Intelligent underwriting risk analysis method, device and equipment based on agricultural products and medium

Abstract

The invention relates to the technical field of intelligent decision making, can be applied to an agricultural financial science and technology business system platform, and discloses an intelligent underwriting risk analysis method, device and equipment based on agricultural products, and a medium, and the method comprises the steps: obtaining the multi-source behavior data of a target farmer and the multi-source environment data of a target agricultural product, performing space-time alignment and data fusion on the multi-source behavior data and the multi-source environment data to obtain multi-source synchronous data; extracting multi-source fusion features of the multi-source synchronous data, and constructing a risk analysis model according to the multi-source fusion features; performing risk quantification on the multi-source synchronous data by using the risk analysis model to obtain a target risk factor; calculating a target risk dimension probability of the target agricultural product according to the target risk factor; and performing probability fusion analysis on the target risk dimension probability to obtain a target underwriting risk control value. According to the invention, the accuracy of target peasant household underwriting risk assessment and the accuracy of underwriting risk control value pricing can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent decision-making, and in particular to a risk intelligent analysis method, device and equipment based on agricultural products and a medium. BACKGROUND

[0002] With the continuous progress of digital technology, agricultural production activities are undergoing profound changes. For target farmers, agricultural production involves many links and behaviors, covering many aspects such as agricultural material procurement, planting management, and sales channel selection. At the same time, the growth of agricultural products is also affected by many environmental factors, such as changes in meteorological conditions, soil quality, and occurrence of diseases and pests. These multi-source data from different links and factors are interwoven and associated with each other, making the analysis of agricultural insurance underwriting risk complex and difficult.

[0003] However, the current agricultural insurance industry generally relies on historical disaster statistical data and regional experience to determine the underwriting risk control value, but agricultural production factors such as meteorological conditions and diseases and pests change over time and with climate change. These changes make the disaster occurrence probability and loss degree estimated based on historical disaster statistical data and regional experience have a large deviation from the actual situation, and thus the determined underwriting risk probability is seriously out of line with the actual risk situation faced by target farmers.

[0004] In addition, traditional agricultural insurance underwriting risk assessment models are mostly limited to structured data such as meteorological conditions and yields, and lack of mining of unstructured data such as soil moisture and farmer behavior, resulting in limited data dimensions, difficulty in capturing micro risk characteristics, and thus poor accuracy of underwriting risk control value.

[0005] Therefore, how to improve the accuracy of target farmer underwriting risk assessment and improve the accuracy of underwriting risk control value pricing has become a problem to be solved. SUMMARY

[0006] The present application provides a risk intelligent analysis method, device, equipment and medium based on agricultural products, which is mainly aimed at solving the problem of low accuracy of target farmer underwriting risk assessment and low accuracy of underwriting risk control value pricing.

[0007] In the first aspect, to achieve the above-mentioned purpose, the present application provides a risk intelligent analysis method based on agricultural products, comprising: obtaining multi-source behavior data of a target farmer and multi-source environmental data of a target agricultural product, and performing spatio-temporal alignment and data fusion on the multi-source behavior data and the multi-source environmental data to obtain multi-source synchronous data; extract a multi-source fusion feature of the multi-source synchronous data, and construct a risk analysis model according to the multi-source fusion feature; quantify risks of the multi-source synchronous data by using the risk analysis model, to obtain a plurality of target risk factors; calculate a plurality of target risk dimension probabilities of the target agricultural product according to the target risk factors; perform probability fusion analysis on the plurality of target risk dimension probabilities, to obtain a target underwriting risk control value of the target peasant household.

[0008] In a second aspect, the present application further provides a device for intelligent analysis of underwriting risks based on agricultural products, comprising: a synchronous data alignment module, configured to acquire multi-source behavior data of a target peasant household and multi-source environment data of a target agricultural product, perform space-time alignment and data fusion on the multi-source behavior data and the multi-source environment data, and obtain multi-source synchronous data; a fusion feature extraction module, configured to extract a multi-source fusion feature of the multi-source synchronous data, and construct a risk analysis model according to the multi-source fusion feature; a data risk quantification module, configured to quantify risks of the multi-source synchronous data by using the risk analysis model, to obtain a plurality of target risk factors; a risk probability analysis module, configured to calculate a plurality of target risk dimension probabilities of the target agricultural product according to the target risk factors; a risk control value determination module, configured to perform probability fusion analysis on the plurality of target risk dimension probabilities, to obtain a target underwriting risk control value of the target peasant household.

[0009] In a third aspect, the present application further provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the above-mentioned intelligent analysis method of underwriting risks based on agricultural products.

[0010] In a fourth aspect, the present application further provides a computer readable storage medium, wherein the computer readable storage medium stores at least one computer program, and the at least one computer program is executed by a processor in an electronic device to implement the above-mentioned intelligent analysis method of underwriting risks based on agricultural products.

[0011] In the embodiment of the present application, by spatio-temporal alignment of multi-source behavior data and multi-source environment data, the matching problem caused by inconsistent time reference and space coordinates of different source data is solved, avoiding the tediousness and errors of manual processing, greatly improving the data processing speed; the multi-source fusion feature integrates the key information of different source data, avoiding the tediousness of processing massive heterogeneous data separately, in terms of model performance, the multi-source fusion feature covers more comprehensive and rich risk-related information, providing better input for the risk analysis model, effectively improving the accuracy and reliability of risk analysis, and reducing the probability of misjudgment and omission; using the risk analysis model to quantify the multi-source synchronous data, avoiding the inefficiency and subjectivity of manual analysis, greatly shortening the risk assessment cycle, and more accurately capturing the risk patterns and correlation in the multi-source synchronous data; at the same time, efficient quantitative analysis of the risk dimension probability of the target agricultural product is realized, the intelligent level of risk assessment is significantly improved, and the accuracy of the risk control value pricing and the reliability of the risk control are improved by combining the multi-dimensional risk probability, historical underwriting data and farmer safety behavior coefficient. BRIEF DESCRIPTION OF DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the description of the embodiments of the present application will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0013] Figure 1 An application environment schematic diagram of an agricultural product-based underwriting risk intelligent analysis method according to an embodiment of the present application; Figure 2 A flowchart of an agricultural product-based underwriting risk intelligent analysis method according to an embodiment of the present application; Figure 3 A flowchart of probability fusion analysis of a plurality of target risk dimension probabilities according to an embodiment of the present application; Figure 4 A module schematic diagram of an agricultural product-based underwriting risk intelligent analysis device according to an embodiment of the present application; Figure 5 A structure schematic diagram of an electronic device for realizing an agricultural product-based underwriting risk intelligent analysis method according to an embodiment of the present application; Figure 6 Another structure schematic diagram of an electronic device for realizing an agricultural product-based underwriting risk intelligent analysis method according to an embodiment of the present application.

[0014] The object implementation, functional characteristics and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0015] In order to better understand the technical solutions of the present disclosure by those skilled in the art, and to fully understand and implement the implementation process of the present disclosure how to apply technical means to solve technical problems and achieve the corresponding technical effects, the technical solutions in the embodiments of the present disclosure will be described clearly and completely in the following with reference to the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, not all. The embodiments of the present disclosure and various features in the embodiments can be combined with each other without conflict, and the technical solutions formed thereby are all within the protection scope of the present disclosure. Based on the embodiments in the present disclosure, all other embodiments obtained by those skilled in the art without creative labor should be within the protection scope of the present disclosure.

[0016] It should be noted that the terms "first", "second" and the like in the specification and claims of the present disclosure and the above-described drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or a chronological sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present disclosure described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, device, product or apparatus that includes a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or apparatuses.

[0017] The embodiment of the present application provides a kind of based on agricultural product's risk intelligent analysis method of insurance, the execution subject of the risk intelligent analysis method of insurance based on agricultural product of the described at least one kind of electronic device including but not limited to server, terminal etc. It can be configured to execute the device provided by the embodiment of the present application.By software or hardware installed in terminal equipment or server equipment.In other words, the risk intelligent analysis method of insurance based on agricultural product can be executed by software or hardware installed in terminal equipment or server equipment.The server includes but is not limited to: single server, server cluster, cloud server or cloud server cluster etc.The server can be independent server, can also be cloud server that provides cloud service, cloud database, cloud computing, cloud function, cloud storage, network service, cloud communication, middleware service, domain name service, security service, content distribution network (Content Delivery Network, CDN), and big data and artificial intelligence platform etc. Foundation cloud computing service.

[0018] The present application provides a kind of based on agricultural product's risk intelligent analysis method of insurance, which can be applied to Figure 1In the application environment, the client communicates with the server through the network. The server can obtain multi-source behavior data of the target farmer and multi-source environment data of the target agricultural product through the client, and solve the matching problem caused by the inconsistency of time benchmarks and space coordinates of different source data by spatiotemporal alignment of the multi-source behavior data and the multi-source environment data, thereby avoiding the tediousness and errors of manual processing and greatly improving the data processing speed. The multi-source fusion feature integrates the key information of different source data, avoids the tediousness of processing massive heterogeneous data respectively, and covers more comprehensive and rich risk-related information in terms of model performance, thereby providing better input for the risk analysis model and effectively improving the accuracy and reliability of risk analysis and reducing the probability of misjudgment and omission. Specifically, the risk analysis model is used to quantize the multi-source synchronous data, thereby avoiding the inefficiency and subjectivity of manual analysis, greatly shortening the risk assessment period, and more accurately capturing the risk patterns and correlation in the multi-source synchronous data. Meanwhile, efficient quantitative analysis of the risk dimension probability of the target agricultural product is realized, the intelligent level of risk assessment is significantly improved, the accuracy of the pricing of the risk control value of the target agricultural product is improved by combining the multi-dimensional risk probability, historical underwriting data and the safety behavior coefficient of the farmer, and the reliability of risk control is improved, and finally the target risk control value of the target agricultural product is output and fed back to the client.

[0019] The client can be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers and portable wearable devices. The server can be implemented by an independent server or a server cluster composed of multiple servers. The application will be described in detail through specific embodiments.

[0020] Referring to FIG. 1, it is a flowchart of a risk intelligent analysis method for underwriting of an agricultural product according to an embodiment of the application. In this embodiment, the risk intelligent analysis method for underwriting of an agricultural product comprises the following steps. Figure 2 S1, obtaining multi-source behavior data of a target farmer and multi-source environment data of a target agricultural product, and performing spatiotemporal alignment and data fusion on the multi-source behavior data and the multi-source environment data to obtain multi-source synchronous data.

[0021] In the embodiment of the application, the obtaining refers to collecting the multi-source behavior data of the target farmer and the multi-source environment data of the target agricultural product from multiple different data sources, and the spatiotemporal alignment refers to adding a unified time mark (time stamp) to the multi-source behavior data and the multi-source environment data of different sources and performing space conversion in the data collection and processing process, so as to ensure the consistency of the multi-source behavior data and the multi-source environment data in the time dimension and the space dimension.

[0022] ​Specifically, multi-source behavior data of target farmers and multi-source environment data of target agricultural products are collected from multiple different channels, such as equipment records, platform interactions, sensor data, etc., wherein the multi-source behavior data of target farmers refers to operation records generated by target farmers in the whole process of agricultural production, including time and space information and decision characteristics from sowing, fertilization, irrigation to harvesting, sales and other links.

[0023] For example, the fertilization time, fertilizer type, dosage and operation duration of target farmers in a certain farmland can be collected in real time through the Beidou positioning module and sensor network of intelligent agricultural equipment, and combined with the agricultural log recorded by the mobile terminal APP to form multi-source behavior data.

[0024] The multi-source environment data of target agricultural products refers to comprehensive environmental parameters affected by natural and human factors during the growth cycle of target agricultural products, including soil temperature and humidity, light intensity, meteorological disasters, occurrence of diseases and insect pests, and other environmental data.

[0025] In the embodiment of the application, the spatiotemporal alignment and data fusion of the multi-source behavior data and the multi-source environment data to obtain multi-source synchronous data comprises: The multi-source behavior data and the multi-source environment data are respectively subjected to time format standardization processing to obtain standard behavior data and standard environment data; The standard behavior data and the standard environment data are respectively subjected to timestamp alignment to obtain aligned behavior data and aligned environment data; The spatial behavior feature information in the aligned behavior data and the spatial environment identifier information in the aligned environment data are extracted; The aligned behavior data and the aligned environment data are respectively subjected to spatial conversion according to the spatial behavior feature information and the spatial environment identifier information to obtain synchronous behavior data and synchronous environment data; The synchronous behavior data and the synchronous environment data are respectively subjected to data cleaning to obtain cleaned behavior data and cleaned environment data; The cleaned behavior data and the cleaned environment data are subjected to redundancy removal and merging to obtain multi-source synchronous data.

[0026] In the embodiment of the application, the time format standardization refers to converting the timestamps in the multi-source behavior data and the multi-source environment data into a unified time format and time zone to ensure the consistency and comparability of time data; and the timestamp alignment refers to matching and integrating data records with the same or similar timestamps in the multi-source behavior data and the multi-source environment data.

[0027] The space conversion refers to the process of mapping the aligned behavior data and the aligned environment data from one space reference system to a standard space reference system, the core of which is to realize the synchronous presentation of multi-source behavior data and multi-source environment data under different space logics by adjusting the spatial coordinates, the visual angle or the dimensional relationship; the redundancy removal and merging refers to further processing the cleaning behavior data and the cleaning environment data to remove redundant information and merge similar records, and then obtaining multi-source synchronous data.

[0028] Specifically, the multi-source behavior data and the multi-source environment data are uniformly converted into an international standard time format, for example, the original record of the intelligent device “2025 / 7 / 3 8:30” is converted into “2025-07-03 08:30:00”; the low-precision timestamp (such as accurate to minutes) is supplemented to the second level (such as “09:00” is supplemented to “09:00:00”), and the high-precision timestamp (such as accurate to milliseconds) is truncated to the second level (such as “09:00:00.123” is retained as “09:00:00”), so as to ensure the consistency of the time granularity; the timestamps of different time zones are converted into the target time zone (such as unified into Beijing time), for example, the UTC time “01:00:00” (8 hours later than Beijing time) is converted into “09:00:00”.

[0029] Further, when aligning the timestamps of the standard behavior data and the standard environment data, the timestamp of the earliest standard behavior data and the standard environment data is taken as the starting point, and the timestamp of the latest standard behavior data and the standard environment data is taken as the ending point, a continuous second-level time axis is generated, each piece of standard behavior data and standard environment data is matched to the time point corresponding to the second-level time axis, so as to obtain the aligned behavior data and the aligned environment data.

[0030] In detail, the spatial behavior characteristic information of the target farmer operation (such as the latitude and longitude) and the operation area range (such as the farmland plot number) is parsed from the aligned behavior data through spatial semantic analysis, and the spatial environment identification information (such as the geographical position corresponding to the Internet of Things node) of the sensor deployment is extracted from the aligned environment data; based on the spatial behavior characteristic information and the spatial environment identification information, the spatial reference system of the aligned behavior data and the aligned environment data is unified into the WGS84 geographical coordinate system by using a coordinate conversion algorithm, so as to realize the alignment of the spatial dimension, and then obtain the synchronous behavior data and the synchronous environment data.

[0031] Specifically, the synchronous behavior data and the synchronous environment data are subjected to an outlier detection algorithm to identify and correct values that obviously deviate from the normal range (such as data whose soil moisture exceeds twice the field water holding capacity), and at the same time, the error records in the synchronous behavior data and the synchronous environment data are filtered through a logical verification rule (such as the fertilizer amount cannot be negative).

[0032] Further, in the redundancy deduplication merging link, a hash algorithm is used to generate a unique behavior identifier and a unique environment identifier for each cleaning behavior data and cleaning environment data, and by comparing the unique behavior identifier and the unique environment identifier, completely repeated data is removed to obtain deduplicated behavior data and deduplicated environment data, and the deduplicated behavior data and the deduplicated environment data are subjected to data aggregation to form multi-source synchronous data which is consistent in time and space and unified in structure.

[0033] Exemplarily, in the field of financial technology, the time-space alignment and data fusion operation of multi-source behavior data and multi-source environment data are implemented in the intelligent risk control scene, first, multi-source behavior data of a target user in agricultural insurance, such as underwriting time, underwriting location, underwriting amount, etc., and multi-source environment data, such as crop growth environment, etc., are collected, and the collected multi-source behavior data and multi-source environment data are subjected to time format standardization processing respectively, and are unified to a specific standard time format to obtain standard behavior data and standard environment data.

[0034] Specifically, according to the unified time standard, time stamp alignment is performed on the two to obtain aligned behavior data and aligned environment data, then, spatial behavior feature information, such as the geographical position range of a transaction, is extracted from the aligned behavior data, and spatial environment identifier information, such as the area where a transaction device is located, is extracted from the aligned environment data; according to the information, spatial conversion is performed on the two to obtain synchronous behavior data and synchronous environment data, then, data cleaning is performed on the synchronous data to remove noise and outliers to obtain cleaned behavior data and cleaned environment data, and finally, redundancy deduplication and merging are performed on the cleaned data to form multi-source synchronous data which can be used for intelligent risk control analysis, and help to more accurately assess financial transaction risks.

[0035] In the embodiment of the application, by performing time-space alignment on multi-source behavior data and multi-source environment data, the matching problem caused by inconsistent time benchmarks and spatial coordinates of different source data is solved, the tediousness and errors of manual processing are avoided, the data processing speed is greatly improved, and the relevance and accuracy of subsequent risk analysis are ensured.

[0036] S2, extract multi-source fusion features of the multi-source synchronous data, and construct a risk analysis model according to the multi-source fusion features.

[0037] In the embodiment of the present application, the extraction of the multi-source fusion features refers to the process of refining different sources and different types of multi-source synchronous data to generate more representative and comprehensive features reflecting the characteristics of target farmers and target agricultural products; the risk analysis model is a tool for identifying, evaluating and predicting the risks that may be faced by the agricultural production process, and by analyzing the internal relationship of the multi-source synchronous data, the corresponding risk analysis algorithm rules are established, so that the risk occurrence probability, impact degree and the like under unknown conditions can be quantitatively or qualitatively analyzed.

[0038] In the embodiment of the present application, the multi-source fusion features of the multi-source synchronous data are extracted, including: performing data type layering on the multi-source synchronous data to obtain layered data; performing preliminary feature extraction on the layered data to obtain a layered feature set; performing cross-source feature association fusion on the layered feature set to obtain an associated fusion feature set; performing dimension reduction processing on the associated fusion feature set to obtain a dimension-reduced fusion feature set; performing feature stability testing on the dimension-reduced fusion feature set, and screening out dimension-reduced fusion features with stability greater than a set threshold value as multi-source fusion features.

[0039] In the embodiment of the present application, the data type layering refers to dividing the multi-source synchronous data into multiple levels or categories according to different dimensions such as internal attributes, generation methods, forms of expression or business logic of the multi-source synchronous data; the preliminary feature extraction refers to mining key information or attributes from each layered data that can represent the characteristics of the layered data, which can be used to describe the internal rules and characteristics of the layered data; the cross-source feature association fusion refers to integrating and associating features with relevance in different layered feature sets to mine potential connections and complementary information between different data sources, and forming an associated fusion feature set.

[0040] Specifically, the multi-source synchronous data is classified into three layered data of numerical data, text data and image data according to data types, and for the numerical data, it can be further subdivided into continuous type (such as temperature, precipitation) and discrete type (such as crop variety number); the text data can be classified according to content themes, such as classifying texts describing diseases and pests into “disease and pest information layer”; the image data can be classified according to shooting objects or scenes, such as “farmland crop image layer” and “farmland facility image layer”.

[0041] Among them, for numerical data, statistical feature extraction and time series feature extraction can be based on, statistical feature extraction such as calculating mean and median and other basic statistics to describe the central tendency and dispersion degree of numerical data in hierarchical data, for example, for the numerical data of the soil moisture layer, the average humidity, humidity fluctuation range and the like in a preset period of time are calculated; the time series feature extraction can extract the trend, seasonality, periodicity and other features of the numerical data, for example, the change trend of the multi-source environmental data with time is analyzed, and the seasonal fluctuation feature is extracted.

[0042] For text data, the text can be represented as a word frequency vector, and the frequency of each word appearing in the text is counted, for example, for the text of the pest information layer, the occurrence times of keywords such as "aphid" and "powdery mildew" are counted; for image data, color feature extraction and texture feature extraction can be based on, color feature extraction calculates the color histogram of image data, and counts the distribution of different colors in image data, and based on texture feature extraction, the texture information of image data is extracted using a gray level co-occurrence matrix, wavelet transform and the like to describe the spatial relationship between pixels in image data.

[0043] For example, the proportion of green, yellow and other colors in the crop image of the farmland is analyzed to judge the growth status of the target agricultural product; the texture feature of the farmland soil image is analyzed to judge the looseness of the soil.

[0044] Further, the cross-source feature correlation fusion of the hierarchical feature set can calculate the correlation coefficient (such as cosine similarity) between different hierarchical features, find out the feature pairs or feature groups with strong correlation and perform feature fusion to generate a correlation fusion feature set.

[0045] In detail, the dimension reduction processing refers to reducing the dimension of the correlation fusion feature set, removing redundant and irrelevant features, and retaining the most important information to reduce the calculation complexity, and specifically, the correlation fusion feature set is projected into a new coordinate system through linear transformation, and several fusion features with the largest variance are selected as new features to realize dimension reduction of the features; the feature stability test is to evaluate the change degree of the dimension reduction fusion features under different data sets, different environmental conditions or different times, and to select features with high stability to improve the robustness and generalization ability of the subsequent risk analysis model.

[0046] Specifically, the correlation coefficient between the values of the dimension reduction fusion feature set in different time periods is calculated, the closer the correlation coefficient is to 1, the more stable the dimension reduction fusion feature is, and according to a pre-set threshold, the dimension reduction fusion features with stability greater than the threshold are selected as multi-source fusion features.

[0047] In the embodiment of the present application, the cross-source feature correlation fusion of the hierarchical feature set to obtain a correlation fusion feature set comprises: calculate the similarity between each hierarchical feature in the hierarchical feature set, and determine the association strength between the hierarchical features according to the similarity; construct a cross-source feature association matrix according to the association strength, and perform hierarchical alignment processing on the cross-source feature association matrix to obtain an aligned association matrix; perform weight distribution and feature fusion on the aligned association matrix by using a preset attention mechanism to obtain an association fusion feature set.

[0048] In the embodiment of the application, the Jaccard similarity can be used to calculate the similarity for the hierarchical feature set, the Jaccard similarity is defined as the ratio of the size of the intersection of each two hierarchical features to the size of the union, according to the similarity calculation result, the hierarchical feature pair with a similarity greater than a preset similarity threshold is regarded as having a strong association, and the association strength can be set as the similarity value itself; for example, if the Jaccard similarity of two hierarchical features is 0.8, the association strength of the two hierarchical features is also 0.8, provided that the similarity threshold is set to 0.7.

[0049] Specifically, the hierarchical features are taken as rows and columns, the association strength between the hierarchical feature pairs is filled into the corresponding positions of the matrix to obtain a cross-source feature association matrix, if the hierarchical features come from different hierarchical structures, hierarchical alignment needs to be performed, a tree structure-based alignment method can be used to map the hierarchical features of different levels to a unified tree structure, and then the hierarchical features are aligned according to their positions in the tree structure.

[0050] For example, for the crop growth stage features (high level) and soil nutrient features (low level) in the agricultural data, the crop growth and soil nutrient association tree is constructed to align the crop growth stage features and the soil nutrient features to the corresponding nodes, so that the correspondence relationship of the features in the cross-source feature association matrix is correct, and thus the aligned association matrix is obtained.

[0051] The self-attention mechanism (Self-Attention) is used, for each aligned feature in the aligned association matrix, the attention weight between the aligned feature and other features is calculated, the self-attention mechanism determines the attention weight by calculating the similarity between the query vector (Query), the key vector (Key) and the value vector (Value); according to the calculated attention weight, the aligned features in the aligned association matrix are weighted and summed to obtain all the fused feature vectors, which form the association fusion feature set.

[0052] In the embodiment of the application, the training data set of the risk analysis model is constructed according to the multi-source fusion features and a preset risk label vector, in the training process, a suitable optimization algorithm such as stochastic gradient descent is used to continuously adjust the parameters of the risk analysis model, so that the error between the output of the risk analysis model and the real risk label is minimized.

[0053] Optionally, the trained risk analysis model is comprehensively evaluated by accuracy, recall rate, F1 value and the like to measure the performance of the risk analysis model, and the risk analysis model is optimized and adjusted according to the evaluation results, such as adjusting the model structure, increasing or reducing features and the like, so as to finally build a model capable of accurately and reliably performing risk analysis, that is, a risk analysis model.

[0054] In the embodiment of the application, the multi-source fusion feature integrates key information of data from different sources, avoids the tediousness of separately processing massive heterogeneous data, improves the data processing speed, and in terms of model performance, the multi-source fusion feature covers more comprehensive and rich risk-related information, provides better input for the risk analysis model, effectively improves the accuracy and reliability of risk analysis, and reduces the probability of misjudgment and omission.

[0055] S3, quantifying the risk of the multi-source synchronous data by using the risk analysis model to obtain a plurality of target risk factors.

[0056] In the embodiment of the application, the target risk factor refers to various factors that have potential influence on the production, circulation, sales and the like of target agricultural products, and then may cause risk events or lead to losses. For fruit agricultural products, frost and plant diseases and insect pests are common target risk factors. For grain agricultural products, drought, flood and fluctuation of chemical fertilizer price are main target risk factors.

[0057] In the embodiment of the application, the risk quantification of the multi-source synchronous data by using the risk analysis model to obtain a plurality of target risk factors comprises: performing feature importance analysis on the multi-source synchronous data to obtain a weight coefficient of the multi-source synchronous data; filtering out a key risk data subset greater than a preset weight threshold according to the weight coefficient; calculating an initial risk score of the key risk data subset by using the risk analysis model; performing time series smoothing processing on the initial risk score to obtain a target risk score; performing factor mapping on the target risk score according to a preset risk threshold to obtain a plurality of target risk factors.

[0058] In the embodiment of the application, the random forest is used to calculate the total decrease of impurity (such as Gini coefficient) brought by each multi-source synchronous data when the decision tree node is split, so as to determine the importance degree of the multi-source synchronous data, that is, the weight coefficient. The higher the weight coefficient is, the greater the influence of the multi-source synchronous data on the risk analysis model is. All multi-source synchronous data and their weight coefficients are traversed, and the multi-source synchronous data with a weight coefficient greater than a preset weight threshold are summarized to form a key risk data subset.

[0059] In detail, a subset of key risk data is input into a pre-trained risk analysis model, which outputs a corresponding initial risk score. The initial risk score is then subjected to time-series smoothing to eliminate short-term fluctuations and noise, making it more stable and reliable. Specifically, a fixed-size window can be selected, and the average value of the initial risk scores within the window is calculated as the target risk score at the current moment.

[0060] For example, by using a 5-day moving average window, the initial risk scores for each day and the previous 4 days are added together and then divided by 5 to obtain the target risk score for that day, which can effectively smooth out short-term fluctuations.

[0061] Specifically, based on preset risk thresholds and preset risk labels, multi-source synchronous data corresponding to continuous target risk scores are converted into discrete target risk factors. Multiple preset risk thresholds are set according to business rules and risk level classification standards. For example, risk labels are divided into three levels: low, medium, and high, and a first risk threshold and a second risk threshold are set. Multi-source synchronous data with target risk scores lower than the first risk threshold are mapped to low-risk factors, multi-source synchronous data with target risk scores between the first and second risk thresholds are mapped to medium-risk factors, and multi-source synchronous data with target risk scores higher than the second risk threshold are mapped to high-risk factors.

[0062] For example, in the agricultural insurance approval scenario in the fintech field, feature importance analysis is performed on the collected multi-source synchronous data to determine the weight coefficient of each multi-source synchronous data. Based on the preset weight threshold, a subset of key risk data, such as frequent overdue consumption records and high debt credit information, is selected. Then, an initial risk score for the key risk data subset is calculated using a pre-trained risk analysis model.

[0063] Considering the fluctuations in data over time, the initial risk score is smoothed over time to obtain a more stable target risk score. Finally, based on the preset risk threshold, the target risk score is mapped to multiple target risk factors such as high risk, medium risk, and low risk to assist in agricultural insurance underwriting decisions.

[0064] In this embodiment of the invention, a risk analysis model is used to quantify the risk of multi-source synchronous data, avoiding the inefficiency and subjectivity of manual analysis, greatly shortening the risk assessment cycle, and more accurately capturing the risk patterns and correlations in multi-source synchronous data. The quantified target risk factors can be used in multiple stages such as risk warning and decision support, improving the accuracy of subsequent risk control values ​​for target farmers.

[0065] S4. Calculate the probabilities of multiple target risk dimensions of the target agricultural product based on the target risk factors.

[0066] In the embodiment of the present application, by dimension mapping and probability modeling of the target risk factor, the weight-distributed multiple target risk dimension probabilities are finally calculated, thereby improving the precision of target agricultural product insurance risk assessment.

[0067] Among them, the target risk dimension is a different perspective or level for classifying and describing the target risk factor, including natural risk, market risk, technical risk, policy risk and multiple dimensions.

[0068] In the embodiment of the present application, the multiple target risk dimension probabilities of the target agricultural product are calculated according to the target risk factor, comprising: Obtaining multiple target risk dimensions corresponding to the target agricultural product, distributing the target risk factor to the corresponding target risk dimension, and generating a dimension risk factor corresponding to each target risk dimension; Normalizing each dimension risk factor to obtain multiple standardized risk factors; Probability aggregation analysis is performed on the standardized risk factor to construct a probability density function of the target risk dimension; According to the probability density function, the cumulative distribution probability value of each target risk dimension is calculated; The cumulative distribution probability value is weighted and distributed to obtain multiple target risk dimension probabilities.

[0069] In the embodiment of the present application, according to the production, circulation, sales and other links of the target agricultural product, combined with industry experience and expert knowledge, multiple target risk dimensions are determined, for example, for fruit agricultural products, the risk dimensions can include climate risk (such as the influence of heavy rain, drought, extreme temperature, etc. on fruit growth), disease and pest risk (damage of specific diseases and pests to fruit yield and quality), market risk (market price fluctuation, supply and demand change, etc.), transportation risk (loss and delay in transportation process, etc.).

[0070] Specifically, the nature and influence range of each target risk factor are analyzed, and it is distributed to the corresponding target risk dimension, for example, if a target risk factor is "recent rain frequency", it is distributed to the climate risk dimension; if the risk factor is "the infection rate of a certain common disease and pest", it is distributed to the disease and pest risk dimension, thereby obtaining the dimension risk factor corresponding to each target risk dimension.

[0071] Further, by normalizing processing to eliminate the dimensional difference between different dimensions, the value of each dimension risk factor is linearly mapped to a fixed interval, ensuring that the dimension risk factor obeys the normal distribution, thereby obtaining multiple standardized risk factors; the maximum likelihood estimation method is used to estimate the parameters of the standardized risk factor, thereby constructing the probability density function.

[0072] Specifically, if the market risk dimension obeys a normal distribution, the probability density function of the market risk dimension can be obtained by calculating the mean and standard deviation of the standardized risk factor as the parameters of the normal distribution; for a continuous probability density function, the cumulative distribution function (CDF) can be calculated by numerical integration.

[0073] For example, the probability density function is integrated using the trapezoidal rule or Simpson's rule to obtain the cumulative distribution probability value, and the cumulative distribution probability values of each dimension are weighted to generate the final target risk dimension probability.

[0074] In the embodiment of the present application, efficient quantitative analysis of the target agricultural product risk dimension probability is realized, which significantly improves the intelligent level of risk assessment and improves the accuracy of subsequent guaranteed risk control value pricing.

[0075] S5, probability fusion analysis is performed on a plurality of target risk dimension probabilities to obtain a target guaranteed risk control value of the target farmer.

[0076] In the embodiment of the present application, three guaranteed risk control values are obtained by comprehensively analyzing the risk dimension probability, the historical guarantee factor and the safety behavior, and the final target guaranteed risk control value is obtained by fusing the three guaranteed risk control values, thereby improving the accuracy of the guaranteed risk control value pricing.

[0077] As shown in Figure 3 In the embodiment of the present application, the probability fusion analysis on a plurality of target risk dimension probabilities to obtain a target guaranteed risk control value of the target farmer comprises: controlling value analysis is performed on the target risk dimension probability by using a preset guaranteed risk control strategy to obtain a first guaranteed risk control value; obtaining a historical guarantee factor of the target agricultural product, and calculating a second guaranteed risk control value of the target farmer according to the historical guarantee factor; performing safety behavior analysis on the target farmer to obtain a safety behavior coefficient of the target farmer; determining a third guaranteed risk control value of the target farmer according to the safety behavior coefficient; weighting the first guaranteed risk control value, the second guaranteed risk control value and the third guaranteed risk control value to obtain a target guaranteed risk control value.

[0078] Specifically, the target guaranteed risk control value is specifically as follows:

[0079] wherein, represents the target guaranteed risk control value at the moment, represents a preset initial underwriting risk control value, represents a first underwriting risk control value at a time point, represents a third underwriting risk control value at a time point, represents a second underwriting risk control value at a time point.

[0080] In the embodiment of the present application, a corresponding risk control strategy is selected according to the size of the risk dimension probability, and a corresponding first risk control value is calculated; historical underwriting data of the target agricultural product is obtained, including underwriting amount, claim rate, loss condition, etc., historical underwriting factors affecting underwriting risk, such as climate conditions, disease and pest occurrence rate, etc., are extracted according to the historical underwriting data, and a second underwriting risk control value of the target farmer is calculated according to the historical underwriting factors.

[0081] Specifically, the safety behavior of the target farmer is analyzed by means of field investigation, questionnaire survey, etc., for example, whether the agricultural production operation of the farmer is standardized, whether effective disease and pest control measures are taken, etc., a safety behavior coefficient of the target farmer is given according to the safety behavior analysis result, and a third underwriting risk control value is calculated according to the safety behavior coefficient.

[0082] Further, the first underwriting risk control value, the second underwriting risk control value and the third underwriting risk control value are weighted and calculated by using weighted average to obtain a target underwriting risk control value.

[0083] Exemplarily, in the field of agricultural insurance, taking the underwriting of rice planting farmers in a certain region as an example, first, the preset underwriting risk control strategy is used to analyze the target risk dimension probability such as natural disasters, diseases and pests, etc., to obtain a first underwriting risk control value, then the historical underwriting factors of rice in this region in the past few years are obtained, such as the loss claim rate in recent years, etc., and a second underwriting risk control value of the target farmer is calculated accordingly.

[0084] Specifically, safety behavior analysis is carried out on the target farmer, such as whether irrigation and fertilization are carried out according to the standard, a safety behavior coefficient is obtained, a third underwriting risk control value is determined, finally, the weights of the control values are set according to the importance, the three underwriting risk control values are weighted and calculated, and a target underwriting risk control value of the target farmer is obtained, which provides a basis for reasonable underwriting.

[0085] In the embodiment of the present application, by combining the multi-dimensional risk probability, the historical underwriting data and the safety behavior coefficient of the farmer, the accuracy of the underwriting risk control value pricing and the reliability of the risk control are improved.

[0086] It should be understood that the size of the serial number of each step in the above embodiments does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0087] As Figure 4 shown, it is a functional module diagram of an agricultural product-based insurance risk intelligent analysis device provided by an embodiment of the present application.

[0088] In the embodiments of the present disclosure, an agricultural product-based insurance risk intelligent analysis device is provided, which corresponds one-to-one to the above-mentioned agricultural product-based insurance risk intelligent analysis method. As Figure 4 shown, the agricultural product-based insurance risk intelligent analysis device 100 can be installed in an electronic device, and according to the implemented functions, the agricultural product-based insurance risk intelligent analysis device 100 includes a synchronous data alignment module 101, a fusion feature extraction module 102, a data risk quantification module 103, a risk probability analysis module 104, and a risk control value determination module 105. The detailed description of each functional module is as follows: The synchronous data alignment module 101 is configured to obtain multi-source behavior data of a target farmer and multi-source environment data of a target agricultural product, perform spatio-temporal alignment and data fusion on the multi-source behavior data and the multi-source environment data, and obtain multi-source synchronous data. The fusion feature extraction module 102 is configured to extract multi-source fusion features of the multi-source synchronous data, and construct a risk analysis model according to the multi-source fusion features. The data risk quantification module 103 is configured to perform risk quantification on the multi-source synchronous data by using the risk analysis model, and obtain a plurality of target risk factors. The risk probability analysis module 104 is configured to calculate a plurality of target risk dimension probabilities of the target agricultural product according to the target risk factors. The risk control value determination module 105 is configured to perform probability fusion analysis on the plurality of target risk dimension probabilities, and obtain a target insurance risk control value of the target farmer.

[0089] In an embodiment, when the synchronous data alignment module 101 performs spatio-temporal alignment and data fusion on the multi-source behavior data and the multi-source environment data to obtain multi-source synchronous data, it is configured to: perform time format standardization processing on the multi-source behavior data and the multi-source environment data respectively, to obtain standard behavior data and standard environment data; perform timestamp alignment on the standard behavior data and the standard environment data respectively, to obtain aligned behavior data and aligned environment data; extract spatial behavior feature information in the alignment behavior data and spatial environment identification information in the alignment environment data; perform spatial conversion on the alignment behavior data and the alignment environment data respectively according to the spatial behavior feature information and the spatial environment identification information, to obtain synchronous behavior data and synchronous environment data; perform data cleaning on the synchronous behavior data and the synchronous environment data respectively, to obtain cleaned behavior data and cleaned environment data; perform redundancy deduplication and merging on the cleaned behavior data and the cleaned environment data, to obtain multi-source synchronous data.

[0090] In an embodiment, the fusion feature extraction module 102, when performing extraction of multi-source fusion features of the multi-source synchronous data, is configured to: perform data type layering on the multi-source synchronous data, to obtain layered data; perform preliminary feature extraction on the layered data, to obtain a layered feature set; perform cross-source feature association fusion on the layered feature set, to obtain an associated fusion feature set; perform dimension reduction processing on the associated fusion feature set, to obtain a dimension-reduced fusion feature set; perform feature stability inspection on the dimension-reduced fusion feature set, and filter out dimension-reduced fusion features with stability greater than a set threshold value as multi-source fusion features.

[0091] In an embodiment, the fusion feature extraction module 102, when performing cross-source feature association fusion on the layered feature set to obtain an associated fusion feature set, is configured to: calculate similarity between each layered feature in the layered feature set, and determine association strength between the layered features according to the similarity; construct a cross-source feature association matrix according to the association strength, and perform hierarchical alignment processing on the cross-source feature association matrix, to obtain an aligned association matrix; perform weight distribution and feature fusion on the aligned association matrix by using a preset attention mechanism, to obtain an associated fusion feature set.

[0092] In an embodiment, the data risk quantification module 103, when performing risk quantification on the multi-source synchronous data by using the risk analysis model to obtain a plurality of target risk factors, is configured to: perform feature importance analysis on the multi-source synchronous data, to obtain a weight coefficient of the multi-source synchronous data; filter out a key risk data subset greater than a preset weight threshold value according to the weight coefficient; calculate an initial risk score of the key risk data subset by using the risk analysis model; time series smoothing processing is performed on the initial risk score to obtain a target risk score; According to the preset risk threshold, the target risk score is factor mapped to obtain a plurality of target risk factors.

[0093] In an embodiment, the risk probability analysis module 104 is configured to: obtain a plurality of target risk dimensions corresponding to the target agricultural product, distribute the target risk factors to the corresponding target risk dimensions, and generate a dimension risk factor corresponding to each target risk dimension; normalizing each dimension risk factor to obtain a plurality of standardized risk factors; performing probability aggregation analysis on the standardized risk factors to construct a probability density function of the target risk dimension; According to the probability density function, the cumulative distribution probability value of each target risk dimension is calculated; weighting the cumulative distribution probability value to obtain a plurality of target risk dimension probabilities.

[0094] In an embodiment, the risk control value determination module 105 is configured to: using a preset underwriting risk control strategy to analyze the target risk dimension probability to obtain a first underwriting risk control value; obtaining a historical underwriting factor of the target agricultural product, and calculating a second underwriting risk control value of the target farmer according to the historical underwriting factor; performing safety behavior analysis on the target farmer to obtain a safety behavior coefficient of the target farmer; determining a third underwriting risk control value of the target farmer according to the safety behavior coefficient; weighting the first underwriting risk control value, the second underwriting risk control value and the third underwriting risk control value to obtain a target underwriting risk control value.

[0095] In the present application, the specific limitations of the agricultural product-based insurance risk intelligent analysis device can be referred to the limitations of the agricultural product-based insurance risk intelligent analysis method as described above, which will not be repeated here. Each module in the above-mentioned agricultural product-based insurance risk intelligent analysis device can be realized by software, hardware and their combinations in whole or in part. The above-mentioned modules can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory in the computer device in software form, so as to be called and executed by the processor to perform the operations corresponding to each module.

[0096] In one embodiment, a computer device is provided, which can be a server, and its internal structure diagram can be as shown in Figure 5 The computer device includes a processor, a memory, a network interface and a database connected by a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium, an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with the external client through the network connection. The computer program is executed by the processor to implement the functions or steps of the server side of the agricultural product-based insurance risk intelligent analysis method.

[0097] In one embodiment, a computer device is provided, which can be a client, and its internal structure diagram can be as shown in Figure 6 The computer device includes a processor, a memory, a network interface, a display screen and an input device connected by a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with the external server through the network connection. The computer program is executed by the processor to implement the functions or steps of the client side of the agricultural product-based insurance risk intelligent analysis method.

[0098] In one embodiment, a computer device is provided, which includes a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor executes the computer program to implement the following steps: Obtaining multi-source behavior data of a target farmer and multi-source environment data of a target agricultural product, spatiotemporally aligning and fusing the multi-source behavior data and the multi-source environment data to obtain multi-source synchronous data; extract a multi-source fusion feature of the multi-source synchronous data, and construct a risk analysis model according to the multi-source fusion feature; quantify risks of the multi-source synchronous data by using the risk analysis model, to obtain a plurality of target risk factors; calculate a plurality of target risk dimension probabilities of the target agricultural products according to the target risk factors; perform probability fusion analysis on the plurality of target risk dimension probabilities, to obtain a target risk control value of the target peasant household.

[0099] In several embodiments provided in the present application, it should be understood that the disclosed devices and apparatuses can be implemented in other manners. For example, the above-described system embodiments are merely illustrative. For example, the division of the modules is merely a logical function division. In actual implementation, another division manner can be adopted.

[0100] In addition, each function module in each embodiment of the present application can be integrated in a processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be realized in the form of hardware, or in the form of hardware plus software function modules.

[0101] Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting, and the scope of the present application is defined by the appended claims rather than the above description, and therefore all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present application. Any reference signs in the claims should not be regarded as limiting the claims involved.

[0102] In some embodiments of the present embodiment, a computer readable storage medium is provided, and a computer program is stored on the computer readable storage medium. The computer program is executed by a processor to implement the steps of the method described in the above embodiments.

[0103] The readable storage medium of the present application stores a computer program, and the computer program can be implemented when executed by a processor of an electronic device: obtain multi-source behavior data of a target peasant household and multi-source environment data of a target agricultural product, perform spatio-temporal alignment and data fusion on the multi-source behavior data and the multi-source environment data, to obtain multi-source synchronous data; extract a multi-source fusion feature of the multi-source synchronous data, and construct a risk analysis model according to the multi-source fusion feature; quantify risks of the multi-source synchronous data by using the risk analysis model, to obtain a plurality of target risk factors; calculate a plurality of target risk dimension probabilities of the target agricultural products according to the target risk factors; The target risk control value of the target farmer is obtained by performing probability fusion analysis on the plurality of target risk dimension probabilities.

[0104] It should be noted that the functions or steps described above in relation to the computer-readable storage medium or the computer device can correspond to the relevant descriptions of the server side and the client side in the foregoing method embodiments, and for the sake of brevity, will not be described again here.

[0105] The computer-readable storage medium can also store at least one computer-executable program / instruction, such as computer-readable instructions. The computer-readable storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. The volatile memory may, for example, include random access memory (RAM) and / or cache memory, etc. The computer-readable storage medium may, for example, include read-only memory (ROM), a hard disk, a flash memory, etc. For example, the non-transitory computer-readable storage medium can be connected to a computing device such as a computer, and then when the computing device runs the computer-readable instructions stored on the computer-readable storage medium, the various methods described above can be performed.

[0106] In addition, the computer device can also include (but is not limited to) a data bus, an input / output (I / O) bus, a display, and an input / output device (for example, a keyboard, a mouse, a speaker, etc.), etc.

[0107] In one embodiment, the at least one computer-executable instruction can also be compiled or composed into a software product / computer program product, wherein one or more computer-executable instructions are executed by a processor to perform the steps of the various functions and / or methods described in the embodiments of the present technology.

[0108] A person of ordinary skill in the art can understand that all or part of the processes in the above-described embodiments can be completed by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium and can include the processes of the above-described embodiments when executed.

[0109] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is exemplified, and in actual applications, the above-mentioned functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above.

[0110] In the embodiments provided by the present disclosure, it should be understood that the disclosed apparatus and method can also be implemented in other manners. The embodiments described above are merely exemplary, and the apparatus embodiments described above can be implemented in other manners. For example, the flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architectures, functions and operation of the apparatus, method and computer program product according to the embodiments of the present disclosure. In this regard, each block in the flowcharts or block diagrams can represent a module, a program segment or a part of code, which contains one or more executable instructions for implementing the specified logic function. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur in a different order than that noted in the accompanying drawings. It should also be noted that each block in the block diagrams and / or flowcharts, and the combination of blocks in the block diagrams and / or flowcharts, can be implemented by a dedicated hardware-based system, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0111] The above-described embodiments are merely intended for describing the technical solutions of the present disclosure, but not for limiting the same; even though the present disclosure has been described in detail with reference to the foregoing embodiments, it should be understood by those skilled in the art that the technical solutions recorded in the foregoing embodiments can be modified, or some of the technical features can be replaced by equivalent technical features; and such modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure, and should be included in the protection scope of the present disclosure.

[0112] It should be noted that if a software tool or component that is not from the company appears in the embodiments of the present disclosure, it is merely used for example introduction, and does not represent actual use.

Claims

1. An agricultural product-based insurance risk intelligent analysis method, characterized in that, The method comprises: obtaining multi-source behavior data of a target farmer and multi-source environment data of a target agricultural product, performing spatio-temporal alignment and data fusion on the multi-source behavior data and the multi-source environment data to obtain multi-source synchronous data; extracting multi-source fusion features of the multi-source synchronous data, and constructing a risk analysis model according to the multi-source fusion features; quantifying risks of the multi-source synchronous data by using the risk analysis model to obtain a plurality of target risk factors; calculating a plurality of target risk dimension probabilities of the target agricultural product according to the target risk factors; performing probability fusion analysis on the plurality of target risk dimension probabilities to obtain a target underwriting risk control value of the target farmer.

2. The agro-based insured risk intelligent analysis method according to claim 1, wherein, The spatio-temporal alignment and data fusion on the multi-source behavior data and the multi-source environment data to obtain multi-source synchronous data comprises: performing time format standardization processing on the multi-source behavior data and the multi-source environment data respectively to obtain standard behavior data and standard environment data; performing timestamp alignment on the standard behavior data and the standard environment data respectively to obtain aligned behavior data and aligned environment data; extracting spatial behavior feature information in the aligned behavior data and spatial environment identifier information in the aligned environment data; performing spatial conversion on the aligned behavior data and the aligned environment data respectively according to the spatial behavior feature information and the spatial environment identifier information to obtain synchronous behavior data and synchronous environment data; performing data cleaning on the synchronous behavior data and the synchronous environment data respectively to obtain cleaned behavior data and cleaned environment data; performing redundancy removal and merging on the cleaned behavior data and the cleaned environment data to obtain multi-source synchronous data.

3. The agricultural product-based underwriting risk intelligent analysis method of claim 1, wherein, The extraction of the multi-source fusion features of the multi-source synchronous data comprises: performing data type layering on the multi-source synchronous data to obtain layered data; performing preliminary feature extraction on the layered data to obtain a layered feature set; performing cross-source feature association fusion on the layered feature set to obtain an associated fusion feature set; performing dimension reduction processing on the associated fusion feature set to obtain a reduced fusion feature set; performing feature stability testing on the reduced fusion feature set, and screening out reduced fusion features with stability greater than a set threshold value as multi-source fusion features.

4. The agro-based insured risk intelligent analysis method according to claim 3, wherein, The cross-source feature association fusion on the layered feature set to obtain an associated fusion feature set comprises: calculating the similarity between each layered feature in the layered feature set, and determining the association strength between the layered features according to the similarity; constructing a cross-source feature association matrix according to the association strength, and performing hierarchical alignment processing on the cross-source feature association matrix to obtain an aligned association matrix; performing weight distribution and feature fusion on the aligned association matrix by using a preset attention mechanism to obtain an associated fusion feature set.

5. The agricultural product-based underwriting risk intelligent analysis method of claim 1, wherein, The risk quantification of the multi-source synchronous data by using the risk analysis model to obtain a plurality of target risk factors comprises: performing feature importance analysis on the multi-source synchronous data to obtain a weight coefficient of the multi-source synchronous data; screening out a key risk data subset greater than a preset weight threshold value according to the weight coefficient; calculating an initial risk score of the key risk data subset by using the risk analysis model; performing time series smoothing on the initial risk score to obtain a target risk score; performing factor mapping on the target risk score according to a preset risk threshold to obtain a plurality of target risk factors.

6. The agro-based underwriting risk intelligent analysis method according to claim 1, wherein, The calculating of the plurality of target risk dimension probabilities of the target agricultural product according to the target risk factors comprises: obtaining a plurality of target risk dimensions corresponding to the target agricultural product, distributing the target risk factors to the corresponding target risk dimensions, and generating a dimension risk factor corresponding to each target risk dimension; performing normalization processing on each dimension risk factor to obtain a plurality of standardized risk factors; performing probability aggregation analysis on the standardized risk factors to construct a probability density function of the target risk dimension; calculating a cumulative distribution probability value of each target risk dimension according to the probability density function; performing weight distribution on the cumulative distribution probability value to obtain a plurality of target risk dimension probabilities.

7. The agro-based underwriting risk intelligent analysis method according to claim 1, wherein, The probability fusion analysis of the plurality of target risk dimension probabilities comprises: performing control value analysis on the target risk dimension probabilities by using a preset underwriting risk control strategy to obtain a first underwriting risk control value; obtaining a historical underwriting factor of the target agricultural product, and calculating a second underwriting risk control value of the target farmer according to the historical underwriting factor; performing safety behavior analysis on the target farmer to obtain a safety behavior coefficient of the target farmer; determining a third underwriting risk control value of the target farmer according to the safety behavior coefficient; performing weighted underwriting risk control on the first underwriting risk control value, the second underwriting risk control value, and the third underwriting risk control value to obtain a target underwriting risk control value.

8. An agricultural product-based risk insurance underwriting intelligent analysis device, characterized by, The device comprises: a synchronous data alignment module configured to obtain multi-source behavior data of a target farmer and multi-source environment data of a target agricultural product, perform space-time alignment and data fusion on the multi-source behavior data and the multi-source environment data, and obtain multi-source synchronous data; a fusion feature extraction module configured to extract multi-source fusion features of the multi-source synchronous data, and construct a risk analysis model according to the multi-source fusion features; a data risk quantification module configured to perform risk quantification on the multi-source synchronous data by using the risk analysis model to obtain a plurality of target risk factors; a risk probability analysis module configured to calculate a plurality of target risk dimension probabilities of the target agricultural product according to the target risk factors; a risk control value determination module configured to perform probability fusion analysis on the plurality of target risk dimension probabilities to obtain a target underwriting risk control value of the target farmer.

9. An electronic device, comprising: The electronic device comprises: at least one processor; and a memory connected with the at least one processor in communication; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the underwriting risk intelligent analysis method based on agricultural products according to any one of claims 1 to 7.

10. A computer readable storage medium storing a computer program, characterized in that, The computer program, when executed by a processor, implements the method for intelligent analysis of underwriting risks based on agricultural products according to any one of claims 1 to 7.

Images