, a risk warning method is provided, which adopts the following technical solution: The process involves acquiring underwriting data, land attribute information, and multi-source remote sensing image data for the target land parcel; performing remote sensing crop identification on the multi-source remote sensing image data to obtain crop identification results for the target land parcel; establishing a spatiotemporal topological relationship model for the target land parcel based on the crop identification results and land attribute information; identifying the risk of inflated claims using the spatiotemporal topological relationship model based on the underwriting data to obtain multi-dimensional inflated risk identification results; obtaining multi-dimensional risk factors for the target land parcel from a pre-set risk factor library; determining a risk threshold set for the target land parcel based on the multi-dimensional risk factors; and performing risk verification on the multi-dimensional inflated risk identification results based on the risk threshold set to generate differentiated early warnings.

[0006] Secondly, a risk warning device is provided, which adopts the following technical solution: The acquisition module is used to acquire the insurance data, land attribute information, and multi-source remote sensing image data of the target land parcel. The crop identification module is used to identify crops from multi-source remote sensing image data and obtain crop identification results for the target plot. A module is established to build a spatiotemporal topological relationship model of the target plot based on crop identification results and plot attribute information; The risk identification module is used to identify the risk of inflated claims based on underwriting data and using a spatiotemporal topological relationship model to obtain multi-dimensional inflated risk identification results. The first determining module is used to obtain multi-dimensional risk factors of the target land plot from a preset risk factor library, and determine the risk threshold set of the target land plot based on the multi-dimensional risk factors; The verification module is used to verify the results of multi-dimensional false risk identification based on a set of risk thresholds and generate differentiated warnings.

[0007] Thirdly, a computer device is provided, which adopts the following technical solution: The process involves acquiring underwriting data, land attribute information, and multi-source remote sensing image data for the target land parcel; performing remote sensing crop identification on the multi-source remote sensing image data to obtain crop identification results for the target land parcel; establishing a spatiotemporal topological relationship model for the target land parcel based on the crop identification results and land attribute information; identifying the risk of inflated claims using the spatiotemporal topological relationship model based on the underwriting data to obtain multi-dimensional inflated risk identification results; obtaining multi-dimensional risk factors for the target land parcel from a pre-set risk factor library; determining a risk threshold set for the target land parcel based on the multi-dimensional risk factors; and performing risk verification on the multi-dimensional inflated risk identification results based on the risk threshold set to generate differentiated early warnings.

[0008] Fourthly, a computer-readable storage medium is provided, which adopts the following technical solution: The process involves acquiring underwriting data, land attribute information, and multi-source remote sensing image data for the target land parcel; performing remote sensing crop identification on the multi-source remote sensing image data to obtain crop identification results for the target land parcel; establishing a spatiotemporal topological relationship model for the target land parcel based on the crop identification results and land attribute information; identifying the risk of inflated claims using the spatiotemporal topological relationship model based on the underwriting data to obtain multi-dimensional inflated risk identification results; obtaining multi-dimensional risk factors for the target land parcel from a pre-set risk factor library; determining a risk threshold set for the target land parcel based on the multi-dimensional risk factors; and performing risk verification on the multi-dimensional inflated risk identification results based on the risk threshold set to generate differentiated early warnings.

[0009] Compared with existing technologies, the embodiments of this application have the following main advantages: By integrating underwriting data, land parcel attribute information, and multi-source remote sensing image data, the limitations of traditional single data dimensions are overcome, achieving comprehensive coverage and identification of inflated risks and significantly reducing the probability of missed detection. A spatiotemporal topological relationship model is constructed based on remote sensing crop identification results and land parcel attribute information, establishing an efficient matching and verification mechanism between declared data and actual data, replacing inefficient manual visual comparison, and improving the accuracy and efficiency of inflated risk identification. A dynamic risk threshold set is determined by combining multi-dimensional risk factors, replacing the traditional static one-size-fits-all threshold, adapting to the differences in insurance characteristics of different crops and regions, generating targeted differentiated early warnings, significantly optimizing the early warning effect, and enabling accurate and efficient identification and early warning of inflated risks in agricultural insurance targets. Attached Figure Description

[0010] To more clearly illustrate the solutions in this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is an exemplary system architecture diagram to which this application can be applied; Figure 2 A flowchart of an embodiment of the risk warning method according to this application; Figure 3 This is a schematic diagram of a structure of an embodiment of the risk warning device according to this application; Figure 4 This is a schematic diagram of the structure of one embodiment of the computer device according to this application. Detailed Implementation

[0012] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0013] like Figure 1 As shown, system architecture 100 may include terminal device 101, network 102, and server 103. Terminal device 101 may be a laptop 1011, tablet 1012, or mobile phone 1013. Network 102 is used as a medium to provide a communication link between terminal device 101 and server 103. Network 102 may include various connection types, such as wired, wireless communication links, or fiber optic cables.

[0014] Users can use terminal device 101 to interact with server 103 via network 102 to receive or send messages, etc. Various communication client applications can be installed on terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social media platform software, etc.

[0015] Terminal device 101 can be various electronic devices with a display screen and support web browsing. In addition to laptops 1011, tablets 1012, or mobile phones 1013, terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 player (Moving Picture Experts Group Audio Layer IV), a laptop computer, and a desktop computer, etc.

[0016] Server 103 can be a server that provides various services, such as a backend server that provides support for the pages displayed on terminal device 101.

[0017] It should be noted that the risk warning method provided in this application embodiment is generally executed by a server / terminal device, and correspondingly, the risk warning device is generally set in the server / terminal device.

[0018] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0019] Continue to refer to Figure 2 A flowchart of an embodiment of the risk warning method according to this application is shown. The risk warning method includes the following steps: Step S201: Obtain the insurance coverage data, land parcel attribute information, and multi-source remote sensing image data of the target land parcel.

[0020] The target plot refers to a specific land area included in the agricultural insurance coverage and subject to risk identification and control. Its scope is defined by the insured object determined by the underwriting business, and its core characteristics include spatial location, boundary range, and planting attributes. For example, a farmer insures 10 mu of rice paddy, or an agricultural cooperative insures 50 mu of wheat planting area.

[0021] Among them, underwriting data refers to the set of business data collected and stored by insurance institutions in the agricultural insurance underwriting process based on the information declared by the insured and their own underwriting review records. It centrally reflects the declaration status of the insured object and key information of the insurance contract. Its core purpose is to compare with actual data to identify the risk of inflated data. Specifically, it may include information such as the boundary information of the declared plot, the declared crop type information, the declared planting area, the insured amount, the insurance period, and the insured information.

[0022] Among them, the land parcel attribute information refers to the sum of data describing the inherent characteristics of the target land parcel and the suitable planting conditions. It reflects the spatial attributes and soil conditions of the land parcel, and also clarifies its planting suitability, such as the land parcel location coordinates, legal boundaries, soil type, fertility level, and information on suitable crops such as corn for dry land and rice for paddy fields.

[0023] Multi-source remote sensing image data refers to a combination of image data collected at different resolutions through different remote sensing platforms such as satellites and drones, which can reflect the surface and planting conditions of a target plot. It can accurately present the actual surface morphology, crop cover, and boundary features of the target plot. For example, basic remote sensing images collected by satellites, high-precision remote sensing images collected by satellites, and auxiliary remote sensing images collected by drones.

[0024] Step S202: Perform remote sensing crop identification on the multi-source remote sensing image data to obtain the crop identification results of the target plot.

[0025] Among them, remote sensing crop identification refers to a series of hierarchical processing procedures for multi-source remote sensing image data that integrate remote sensing image processing, spatial data modeling and field verification technologies. Its core logic is to accurately obtain crop planting information and plot boundary information of the target plot through progressively refined image analysis and multi-source data verification.

[0026] The crop identification result is the core information of the target plot, output through the aforementioned hierarchical remote sensing crop identification process and verified and corrected by multi-source data. It not only accurately represents the actual crop type of the target plot (such as wheat, rice, corn, etc.), but also clearly presents the specific distribution range, actual boundary coordinates, and boundary shape of the crop, while integrating the plot attribute suitability verification results. For example, the identification result of the target plot is "actual planting area of ​​12 mu, boundary coordinates of xxx-xxx, crop type of winter wheat, and suitable for the dryland attribute of the plot."

[0027] Step S203: Based on the crop identification results and plot attribute information, establish a spatiotemporal topological relationship model of the target plot.

[0028] Among them, the spatiotemporal topology model is a mathematical model constructed by integrating real spatiotemporal data and land parcel attribute information from crop identification results, relying on multi-source data fusion modeling technology. Its core function is to characterize the spatial topological association of the target land parcel (such as the adjacency / overlap relationship with the surrounding area), temporal stability (such as the changes in boundaries and planting types in recent years), and attribute adaptation rules (such as the adaptation relationship between land type and crop), providing reliable model support for the accurate verification of underwriting data and the identification of false inflation risks.

[0029] Step S204: Based on the underwriting data, a spatiotemporal topological relationship model is used to identify the risk of inflated claims, and multi-dimensional inflated risk identification results are obtained.

[0030] Among them, the identification of the risk of inflated target is a series of technical processes that combine the risk management needs of agricultural insurance with data verification technology. It uses real benchmark data provided by spatiotemporal topological relationship model to compare and verify the declared information in the underwriting data in multiple dimensions. By analyzing the difference between the declared boundary and the actual boundary, the compatibility between the declared crop and the suitable crop, etc., the risk of fictitious target, inflated area, and falsely reported crop type can be accurately identified. The core is to judge the difference between the declared information and the actual situation and the possibility of inflated target.

[0031] Among them, the multi-dimensional inflated risk identification result is a structured data product generated after the target inflated risk identification process. It originates from the in-depth comparative analysis of underwriting data and spatiotemporal topological relationship model, and comprehensively presents risk points and quantitative levels from spatiotemporal and attribute dimensions.

[0032] Step S205: Obtain multi-dimensional risk factors for the target land parcel from the preset risk factor library, and determine the risk threshold set for the target land parcel based on the multi-dimensional risk factors.

[0033] Among them, the risk factor library is a database that is pre-built through the analysis of historical risk cases of agricultural insurance, the survey of regional agricultural characteristics, and the sorting out of insurance product design specifications. Its core function is to store various influencing factors related to the inflated risk of agricultural insurance targets. These factors serve as key variables affecting the inflated risk in different scenarios, providing data support for the determination of differentiated risk thresholds. Specifically, it can include a set of factors stored in categories such as regional characteristic factors, crop attribute factors, farmer risk factors, insurance product suitability factors, and topological risk factors.

[0034] Among them, multi-dimensional risk factors are various key factors extracted from the risk factor library that affect the inflated risk level of the target plot. The data comes from the classified storage content of the risk factor library. Different factors correspond to different influence weights. The core purpose is to calculate the dynamic risk threshold through quantitative analysis.

[0035] Among them, the risk threshold set is a set of critical values ​​determined by the weight analysis and quantitative calculation of multi-dimensional risk factors, after statistical modeling and threshold calibration. Each critical value corresponds to a risk dimension. Its core function is to define the acceptable risk boundary under different scenarios and provide a judgment standard for the verification of the multi-dimensional false risk identification results. Specifically, it can include critical values ​​of different dimensions such as boundary difference threshold, area false ratio threshold, and land type crop mismatch risk threshold.

[0036] Step S206: Based on the risk threshold set, perform risk verification on the multi-dimensional false risk identification results and generate differentiated early warnings.

[0037] Risk verification is a technical process that applies risk assessment and data verification technologies to compare and analyze the risk data of each dimension in the multi-dimensional false risk identification results with the corresponding critical values ​​in the risk threshold set. The core purpose is to determine whether the risk level of the identification results exceeds the acceptable range, thereby filtering out truly high-risk cases. For example, comparing the boundary declaration difference value with the boundary difference threshold, and comparing the area false increase ratio with the area threshold, all belong to this process.

[0038] Differentiated early warning, in particular, combines risk-based early warning technology with the management needs of agricultural insurance. Based on risk verification results, it generates early warning information and handling suggestions after comprehensively considering the risk level, causes, and scope of impact. Its core feature is its targeted nature, clearly defining the priority and methods for handling different types of inflated risks. For example, high-risk levels correspond to on-site verification early warnings, medium-risk levels to supplementary material early warnings, and low-risk levels to system-marked early warnings.

[0039] This application's embodiments overcome the limitations of traditional single-data-dimensional approaches by integrating underwriting data, land parcel attribute information, and multi-source remote sensing image data, achieving comprehensive coverage and identification of inflated risks and significantly reducing the probability of missed detections. Based on remote sensing crop identification results and land parcel attribute information, a spatiotemporal topological relationship model is constructed, establishing an efficient matching and verification mechanism between declared and actual data. This replaces inefficient manual visual comparison, improving the accuracy and efficiency of inflated risk identification. A dynamic risk threshold set is determined by combining multi-dimensional risk factors, replacing the traditional static one-size-fits-all threshold. This adapts to the differences in insurance characteristics of different crops and regions, generating targeted and differentiated early warnings, significantly optimizing the early warning effect. This enables accurate and efficient identification and early warning of inflated risks in agricultural insurance targets.

[0040] In some optional implementations of this embodiment, the multi-source remote sensing image data includes basic remote sensing image data, target-precision remote sensing image data, and auxiliary image data. Step 202 involves performing remote sensing crop identification on the multi-source remote sensing image data to obtain the crop identification results for the target plot. This specifically includes the following steps: Based on basic remote sensing image data, a base map for crop identification is constructed, which is used to initially delineate crop distribution areas. For the first target area in the base map, local details are supplemented using target-precision remote sensing image data to obtain an enhanced crop distribution map. Based on auxiliary image data, image enhancement processing is performed on the second target area in the enhanced crop distribution map to generate a set of locally enhanced images. Ground survey data of the target plots is obtained. Based on the set of locally enhanced images, ground survey data, and plot attribute information, the enhanced crop distribution map is verified and corrected to determine the crop identification results of the target plots.

[0041] Basic remote sensing image data is collected by general-purpose sensors carried on remote sensing platforms such as satellites and aircraft, covering ground features of the target area. Target precision remote sensing image data is collected by remote sensing equipment equipped with high-resolution imaging sensors (such as high-orbit satellites and specialized remote sensing aircraft), providing refined imaging of the target area.

[0042] The auxiliary image data comes from multiple sources, including drone aerial photography and low-altitude remote sensing equipment. It can characterize additional information such as topographic relief, seasonal changes, and multi-angle landform morphology of the target area. This information is used to enhance key areas in the augmented crop distribution map, enriching the data dimensions to support subsequent verification and correction. The crop identification base map is an image carrier generated from preprocessed basic remote sensing image data. It primarily represents the approximate range and preliminary distribution outline of crop planting over a large area, clearly defining the basic boundaries of different land types. It serves as the initial benchmark for supplementing subsequent target precision image details and refining crop distribution analysis.

[0043] The first target area refers to regions in the base map for crop identification where, due to the resolution limitations of the basic remote sensing image data, the details of crop distribution (such as the overall planting area, boundaries between plots, and preliminary determination of crop type) are blurred or lack precision. The low-resolution data in these areas cannot meet the accuracy requirements for subsequent crop identification. By supplementing the local details with comprehensive remote sensing image data of the target precision, the crop distribution of the entire target plot can be depicted more accurately and completely.

[0044] Local detail enhancement, by overlaying high-resolution ground feature data, compensates for the deficiencies in small-scale details in the base map, making the representation of crop distribution boundaries and internal plot structures more accurate. Its core purpose is to improve the detail integrity and data reliability of the crop distribution map. The enhanced crop distribution map is a refined image result formed on the basis of the crop identification base map after local detail enhancement of the first target area. It integrates the advantages of the large-scale coverage of the base remote sensing data with the detail advantages of the target precision data, and can accurately represent the specific planting range, plot boundary coordinates, crop distribution density, and other information of different crops.

[0045] The second target area is a region selected from the enhanced crop distribution map that has identification questions or requires key verification, such as plots with ambiguous crop type determination, conflicting boundaries and plot attribute information, or abnormal insurance declaration information. For example, plots in the enhanced distribution map that are "declared as rice cultivation but whose image features suggest dryland crops" are designated as the second target area.

[0046] Among them, image enhancement processing is an information supplementation and image optimization technology process based on auxiliary image data obtained from channels such as UAV aerial photography and multispectral imaging. It addresses issues such as information ambiguity and lack of details (e.g., unclear crop boundaries and insufficient basis for type determination) in the second target area of ​​the enhanced crop distribution map.

[0047] The local enhanced image set is a multi-dimensional image data combination specifically corresponding to the second target region, formed after image enhancement processing. It represents the clear ground features of the second target region after supplementing key information (such as clear crop boundaries, distinguishable crop spectral details, and unobstructed landforms). It is mainly used to provide a refined verification basis for the enhanced crop distribution map, and the accuracy of subsequent correction is ensured by cross-comparison of multiple sets of images. For example, for the second target region with blurred boundaries due to vegetation occlusion, the local enhanced image set formed after image enhancement processing includes multiple sets of data such as multispectral images that supplement the details of the occluded area and clear boundary images from multiple angles.

[0048] The ground survey data consists of real ground data obtained through on-site visits, measurements, and sample collection by staff. This data covers core information such as actual plot area, crop type, growth stage, plot boundary coordinates, and soil conditions. Verification and correction involves combining locally enhanced image sets, ground survey data, and plot attribute information to verify the accuracy and correct errors in the enhanced crop distribution map. By comparing the consistency between remote sensing analysis results and actual field conditions and attribute data, misjudged areas in the image (such as misidentified crop type or boundary offset) are identified and adjusted.

[0049] In one example, using the insurance coverage of a contiguous 1,500-mu (approximately 100 hectares) corn plantation in a plain area as a scenario, crop identification was used to support the control of inflated risks. First, a 20-meter resolution satellite remote sensing image was acquired. After geometric correction and noise reduction, a base map for crop identification was constructed, initially delineating suspected corn planting areas. However, due to insufficient resolution, the boundaries and internal divisions of the plots were blurred. This entire 1,500-mu contiguous plot was designated as the first target area. A 0.5-meter resolution remote sensing image was used to supplement details, forming an enhanced crop distribution map covering all target plots. Two second target areas were selected: a 60-mu plot declared as corn but suspected of being soybeans, and a 45-mu plot whose boundary deviated by 20 meters from the land ownership information. A set of locally enhanced images was generated through UAV multispectral assisted image enhancement processing. Combined with field survey data (the first area confirmed as soybeans) and plot attribute information, the misjudgments and boundary deviations in the distribution map were corrected, yielding the crop identification results.

[0050] This application first constructs a base map for crop identification using basic remote sensing images, completing the initial division of crop distribution areas. Then, for the first target area with insufficient resolution in the base map, local details are supplemented using target-precision remote sensing images to form an enhanced crop distribution map covering the complete target plot. Subsequently, focusing on the second target area in the enhanced image where identification is questionable, a set of locally enhanced images is generated using auxiliary image enhancement processing. This is combined with ground survey data and plot attribute information obtained from the field to conduct multi-dimensional verification and correction, ultimately determining the accurate crop identification result. The entire process, through multi-source remote sensing data hierarchical optimization, precise breakthrough of questionable areas, and multi-data cross-verification, not only overcomes the shortcomings of traditional methods with their single data dimension but also solves the problem of the disconnect between manual verification and remote sensing data, significantly improving the comprehensiveness and accuracy of crop identification.

[0051] In some optional implementations, step 203, based on crop identification results and land parcel attribute information, establishes a spatiotemporal topological relationship model of the target land parcel, specifically including the following steps: Extracting plot boundary information and crop type data from crop identification results; constructing plot closure and adjacency / overlap relationships between the target plot and surrounding plots based on plot boundary information and coordinate comparison, thus obtaining a preliminary spatial topological framework; using crop type data, functionally partitioning the spatial topological framework to generate spatial functional zoning results; associating plot attribute information with spatial functional zoning results to construct a spatiotemporally related topological structure; extracting dual features of spatial relationships and attribute matching from the topological structure to determine boundary compliance verification rules and land use crop matching verification rules; filtering and correcting the topological structure using boundary compliance verification rules and land use crop matching verification rules to generate a spatiotemporal topological relationship model of the target plot.

[0052] Among them, the land parcel boundary information represents the set of geographic coordinates of the actual spatial range of the target land parcel, including core data such as the coordinates of the inflection points of the land parcel outline and the direction of the boundary lines. Crop type data is extracted from the remote sensing crop identification results and represents the core data of the actual types and distribution of crops planted on the target land parcel, covering related information such as crop varieties, planting area ratio, and growth stage.

[0053] Among them, parcel closure is a spatial characteristic constructed by comparing the extracted parcel boundary information with coordinates one by one, characterizing whether the target parcel boundary forms a complete and closed geographical area. The spatial topology framework is constructed based on the parcel closure characteristics, combined with the adjacency and overlap relationships between the target parcel and surrounding parcels through coordinate comparison. It is a basic model representing the spatial location relationship between multiple parcels, and its core purpose is to provide skeletal support for subsequent functional zoning and topology optimization.

[0054] Functional zoning is a process of selectively dividing the initially constructed spatial topology framework using crop type data extracted from crop identification results. Specifically, it refers to the functional division of land parcels according to crop use, growth characteristics, or planting type.

[0055] The spatial functional zoning result is the final output after functional zoning processing. It is a dataset representing the spatial distribution of the target plot and its surrounding area after being divided according to planting functions, including information such as the boundary coordinates, crop type, and area of ​​each functional area. The spatiotemporal relational topology represents the dynamic relationship between the spatial relationships of the plots and their temporal attributes and attribute information. Specifically, it refers to a comprehensive structure that simultaneously includes the spatial layout of the plots, functional zoning, attribute information, and changes over time.

[0056] Among them, the dual characteristics of spatial relationship and attribute adaptation are the core information set extracted from the spatiotemporal topological structure. On the one hand, it represents the spatial logical relationship between the target plot and the surrounding plots, such as adjacency and overlap. On the other hand, it represents the degree of adaptation between plot attributes (such as soil type and land use) and functional zoning (such as crop planting type).

[0057] Among them, the boundary compliance verification rule is a standardized judgment criterion formulated based on the extracted spatial relationship and attribute adaptation dual features, used to verify the legality, accuracy, and uniqueness of the target plot boundary. The land type crop matching verification rule is a judgment standard determined based on the spatial relationship and attribute adaptation dual features, used to verify whether the land type attributes (such as dry land, paddy field, and forest land) of the target plot and the actual crop type planted conform to agricultural production laws and land use requirements.

[0058] Among them, screening and correction refers to the process of systematically processing the spatiotemporal topology based on boundary compliance verification rules and land type crop matching verification rules. The core is to screen out abnormal data such as boundary violations and land type crop mismatches in the topology, and supplement, adjust or delete abnormal data.

[0059] In one example, a spatiotemporal topological model was constructed to support the identification of inflated risks, using a contiguous 1200-mu (approximately 80 hectares) agricultural insurance-insured plot in a county as the object. After obtaining crop identification results through remote sensing, the plot boundary information, including the latitude and longitude coordinates of 15 inflection points, and crop type data (900 mu for wheat and 300 mu for corn) were extracted. Coordinate comparison confirmed that the plot boundaries were closed and adjacent to the eastern woodland, with no overlap with the southern farmland, thus constructing a preliminary spatial topological framework. The framework was then divided into zones according to crop type, generating spatial functional zoning results for "wheat planting functional zone" and "corn planting functional zone." The attribute information of plots with loam soil type, a confirmed area of ​​1200 mu (approximately 80 hectares), and a contract period of 2023-2028 was associated with the zoning results to construct a spatiotemporal relational topological structure. From the structure, the dual features of "adjacent to the eastern woodland" and "loam soil suitable for wheat / corn" were extracted to determine boundary compliance verification rules (deviation from confirmed coordinates ≤ 5 meters) and land type crop matching verification rules (loam soil suitable for wheat / corn, unsuitable for rice). One abnormal data point with a boundary deviation of 8 meters was selected. After supplementing the remote sensing image and re-extracting the coordinates, an accurate spatiotemporal topological relationship model was generated.

[0060] This application extracts plot boundary information and crop type data from crop identification results, and constructs plot closure and surrounding adjacency and overlap relationships by combining coordinate comparison, forming a preliminary spatial topological framework. Then, functional zoning is completed using crop type data, and a spatiotemporal relational topological structure is constructed by associating plot attribute information. Dual features are extracted from the structure to determine two types of verification rules, which are then filtered and corrected to generate a spatiotemporal topological relationship model. The entire process, through deep fusion of multi-source data and spatial logic construction, breaks through the limitations of traditional methods with a single data dimension, solves the problem of disconnect between manual verification and remote sensing data, and allows plot spatial relationships, crop types, and attribute information to form a closed-loop verification, significantly improving the accuracy and comprehensiveness of subsequent false inflation risk identification.

[0061] In some optional implementations, step S204, based on underwriting data, uses a spatiotemporal topological relationship model to identify the risk of inflated claims, obtaining multi-dimensional inflated risk identification results, specifically including the following steps: A spatiotemporal topology model is used to perform structured parsing of the underwriting data to obtain key verification fields, including the declared land parcel boundary information and the declared crop type information. Based on boundary compliance verification rules, the declared land parcel boundary information is compared with the actual boundary information in the spatiotemporal topology model to obtain boundary verification results. Based on land type and crop matching verification rules, the declared crop type information is matched with the actual land type information of the land parcel in the spatiotemporal topology model to obtain land type and crop verification results. Based on the boundary verification results and land type and crop verification results, risk points of the target land parcel in the spatiotemporal and attribute dimensions are identified. A comprehensive risk quantification assessment is performed on the risk points to generate multi-dimensional false risk identification results.

[0062] Among them, the key verification fields represent the set of fields that represent the core information of the insurance application for the target plot and require key verification. Their core purpose is to provide clear comparison basis for subsequent boundary verification and land type crop matching verification, ensuring that the verification work is carried out in a targeted manner. The declared plot boundary information represents the spatial range of the target plot declared by the insured at the time of insurance application. Specifically, it includes geographical information such as the coordinates of the declared plot inflection points and the direction of the boundary. It is mainly used to compare with the actual boundary information in the spatiotemporal topological relationship model to complete the boundary compliance verification.

[0063] The declared crop type information represents the types and distribution of crops actually planted on the target plot declared by the insured, including crop varieties and planting area percentages. Its core purpose is to match and verify this information with the actual land use information in the spatiotemporal topological relationship model to determine whether the declared crops and land use are compatible. The boundary verification result represents the consistency and compliance of the declared boundary with the actual boundary, primarily used to identify risks such as inflated or offset boundaries in the declared plot. Specifically, it can include judgments such as "compliance," "excessive boundary deviation," and "boundary overlap," along with relevant quantitative data.

[0064] Among them, the land type and crop verification result characterizes the compatibility between the declared crop type and the actual land type of the plot. Its core purpose is to determine whether there is any risk of inflated claims related to "mismatch between land type and crop". For example, if the actual land type of a plot is dry land (extracted from the spatiotemporal topological relationship model), while the declared crop type is rice (which requires paddy field planting), after matching and verification, the generated land type and crop verification result is "mismatch between land type and crop, with the risk of fictitious insurance claims", which provides a direct basis for subsequent risk point identification.

[0065] Among them, the risk points in the spatiotemporal dimension and the attribute dimension are risk information derived from the analysis of boundary verification results and land type and crop verification results. The spatiotemporal dimension risk points characterize the abnormal risks of the target plot in terms of spatial range (such as boundary deviation and overlap) and temporal adaptability. The attribute dimension risk points characterize the abnormal risks associated with attributes such as the adaptability of the plot's land type and the declared crop type. The core purpose is to clarify the specific manifestations and distribution dimensions of the inflated risk. For example, after verification, it was found that "the declared boundary is 20 mu larger than the actual boundary (spatiotemporal dimension risk point)" and "rice was declared to be planted on dry land (attribute dimension risk point)". These two types of risk points together constitute the core clues of the inflated risk of the plot.

[0066] Among them, comprehensive risk quantitative assessment is a systematic assessment process based on risk points in the spatiotemporal and attribute dimensions. It refers to the use of a pre-set quantitative assessment model and indicator system to quantitatively score and comprehensively analyze the severity, scope of impact, and probability of occurrence of various risk points. Its core purpose is to transform scattered risk points into multi-dimensional quantitative risk results. For example, if a plot of land has two risk points, through comprehensive risk quantitative assessment, "20 mu of falsely added boundary" is assigned a quantitative score of 85 points (high risk), and "mismatch between land type and crop" is assigned a quantitative score of 70 points (medium-high risk). Finally, a multi-dimensional false risk identification result containing risk level and quantitative score is generated.

[0067] In one example, a 1000-mu (approximately 67 hectares) agricultural insurance-insured plot in a county was used as the subject to identify the risk of inflated insured land area. Based on the crop identification results and plot attribute information such as soil type and land ownership information, a spatiotemporal topological model was established. This model was used to perform structured parsing of the insured data, extracting key verification fields, including the declared plot boundary information (latitude and longitude coordinates of 8 inflection points) and the declared crop type information (1000 mu of wheat). According to the boundary compliance verification rule (deviation ≤ 5 meters), the declared boundary was compared with the actual boundary in the model. It was found that the declared boundary was offset 8 meters to the north, resulting in a boundary verification result of "boundary deviation exceeding the standard". Based on the land type crop matching verification rule (dryland is suitable for planting wheat and corn), the declared crop type was compared with the actual land type (dryland) in the model. Because wheat is suitable for dryland, a verification result of "land type crop matching compliance" was obtained. Combining the two results, a spatiotemporal risk point (approximately 60 mu of inflated boundary area) was identified, while no risk points were found in the attribute dimension. A comprehensive risk quantification assessment was conducted on this risk point, and it was assigned a score of 82 (high risk) based on indicators such as the proportion of deviation area and compliance threshold. Finally, a multi-dimensional risk identification result of "high risk caused by boundary inflation" was generated.

[0068] This application embodiment uses a spatiotemporal topological relationship model to perform structured analysis of the underwriting data, accurately extracting key verification fields such as the boundary information of the declared land parcels and the declared crop type information. Then, through boundary compliance verification rules and land type / crop matching verification rules, the declared information is compared with the actual boundaries and actual land type information in the model, generating two types of verification results. Based on the verification results, risk points in the spatiotemporal dimension (such as boundary deviation) and attribute dimension (such as land type / crop mismatch) are accurately identified, and then a comprehensive risk quantification assessment is carried out to form a multi-dimensional false risk identification result. The entire process achieves closed-loop verification between declared data and actual data, breaking the limitations of the single data dimension of traditional methods, solving the problem of disconnect between manual verification and remote sensing data, and making risk identification more targeted and comprehensive.

[0069] In some optional implementations, step 205, determining the risk threshold set for the target land parcel based on multi-dimensional risk factors, specifically includes the following steps: Based on multi-dimensional risk factors, a basic threshold set and a factor weight set are obtained; the multi-dimensional risk factors are quantified to obtain a standardized score for each risk factor; based on the factor weight set, the standardized scores are weighted and summed to obtain the comprehensive risk score of the target plot; based on the comprehensive risk score, the basic threshold set is adjusted to obtain the initial risk threshold set of the target plot; the initial risk threshold set is verified by calling the preset threshold constraint rules through the preset rule engine to obtain the risk threshold set of the target plot.

[0070] The basic threshold set originates from a pre-set risk factor library. It is an initial risk assessment benchmark set established based on industry risk control standards, historical claims data, and the insurance characteristics of different crops / regions. The factor weight set is a set of weight values ​​determined through expert evaluation and big data statistical analysis, based on the degree of influence of multi-dimensional risk factors on the identification of inflated risks. It represents the importance proportion of each risk factor in the comprehensive risk assessment.

[0071] The standardized score is a uniform dimensional score obtained by normalizing multi-dimensional risk factors obtained from the risk factor library, representing the relative risk level of each risk factor within its dimension. The comprehensive risk score represents the quantitative level of the overall inflated risk of the target land parcel, mainly used to reflect the risk level of the land parcel under the combined effect of multiple risk factors.

[0072] The initial risk threshold set represents the critical values ​​for each dimension after adjustment based on the actual risk level of the target plot. Its core purpose is to replace fixed static thresholds, making risk assessments more closely aligned with the specific risk conditions of the plot. The rule engine is mainly used to invoke preset threshold constraint rules to perform compliance checks and logical corrections on the initial risk threshold set, ensuring that the threshold settings meet risk control requirements.

[0073] Among them, the threshold constraint rule represents the adjustment boundary and compliance standard of the initial risk threshold set. Its core purpose is to verify the rationality and compliance of the initial risk threshold through the rule engine, and to avoid the threshold adjustment deviating excessively from the actual risk control needs.

[0074] In one example, an 800-mu (approximately 53 hectares) corn planting insurance plot in a certain county was used as the subject to determine the risk threshold set. Multi-dimensional risk factors (area difference rate, crop matching degree, boundary compliance) for this plot were obtained from a pre-set risk factor library. Simultaneously, a basic threshold set (area difference rate ≤ 5%, matching degree ≥ 85%, compliance ≥ 90%) and a factor weight set (0.4, 0.3, 0.3) were obtained. The risk factors were quantified: an area difference rate of 10% corresponded to a standardized score of 80 points, a matching degree of 80% corresponded to 60 points, and compliance of 85% corresponded to 70 points. Based on the weighted summation of the factor weight set, a comprehensive risk score of (80 × 0.4 + 60 × 0.3 + 70 × 0.3) = 71 points was obtained. Based on this score, the basic thresholds were adjusted to the initial risk threshold set (area difference rate ≤ 3%, matching degree ≥ 88%, compliance ≥ 92%). By calling the threshold constraint rules through the preset rule engine (adjustment range ≤30%), the verification found that the area difference rate threshold was adjusted from 5% to 3%, which was 40% higher than the constraint standard. Therefore, it was corrected to ≤3.5% (adjustment range 30%, which meets the requirements). The final set of risk thresholds is: area difference rate ≤3.5%, land type crop matching degree ≥88%, and boundary compliance ≥92%, which provides accurate standards for subsequent risk verification.

[0075] This application first obtains a basic threshold set and a factor weight set based on multi-dimensional risk factors. Then, it quantifies each risk factor to obtain a standardized score. A comprehensive risk score is calculated by weighted summation of the factor weight sets. Based on this score, the basic thresholds are dynamically adjusted to obtain an initial risk threshold set. Finally, the initial thresholds are verified by calling threshold constraint rules through a rule engine to form the final risk threshold set. This entire process overcomes the limitations of traditional static thresholds, dynamically optimizing thresholds based on the actual importance of risk factors and the overall risk level of the land parcel. Simultaneously, constraint rules ensure the rationality of the thresholds, making risk assessment more aligned with the insurance characteristics of different crops and regions. This solves the drawbacks of a one-size-fits-all approach to early warning and provides a more adaptable standard for accurate risk verification.

[0076] In some optional implementations, the multi-dimensional inflated risk identification results include assessment data from multiple risk dimensions. Step 206 involves performing risk verification on the multi-dimensional inflated risk identification results based on a set of risk thresholds, generating differentiated warnings. This specifically includes the following steps: The assessment data for each risk dimension is compared with the corresponding risk threshold in the risk threshold set to obtain the comparison results. Based on the comparison results, the target risk dimension with risk is identified from multiple risk dimensions, and the corresponding risk warning information is generated. Based on the risk warning information and the preset warning level rules, the differentiated warning level is determined. According to the differentiated warning level and the risk warning information, the warning strategy for the target plot is generated to issue warnings.

[0077] The assessment data for multiple risk dimensions characterizes the specific risk quantification or status description of the target plot in different risk dimensions (such as area difference, crop matching, and boundary compliance). For example, the assessment data for a corn-growing plot includes "area difference rate of 12%, crop matching degree of 75%, and boundary compliance of 80%", which correspond to the specific risk assessment results of the three risk dimensions.

[0078] The comparison results characterize the degree of fit (meeting or exceeding) between the assessment data and the thresholds for each risk dimension, and are mainly used to accurately locate the target risk dimensions where risks exist. For example, comparing a plot's "area difference rate of 12%" (assessment data) with the corresponding threshold "≤5%" yields a comparison result of "area difference rate exceeding the standard by 7 percentage points"; comparing "land type crop matching degree of 75%" with the threshold "≥85%" yields a comparison result of "land type crop matching degree not meeting the standard".

[0079] The risk warning information represents the specific dimensions of the risk, the degree of risk exceeding the standard, and the core risk characteristics. Its core purpose is to clearly inform the public of key risk information. For example, for the target risk dimensions of "area difference rate exceeding the standard by 7 percentage points" and "land type crop matching degree not meeting the standard", the generated risk warning information is "significant risk of inflated target land area, difference rate exceeding the threshold by 7 percentage points; insufficient matching between land type and declared crop, suspected of fabricating insured object".

[0080] Among them, the warning level rule represents the correspondence between the severity of the risk reflected by the risk warning information and the warning level, and is mainly used to transform abstract risk information into a clear warning level.

[0081] Among them, the differentiated early warning level is a specific early warning level determined based on the severity of the risk corresponding to the risk warning information and the early warning level rules. It represents the urgency and priority of handling the inflated risk of the target plot. Its core purpose is to distinguish the early warning intensity of different risk conditions and provide direction for the formulation of targeted early warning strategies.

[0082] Among them, the early warning strategy is a specific risk management plan formulated by combining differentiated early warning levels and risk warning information, which represents the response measures, implementation process and responsible parties for different early warning levels and risk characteristics.

[0083] In one example, a risk warning for inflated insurance premiums was conducted on a 600-mu (approximately 40 hectares) wheat planting insurance plot in a certain county. The known risk thresholds for this plot are (area difference rate ≤ 3.5%, land type crop matching degree ≥ 88%, boundary compliance ≥ 92%). The assessment data in the multi-dimensional inflated risk identification results are (area difference rate 6%, matching degree 85%, compliance 90%). Comparing the assessment data of each dimension with the corresponding thresholds yields the following results: area difference rate exceeds the standard by 2.5 percentage points, matching degree falls short by 3 percentage points, and compliance falls short by 2 percentage points. Based on these results, area difference, land type matching, and boundary compliance are identified as the target risk dimensions, generating a risk warning message: "The inflated area of ​​the plot exceeds the standard significantly; the land type and wheat matching degree and boundary compliance do not meet the standards, indicating a risk of inflated insurance premiums." According to the preset warning level rules (two or more dimensions failing to meet the standards and one dimension exceeding the standard by more than 2 percentage points constitutes a medium warning), the differentiated warning level is determined to be a medium warning. Based on this level and warning information, an early warning strategy is generated: a manual on-site verification process is initiated, requiring the insured to submit land ownership and planting certificates within 3 working days. The verification results will serve as the basis for whether to underwrite or adjust the insured amount.

[0084] This application first compares the assessment data of each risk dimension with the corresponding risk threshold one by one to form a comparison result that accurately reflects whether the risk of each dimension exceeds the standard. Then, based on the comparison result, it identifies the target risk dimension with risk and generates risk warning information that includes the risk dimension and the degree of exceedance. Combining the preset warning level rules, it determines the differentiated warning level according to the severity of the risk warning information. Finally, it formulates targeted warning strategies based on the warning level and warning information. The whole process realizes a closed loop from risk data comparison to warning generation, level determination and strategy formulation. It not only covers multi-dimensional risk screening, but also avoids one-size-fits-all warnings through differentiated levels and strategies, and is tailored to the insurance characteristics of different crops and regions, solving the problems of weak targeting and untimely risk detection in traditional warnings.

[0085] In some optional implementations, after step 206, which involves performing risk verification on the multi-dimensional false risk identification results based on risk thresholds and generating differentiated warnings, the following steps are also included: Based on the early warning strategy, a corresponding early warning task is generated, and the early warning task is linked to the underwriting data. For the linkage, an operation log is recorded, which includes the operation time and operation content. Based on the operation time and operation content, a full-link data tracking record is generated, which is used to identify the processing flow of the early warning task. If the full-link data tracking record shows that the early warning task is not completed, a prompt message is generated to update the operation log. Based on the update status of the operation log, the closed-loop status of the early warning task is determined.

[0086] Among them, the early warning task is a specific execution item generated based on the differentiated early warning strategy. It represents the implementation instructions for handling the inflated risks of the target plot. Its core purpose is to clarify the specific content of risk handling, the responsible party and the completion time limit, and promote the transformation of the early warning strategy from planning to execution.

[0087] Among them, the correlation represents the corresponding attribution relationship between the early warning task and the underwriting data. It is mainly used to realize the linkage query of risk disposal and original insurance information, so as to ensure that each early warning task can be traced back to the corresponding insurance source.

[0088] Among them, the operation log is a real-time record of key actions in the entire process of early warning handling, which is based on the correlation between early warning tasks and underwriting data.

[0089] Among them, the full-link data tracking record is a complete record document generated by data integration and time-series analysis based on the operation time and operation content in the operation log. It represents the entire process nodes of the early warning task from generation to disposal. It is mainly used to intuitively display the processing progress and unfinished links of the early warning task, and provide a visual basis for the closed-loop management of risk disposal.

[0090] The notification message is an automatically generated reminder message that is triggered by the system when the end-to-end data tracking records show that the warning task has not been completed within the specified time limit. For example, for a task where the verification report has not been submitted on time, a system pop-up window or text message will be generated saying "The warning verification task for plot A001 is overdue by 2 days. Please submit the verification report as soon as possible."

[0091] The closed-loop status is the final handling result of the early warning task determined by the update status of the operation log (whether the incomplete operation has been completed, whether the task objective has been achieved). It represents whether the entire process of the early warning task from generation to completion is closed-loop. It is mainly used to judge whether the risk handling has been implemented and effective, and to ensure that each early warning task can be fully processed.

[0092] In one example, taking a 500-mu (approximately 33 hectares) rice planting insurance plot in a certain county as the object, the early warning strategy generated after risk verification is "Initiate secondary verification, complete on-site verification of the plot within 7 working days." Based on this strategy, a corresponding early warning task is generated, including the verification timeframe, responsible department, and verification indicators. Simultaneously, a correlation is established between this task and the underwriting data (policyholder information, plot coordinates, declared area, etc.) corresponding to policy number C202406, ensuring the task is traceable back to the source of the insurance. For this correlation, the operation log is recorded in real time: the early warning task was generated at 10:00 on June 1, 2024, and assigned to the verification team at 09:15 on June 3, 2024. Based on the operation time and content, a full-link data tracking record is generated, identifying the "task generation - assignment - pending verification" process. As of 17:00 on June 8, 2024, the tracking record shows that the task was not completed, and the system automatically generates the message "Verification task is overdue by 1 day, please proceed as soon as possible" and updates the operation log. The subsequent verification team submitted a verification report at 16:00 on June 9, 2024. The log was updated synchronously. Based on the complete update status of the log, the closed-loop status of the early warning task was determined to be "closed".

[0093] This application's embodiments generate targeted early warning tasks based on early warning strategies and establish a correlation with underwriting data to ensure that risk handling is traceable back to the source of insurance. By recording operation logs containing operation time and content, complete data support is provided for early warning task processing. Then, based on the logs, a full-link data tracking record is generated, clearly identifying the entire task processing process. When the tracking record shows that the task is not completed, a prompt message is generated to update the log, urging timely action. Finally, the closed-loop status is determined based on the log update status. The entire process achieves full-process control of early warning tasks from generation and tracking to closed-loop, solving the problems of traceable risk handling and delayed progress in traditional methods. Simultaneously, by linking underwriting data to ensure traceability, it improves the standardization and efficiency of controlling inflated risks in agricultural insurance.

[0094] It should be emphasized that, in order to further ensure the privacy and security of the aforementioned underwriting data, land parcel attribute information, multi-source remote sensing image data, crop identification results, land parcel attribute information, multi-dimensional inflated risk identification results, and risk threshold set, the aforementioned underwriting data, land parcel attribute information, multi-source remote sensing image data, crop identification results, land parcel attribute information, multi-dimensional inflated risk identification results, and risk threshold set can also be stored in a blockchain node.

[0095] The blockchain referred to in this application is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked together using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include an underlying blockchain platform, a platform product service layer, and an application service layer.

[0096] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0097] Further reference Figure 3 As a response to the above Figure 2 To implement the method shown, this application provides an embodiment of a risk warning device, which is similar to... Figure 2 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0098] like Figure 3 As shown, the risk warning device 400 of this embodiment includes: an acquisition module 401, a crop identification module 402, an establishment module 403, a risk identification module 404, a first determination module 405, and a verification module 406. Wherein: The acquisition module 401 is used to acquire the insurance data, land attribute information and multi-source remote sensing image data of the target land parcel; The crop identification module 402 is used to perform remote sensing crop identification on multi-source remote sensing image data to obtain crop identification results for the target plot; Module 403 is established to build a spatiotemporal topological relationship model of the target plot based on crop identification results and plot attribute information; The risk identification module 404 is used to identify the risk of inflated claims based on underwriting data and using a spatiotemporal topological relationship model to obtain multi-dimensional inflated risk identification results. The first determining module 405 is used to obtain multi-dimensional risk factors of the target land plot from a preset risk factor library, and determine the risk threshold set of the target land plot based on the multi-dimensional risk factors. The verification module 406 is used to perform risk verification on the multi-dimensional false risk identification results based on the risk threshold set, and generate differentiated early warnings.

[0099] This application's embodiments overcome the limitations of traditional single-data-dimensional approaches by integrating underwriting data, land parcel attribute information, and multi-source remote sensing image data, achieving comprehensive coverage and identification of inflated risks and significantly reducing the probability of missed detections. Based on remote sensing crop identification results and land parcel attribute information, a spatiotemporal topological relationship model is constructed, establishing an efficient matching and verification mechanism between declared and actual data. This replaces inefficient manual visual comparison, improving the accuracy and efficiency of inflated risk identification. A dynamic risk threshold set is determined by combining multi-dimensional risk factors, replacing the traditional static one-size-fits-all threshold. This adapts to the differences in insurance characteristics of different crops and regions, generating targeted and differentiated early warnings, significantly optimizing the early warning effect. This enables accurate and efficient identification and early warning of inflated risks in agricultural insurance targets.

[0100] In one embodiment, the crop identification module 402 includes: The first construction submodule is used to construct a basic base map for crop identification based on basic remote sensing image data. The basic base map is used to initially delineate the crop distribution area. The supplementary submodule is used to supplement local details of the first target area in the base map using target-precision remote sensing image data to obtain an enhanced crop distribution map. The acquisition submodule is used to perform image enhancement processing on the second target region in the enhanced crop distribution map based on auxiliary image data, and generate a set of locally enhanced images; The first acquisition submodule is used to acquire ground survey data of the target plot; The verification submodule is used to verify and correct the enhanced crop distribution map based on the local enhanced image set, ground survey data and plot attribute information, and determine the crop identification result of the target plot.

[0101] This application first constructs a base map for crop identification using basic remote sensing images, completing the initial division of crop distribution areas. Then, for the first target area with insufficient resolution in the base map, local details are supplemented using target-precision remote sensing images to form an enhanced crop distribution map covering the complete target plot. Subsequently, focusing on the second target area in the enhanced image where identification is questionable, a set of locally enhanced images is generated using auxiliary image enhancement processing. This is combined with ground survey data and plot attribute information obtained from the field to conduct multi-dimensional verification and correction, ultimately determining the accurate crop identification result. The entire process, through multi-source remote sensing data hierarchical optimization, precise breakthrough of questionable areas, and multi-data cross-verification, not only overcomes the shortcomings of traditional methods with their single data dimension but also solves the problem of the disconnect between manual verification and remote sensing data, significantly improving the comprehensiveness and accuracy of crop identification.

[0102] In one embodiment, the establishment module 403 includes: The first extraction submodule is used to extract plot boundary information and crop type data from the crop identification results; The second construction submodule is used to construct the closure of the target plot and the adjacency and overlap relationship between the target plot and surrounding plots based on the plot boundary information and coordinate comparison, so as to obtain a preliminary spatial topology framework. The partitioning submodule is used to perform functional partitioning of the spatial topology framework using crop type data, and generate spatial functional partitioning results. The association submodule is used to associate land parcel attribute information with spatial functional zoning results to construct a spatiotemporal association topology. The second extraction submodule is used to extract dual features of spatial relationship and attribute adaptation from the topology, and to determine the boundary compliance verification rules and land type crop matching verification rules. The correction submodule is used to filter the topology structure and generate a spatiotemporal topological relationship model of the target plot by using boundary compliance verification rules and land type crop matching verification rules.

[0103] This application extracts plot boundary information and crop type data from crop identification results, and constructs plot closure and surrounding adjacency and overlap relationships by combining coordinate comparison, forming a preliminary spatial topological framework. Then, functional zoning is completed using crop type data, and a spatiotemporal relational topological structure is constructed by associating plot attribute information. Dual features are extracted from the structure to determine two types of verification rules, which are then filtered and corrected to generate a spatiotemporal topological relationship model. The entire process, through deep fusion of multi-source data and spatial logic construction, breaks through the limitations of traditional methods with a single data dimension, solves the problem of disconnect between manual verification and remote sensing data, and allows plot spatial relationships, crop types, and attribute information to form a closed-loop verification, significantly improving the accuracy and comprehensiveness of subsequent false inflation risk identification.

[0104] In one embodiment, the risk identification module 404 includes: The parsing submodule is used to perform structured parsing of the underwriting data using a spatiotemporal topological relationship model to obtain key verification fields, including the boundary information of the declared land parcel and the information of the declared crop type. The boundary verification submodule is used to perform boundary verification between the declared land parcel boundary information and the actual boundary information in the spatiotemporal topological relationship model based on the boundary compliance verification rules, and obtain the boundary verification result. The matching and verification submodule is used to match and verify the declared crop type information with the actual land type information of the plot in the spatiotemporal topological relationship model based on the land type crop matching verification rules, and obtain the land type crop verification result. The risk point identification submodule is used to identify risk points of the target plot in the spatiotemporal and attribute dimensions based on the boundary verification results and land type and crop verification results. The assessment submodule is used to conduct a comprehensive quantitative assessment of risk points and generate multi-dimensional results for identifying inflated risks.

[0105] This application embodiment uses a spatiotemporal topological relationship model to perform structured analysis of the underwriting data, accurately extracting key verification fields such as the boundary information of the declared land parcels and the declared crop type information. Then, through boundary compliance verification rules and land type / crop matching verification rules, the declared information is compared with the actual boundaries and actual land type information in the model, generating two types of verification results. Based on the verification results, risk points in the spatiotemporal dimension (such as boundary deviation) and attribute dimension (such as land type / crop mismatch) are accurately identified, and then a comprehensive risk quantification assessment is carried out to form a multi-dimensional false risk identification result. The entire process achieves closed-loop verification between declared data and actual data, breaking the limitations of the single data dimension of traditional methods, solving the problem of disconnect between manual verification and remote sensing data, and making risk identification more targeted and comprehensive.

[0106] In one embodiment, the first determining module 405 includes: The second acquisition submodule is used to acquire the basic threshold set and factor weight set based on multi-dimensional risk factors; The quantification submodule is used to quantify multi-dimensional risk factors and obtain a standardized score for each risk factor. The weighted submodule is used to perform a weighted summation of standardized scores based on the set of factor weights to obtain the comprehensive risk score of the target land parcel. The adjustment submodule is used to adjust the basic threshold set based on the comprehensive risk score to obtain the initial risk threshold set for the target land parcel. The verification submodule is used to verify the initial risk threshold set by calling the preset threshold constraint rules through the preset rule engine, and obtain the risk threshold set of the target land parcel.

[0107] This application first obtains a basic threshold set and a factor weight set based on multi-dimensional risk factors. Then, it quantifies each risk factor to obtain a standardized score. A comprehensive risk score is calculated by weighted summation of the factor weight sets. Based on this score, the basic thresholds are dynamically adjusted to obtain an initial risk threshold set. Finally, the initial thresholds are verified by calling threshold constraint rules through a rule engine to form the final risk threshold set. This entire process overcomes the limitations of traditional static thresholds, dynamically optimizing thresholds based on the actual importance of risk factors and the overall risk level of the land parcel. Simultaneously, constraint rules ensure the rationality of the thresholds, making risk assessment more aligned with the insurance characteristics of different crops and regions. This solves the drawbacks of a one-size-fits-all approach to early warning and provides a more adaptable standard for accurate risk verification.

[0108] In one embodiment, the verification module 406 includes: The comparison submodule is used to compare the assessment data of each risk dimension with the corresponding risk threshold in the risk threshold set to obtain the comparison results; The first generation submodule is used to determine the target risk dimension with risk from multiple risk dimensions based on the comparison results, and generate the risk warning information corresponding to the target risk dimension. The determination submodule is used to determine differentiated warning levels based on risk warning information and preset warning level rules; The second generation submodule is used to generate early warning strategies for target plots based on differentiated early warning levels and risk warning information, so as to issue early warnings.

[0109] This application first compares the assessment data of each risk dimension with the corresponding risk threshold one by one to form a comparison result that accurately reflects whether the risk of each dimension exceeds the standard. Then, based on the comparison result, it identifies the target risk dimension with risk and generates risk warning information that includes the risk dimension and the degree of exceedance. Combining the preset warning level rules, it determines the differentiated warning level according to the severity of the risk warning information. Finally, it formulates targeted warning strategies based on the warning level and warning information. The whole process realizes a closed loop from risk data comparison to warning generation, level determination and strategy formulation. It not only covers multi-dimensional risk screening, but also avoids one-size-fits-all warnings through differentiated levels and strategies, and is tailored to the insurance characteristics of different crops and regions, solving the problems of weak targeting and untimely risk detection in traditional warnings.

[0110] In one embodiment, the risk warning device 400 further includes: The first generation module is used to generate corresponding early warning tasks according to the early warning strategy, and the early warning tasks are associated with the underwriting data. The recording module is used to record operation logs for relationships. The operation logs include the operation time and operation content. The second generation module is used to generate full-link data tracking records based on operation time and operation content. The full-link data tracking records are used to identify the processing flow of the early warning task. The third generation module is used to generate a prompt message to update the operation log if the end-to-end data tracking record shows that the warning task has not been completed. The second determination module is used to determine the closed-loop status of the early warning task based on the update status of the operation log.

[0111] This application's embodiments generate targeted early warning tasks based on early warning strategies and establish a correlation with underwriting data to ensure that risk handling is traceable back to the source of insurance. By recording operation logs containing operation time and content, complete data support is provided for early warning task processing. Then, based on the logs, a full-link data tracking record is generated, clearly identifying the entire task processing process. When the tracking record shows that the task is not completed, a prompt message is generated to update the log, urging timely action. Finally, the closed-loop status is determined based on the log update status. The entire process achieves full-process control of early warning tasks from generation and tracking to closed-loop, solving the problems of traceable risk handling and delayed progress in traditional methods. Simultaneously, by linking underwriting data to ensure traceability, it improves the standardization and efficiency of controlling inflated risks in agricultural insurance.

[0112] To address the aforementioned technical problems, embodiments of this application also provide a computer device. Please refer to [link / reference needed]. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0113] Computer device 6 includes a memory 61, a processor 62, and a network interface 63 that are interconnected via a system bus. It should be noted that only computer device 6 with memory 61, processor 62, and network interface 63 is shown in the figure; however, it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented alternatively. Those skilled in the art will understand that the computer device described herein is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0114] Computer devices can include desktop computers, laptops, handheld computers, and cloud servers. These devices allow for human-computer interaction with users through keyboards, mice, remote controls, touchpads, or voice-activated devices.

[0115] The memory 61 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 61 may be an internal storage unit of the computer device 6, such as the hard disk or memory of the computer device 6. In this embodiment, the memory 61 is typically used to store the operating system and various application software installed on the computer device 6, such as computer-readable instructions for risk warning methods. In addition, the memory 61 can also be used to temporarily store various types of data that have been output or will be output.

[0116] In some embodiments, processor 62 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip. This processor 62 is typically used to control the overall operation of the computer device 6. In this embodiment, processor 62 is used to execute computer-readable instructions stored in memory 61 or to process data, such as computer-readable instructions for executing a risk warning method.

[0117] The network interface 63 may include a wireless network interface or a wired network interface, which is typically used to establish a communication connection between the computer device 6 and other electronic devices.

[0118] This application's embodiments overcome the limitations of traditional single-data-dimensional approaches by integrating underwriting data, land parcel attribute information, and multi-source remote sensing image data, achieving comprehensive coverage and identification of inflated risks and significantly reducing the probability of missed detections. Based on remote sensing crop identification results and land parcel attribute information, a spatiotemporal topological relationship model is constructed, establishing an efficient matching and verification mechanism between declared and actual data. This replaces inefficient manual visual comparison, improving the accuracy and efficiency of inflated risk identification. A dynamic risk threshold set is determined by combining multi-dimensional risk factors, replacing the traditional static one-size-fits-all threshold. This adapts to the differences in insurance characteristics of different crops and regions, generating targeted and differentiated early warnings, significantly optimizing the early warning effect. This enables accurate and efficient identification and early warning of inflated risks in agricultural insurance targets.

[0119] This application also provides another embodiment, namely, providing a computer-readable storage medium storing computer-readable instructions that can be executed by at least one processor to cause the at least one processor to perform the steps of the risk warning method described above.

[0120] This application's embodiments overcome the limitations of traditional single-data-dimensional approaches by integrating underwriting data, land parcel attribute information, and multi-source remote sensing image data, achieving comprehensive coverage and identification of inflated risks and significantly reducing the probability of missed detections. Based on remote sensing crop identification results and land parcel attribute information, a spatiotemporal topological relationship model is constructed, establishing an efficient matching and verification mechanism between declared and actual data. This replaces inefficient manual visual comparison, improving the accuracy and efficiency of inflated risk identification. A dynamic risk threshold set is determined by combining multi-dimensional risk factors, replacing the traditional static one-size-fits-all threshold. This adapts to the differences in insurance characteristics of different crops and regions, generating targeted and differentiated early warnings, significantly optimizing the early warning effect. This enables accurate and efficient identification and early warning of inflated risks in agricultural insurance targets.

[0121] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods of the various embodiments of this application.

[0122] The software tools or components not belonging to our company that appear in the embodiments of this application are merely examples and do not represent actual use.

Claims

1. A risk warning method, characterized in that, Includes the following steps: Acquire insurance data, land parcel attribute information, and multi-source remote sensing image data for the target land parcel; Remote sensing crop identification is performed on the multi-source remote sensing image data to obtain the crop identification results of the target plot; Based on the crop identification results and the land parcel attribute information, a spatiotemporal topological relationship model of the target land parcel is established; Based on the underwriting data, the spatiotemporal topological relationship model is used to identify the risk of inflated insured assets, resulting in multi-dimensional inflated risk identification results. Obtain multi-dimensional risk factors for the target land parcel from a preset risk factor library, and determine the risk threshold set for the target land parcel based on the multi-dimensional risk factors; Based on the set of risk thresholds, the multi-dimensional false risk identification results are verified to generate differentiated early warnings.

2. The method according to claim 1, characterized in that, The multi-source remote sensing image data includes basic remote sensing image data, target-precision remote sensing image data, and auxiliary image data. The step of performing remote sensing crop identification on the multi-source remote sensing image data to obtain the crop identification result of the target plot specifically includes: Based on the aforementioned basic remote sensing image data, a basic base map for crop identification is constructed, which is used to initially delineate crop distribution areas. For the first target area in the base map, the target precision remote sensing image data is used to supplement local details to obtain an enhanced crop distribution map; Based on the auxiliary image data, the second target region in the enhanced crop distribution map is subjected to image enhancement processing to generate a set of locally enhanced images; Obtain ground survey data for the target land parcel; Based on the local enhanced image set, the ground survey data, and the plot attribute information, the enhanced crop distribution map is verified and corrected to determine the crop identification result of the target plot.

3. The method according to claim 1, characterized in that, The step of establishing a spatiotemporal topological relationship model of the target plot based on the crop identification results and the plot attribute information specifically includes: Extract plot boundary information and crop type data from the crop identification results; Based on the land parcel boundary information, the land parcel closure of the target land parcel and the adjacency and overlap relationships between the target land parcel and surrounding land parcels are constructed through coordinate comparison, thus obtaining a preliminary spatial topology framework; Using the crop type data, the spatial topology framework is functionally partitioned to generate spatial functional partitioning results; The land parcel attribute information is associated with the spatial functional zoning results to construct a spatiotemporal related topology; Extract the dual features of spatial relationship and attribute adaptation from the topology to determine the boundary compliance verification rules and land type crop matching verification rules; The topology is filtered and corrected using the boundary compliance verification rules and the land type crop matching verification rules to generate a spatiotemporal topological relationship model of the target land parcel.

4. The method according to claim 3, characterized in that, The step of identifying the inflated risk of the target asset based on the underwriting data and using the spatiotemporal topological relationship model to obtain multi-dimensional inflated risk identification results specifically includes: Using the aforementioned spatiotemporal topology model, the underwriting data is structured and parsed to obtain key verification fields, which include the boundary information of the declared land parcel and the information of the declared crop type. Based on the boundary compliance verification rules, the boundary information of the declared land parcel is verified against the actual boundary information in the spatiotemporal topology model to obtain the boundary verification result. Based on the land use crop matching verification rules, the declared crop type information is matched and verified with the actual land use information of the plots in the spatiotemporal topological relationship model to obtain the land use crop verification result; Based on the boundary verification results and the land type and crop verification results, the risk points of the target land parcel in the spatiotemporal dimension and attribute dimension are identified. A comprehensive risk quantification assessment is conducted on the aforementioned risk points to generate multi-dimensional results for identifying inflated risks.

5. The method according to claim 1, characterized in that, The step of determining the risk threshold set of the target land parcel based on the multi-dimensional risk factors specifically includes: Based on the multi-dimensional risk factors, obtain the basic threshold set and the factor weight set; The multi-dimensional risk factors are quantified to obtain a standardized score for each risk factor. Based on the set of factor weights, the standardized scores are weighted and summed to obtain the comprehensive risk score of the target land parcel; Based on the comprehensive risk score, the basic threshold set is adjusted to obtain the initial risk threshold set for the target land parcel; By calling preset threshold constraint rules through a preset rule engine, the initial risk threshold set is verified to obtain the risk threshold set of the target land parcel.

6. The method according to claim 1, characterized in that, The multi-dimensional inflated risk identification result includes assessment data for multiple risk dimensions. The step of performing risk verification on the multi-dimensional inflated risk identification result based on the risk threshold set and generating differentiated warnings specifically includes: The assessment data for each risk dimension is compared with the corresponding risk threshold in the set of risk thresholds to obtain the comparison results; Based on the comparison results, the target risk dimension with risk is determined from the multiple risk dimensions, and risk warning information corresponding to the target risk dimension is generated; Based on the aforementioned risk warning information and the preset warning level rules, differentiated warning levels are determined; Based on the differentiated early warning level and the risk warning information, an early warning strategy for the target land parcel is generated for early warning purposes.

7. The method according to claim 6, characterized in that, After the step of performing risk verification on the multi-dimensional false risk identification results based on the risk threshold and generating differentiated early warnings, the method further includes: A corresponding early warning task is generated based on the early warning strategy, and the early warning task is associated with the underwriting data. For the aforementioned relationship, an operation log is recorded, which includes the operation time and operation content; Based on the operation time and operation content, a full-link data tracking record is generated, which is used to identify the processing flow of the early warning task; If the end-to-end data tracking record shows that the early warning task has not been completed, a prompt message is generated to update the operation log; The closed-loop status of the early warning task is determined based on the update status of the operation log.

8. A risk warning device, characterized in that, include: The acquisition module is used to acquire the insurance data, land attribute information, and multi-source remote sensing image data of the target land parcel. The crop identification module is used to perform remote sensing crop identification on the multi-source remote sensing image data to obtain the crop identification results of the target plot; A module is established to build a spatiotemporal topological relationship model of the target plot based on the crop identification results and the plot attribute information; The risk identification module is used to identify the risk of inflated claims based on the underwriting data and the spatiotemporal topological relationship model, and to obtain multi-dimensional inflated risk identification results. The first determining module is used to obtain multi-dimensional risk factors of the target land parcel from a preset risk factor library, and determine the risk threshold set of the target land parcel based on the multi-dimensional risk factors; The verification module is used to perform risk verification on the multi-dimensional false risk identification results based on the risk threshold set, and generate differentiated warnings.

9. A computer device, characterized in that, It includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the risk warning method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the risk warning method as described in any one of claims 1 to 7.

Images