Resource allocation methods, devices, equipment and storage media
By using multimodal data analysis and dynamic adjustment algorithms, the problem of inaccurate car insurance pricing has been solved, achieving a match between vehicle pricing and risk and value, and ensuring the fairness and accuracy of resource allocation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA PING AN PROPERTY INSURANCE CO LTD
- Filing Date
- 2026-03-09
- Publication Date
- 2026-06-02
Smart Images
Figure CN122134035A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application belongs to the technical field of artificial intelligence, and is applied to the field of financial technology, and particularly relates to a resource allocation method and device, equipment and a storage medium. BACKGROUND
[0002] In the current society, as people's risk awareness improves, more and more people begin to understand and purchase insurance. In recent years, with the rise of new energy vehicles, the vehicle market has ushered in a new wave of sales, and the sales volume of vehicle insurance has also shown explosive growth.
[0003] However, the current pricing of vehicle insurance adopts a relatively extensive pricing method, such as directly pricing according to the selling price of the vehicle. However, the value of the vehicle is affected by various factors, not just the selling price of the vehicle. For example, different vehicle owners will decorate their vehicles differently, some vehicles are more expensive, and some vehicles are less expensive. At this time, if the insurance of these vehicles is priced the same, it will obviously make the vehicle owner whose vehicle is less expensive get more compensation, and the vehicle owner whose vehicle is more expensive get less compensation.
[0004] To sum up, the current pricing of vehicle insurance is too single in considering the influencing factors of vehicle insurance pricing, and the relevant influencing data of vehicle insurance pricing is not fully tapped, resulting in inaccurate current vehicle insurance pricing, i.e. inaccurate current resource allocation for vehicle insurance. SUMMARY
[0005] The present application provides a resource allocation method, device, equipment and storage medium, which can solve the technical problem of inaccurate resource allocation for vehicle insurance in the prior art.
[0006] In a first aspect, the present application provides a resource allocation method, comprising: obtaining multi-modal state data of a target object, an external risk graph faced by the target object, and configuration reference data corresponding to the target object, the multi-modal state data at least including environment residence point data and internal feature data; inputting the environment residence point data and the external risk graph into a first prediction model for spatio-temporal risk analysis, to generate an environment risk level and a risk prediction probability of the target object; inputting the internal feature data and the configuration reference data into a second evaluation model for value analysis, to generate an internal value coefficient of the target object; when the risk prediction probability meets a preset threshold condition, inputting the environment risk level and the internal value coefficient into a preset dynamic adjustment algorithm to calculate and generate a dynamic resource allocation parameter for the target object; wherein the dynamic resource allocation parameter is used to guide the resource allocation decision for the target object.
[0007] In a second aspect, the present application provides a resource allocation device, comprising: a data acquisition module configured to acquire multi-modal state data of a target object, an external risk map faced by the target object, and configuration reference data corresponding to the target object, wherein the multi-modal state data comprises at least the environmental residence point data and the internal feature data; a first analysis module configured to input the environmental residence point data and the external risk map into a first prediction model to perform spatio-temporal risk analysis, and generate an environmental risk level and a risk prediction probability of the target object; a second analysis module configured to input the internal feature data and the configuration reference data into a second evaluation model to perform value analysis, and generate an internal value coefficient of the target object; a resource allocation module configured to input the environmental risk level and the internal value coefficient into a preset dynamic adjustment algorithm to calculate and generate a dynamic resource allocation parameter for the target object when the risk prediction probability meets a preset threshold condition, wherein the dynamic resource allocation parameter is used to guide a resource allocation decision for the target object.
[0008] In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above resource allocation method when executing the computer program.
[0009] In a fourth aspect, a computer readable storage medium is provided, which stores a computer program, wherein the computer program is executable by a processor to implement the steps of the above resource allocation method.
[0010] The traditional pricing only relies on the vehicle selling price, and the data dimension is extremely single; the three types of core data acquired by the present application scheme lay a foundation for fine pricing; the multi-modal state data of the present application scheme covers the environmental residence point data and the internal feature data, directly capturing two key information ignored by the traditional pricing, i.e., the use environment and the added value; the external risk map integrates the risk data such as theft, robbery and collision in the area where the vehicle is frequently parked, and quantifies the environmental risk difference; the configuration reference data interfaces the configuration information of the host factory and the industry basic premium standard, and ensures the accuracy of the basic value evaluation.
[0011] In addition, the traditional pricing ignores the risk difference of the vehicle use environment; the embodiment scheme realizes accurate quantification of the risk through a first prediction model, the model input combines vehicle residence point data with an external risk atlas; the model output generates an environmental risk level and a risk prediction probability, converting intangible environmental risk into quantifiable pricing basis; realizing reasonable upward adjustment of the vehicle pricing in a high-risk environment, more favorable vehicle pricing in a low-risk environment, avoiding the problem of loss for the insurance company caused by low-price insurance of high-risk vehicles, and matching the pricing with the actual risk.
[0012] In addition, the traditional pricing ignores the risk difference of the vehicle use environment; the embodiment scheme realizes accurate quantification of the risk through a first prediction model, the model input combines vehicle residence point data with an external risk atlas; the model output generates an environmental risk level and a risk prediction probability, converting intangible environmental risk into quantifiable pricing basis; realizing reasonable upward adjustment of the vehicle pricing in a high-risk environment, more favorable vehicle pricing in a low-risk environment, avoiding the problem of loss for the insurance company caused by low-price insurance of high-risk vehicles, and matching the pricing with the actual risk.
[0013] Finally, the traditional pricing is a static mode of fixed selling price corresponding to fixed premium; the embodiment scheme realizes multi-dimensional fusion pricing of the basic value, additional value and environmental risk through a dynamic adjustment algorithm, replaces the traditional single selling price pricing, and ensures that various factors affecting the vehicle insurance are fully considered when the vehicle insurance is priced, so that the final resource allocation of the vehicle insurance is accurate enough. BRIEF DESCRIPTION OF DRAWINGS
[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the description of the embodiments of the present application. 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.
[0015] Figure 1 is a flowchart of a resource allocation method in an embodiment of the present application.
[0016] Figure 2 is Figure 1 is a flowchart of step S110 in the embodiment.
[0017] Figure 3 is Figure 2 is a flowchart of step S115 in the embodiment.
[0018] Figure 4 is Figure 1 is a flowchart of step S120 in the embodiment.
[0019] Figure 5 yes Figure 1 A flowchart of step S150.
[0020] Figure 6 This is a schematic diagram of a resource allocation device in one embodiment of the present invention.
[0021] Figure 7 This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention.
[0022] Figure 8 This is another structural schematic diagram of a computer device according to one embodiment of the present invention. Detailed Implementation
[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0024] Figure 1 A flowchart of the resource allocation method provided in the embodiments of the present invention is shown below. Figure 1 As shown, the resource allocation method provided in this embodiment of the invention includes the following steps.
[0025] Step S110: Obtain multimodal state data of the target object, external risk map of the target object, and configuration baseline data corresponding to the target object. The multimodal state data includes at least the environmental dwell point data and internal feature data.
[0026] Specifically, the core objective of this step is to collect comprehensive data needed for pricing vehicle theft insurance, providing a foundation for risk assessment and value analysis.
[0027] Specifically, the environmental location data includes the historical location sequence of vehicles (such as residential areas, commercial areas, urban villages, roadsides, etc.) obtained through vehicle navigation and GPS positioning, as well as the location time pattern (nighttime location duration from 22:00 to 5:00 and weekly average location frequency). Abnormal data during holidays is removed, and high-frequency daily scene data is retained. Internal feature data includes interior photos uploaded by car owners (such as those covering seats, center console, and dashboard) and images of high-value items inside the vehicle (electronic devices, luxury goods). Sensitive information such as faces and license plates can be masked through OpenCV template matching. Data such as vehicle type (new energy or fuel) and years of use can be obtained by connecting to the vehicle system.
[0028] External risk maps can be constructed based on heterogeneous data sources such as security data platforms to create a theft and robbery risk map of areas where vehicles are frequently parked.
[0029] The configuration baseline data can be connected to the OEM's Application Programming Interface (API) to obtain the vehicle's basic configuration (such as whether it has a smart cockpit), model guide price, and industry-standard basic insurance premium (such as the baseline insurance premium B0 = 300 yuan for popular models priced between 150,000 and 200,000 yuan).
[0030] In some embodiments of the present invention, such as Figure 2 As shown, obtaining the external risk map faced by the target object includes: Step S111: Periodically acquire historical risk event data within the target geographic area from multiple heterogeneous data sources.
[0031] Specifically, in this step, theft and robbery case data for the target geographical area (within 3km of the vehicle's usual parking location) can be obtained periodically (every hour) from the regional security data platform. This includes the time of the incident, latitude and longitude coordinates, and event description.
[0032] Step S112: Perform spatiotemporal alignment and standardization processing on the acquired historical risk event data; Specifically, this step converts theft and robbery case data from different data sources into a unified coordinate system (WGS-84), based on 100m... Geographic regions are divided into 100m grids, and the grid unit to which each case belongs is marked; duplicate and false case data are removed to ensure data accuracy.
[0033] Step S113: The kernel density estimation algorithm is used to calculate the spatial distribution density of the historical risk event data in each grid cell of the target geographic area. Specifically, this step can use a kernel density estimation algorithm to calculate the spatial distribution density of theft cases within each grid cell. The higher the density value, the higher the risk of theft in that area (e.g., the grid density in urban villages can reach 0.8, and the grid density in underground parking garages in commercial areas is 0.1).
[0034] Step S114: The processed historical risk event data is calculated using a time decay function to obtain the weight of the historical risk event data.
[0035] The weight of the historical risk event data decays more rapidly over time; Specifically, this step can use a time decay function (such as an exponential decay model) to calculate the case weight, with higher weights closer to the current time.
[0036] Step S115: Associate the weights of the historical risk event data and the spatial distribution density of the historical risk event data with the corresponding grid cells to form a real-time risk density grid.
[0037] In some embodiments of the present invention, such as Figure 3 As shown, step S115 includes the following steps.
[0038] Step S1151: Establish a unified event severity quantification scoring model for historical risk event data from different data sources, and generate a severity score for each historical risk event data.
[0039] Specifically, in this step, a unified scoring model can be established to assign severity scores based on the amount of loss and frequency of theft cases (e.g., 5 points for the theft of an entire vehicle and 2 points for the theft of items inside the vehicle).
[0040] Step S1152: Assign different confidence weights to the historical risk event data from different sources based on the data update frequency and authority level of each data source.
[0041] The data update frequency is directly proportional to the confidence weight, and the authority level is directly proportional to the confidence weight.
[0042] Specifically, in this step, confidence weights can be configured according to the data source update frequency and authority level, with the data update frequency, authority level, and confidence weights being positively correlated.
[0043] Step S1153: For the historical risk event data in each grid cell, the historical risk event data is weighted based on the confidence weight corresponding to each historical risk event data to obtain the real-time risk corresponding to each grid cell.
[0044] Specifically, in this step, the case data within each grid cell can be weighted according to confidence level. Severity score The time decay weights are used to perform a weighted summation to obtain the real-time risk value of the grid.
[0045] Step S1154: Associate the real-time risk, the weight of the historical risk event data, and the spatial distribution density of the historical risk event data with the corresponding grid cell to form the real-time risk density grid corresponding to the grid cell.
[0046] Specifically, in this step, the real-time risk value, spatial distribution density, and time decay weight of the grid can be associated with the corresponding grid cell to form a real-time risk density grid.
[0047] Historical risk events are quantitatively scored based on severity, and reliability weights are assigned according to the update frequency and authority of the data source. The real-time risk of the grid is then calculated using a weighted average to form a real-time risk density grid.
[0048] Understandably, this embodiment addresses the pain points of inconsistent data quality and distorted assessment results in traditional risk assessment, achieving objectivity and reliability in risk assessment. Traditional risk assessment treats risk data from different data sources equally, failing to differentiate between event severity and data credibility; this embodiment filters out low-quality data through a unified scoring model and confidence level weighting.
[0049] Step S116: Combine the real-time risk density grids corresponding to all grid cells in the target geographic area to form the external risk map.
[0050] Specifically, in this step, the real-time risk density grid of all grid units in the target geographic area can be integrated and divided into three levels according to risk value (low risk: not more than 0.3, medium risk: 0.3-0.6, high risk: greater than 0.6) to generate a visualized theft risk map, which is dynamically updated every hour.
[0051] Understandably, the above embodiments overcome the limitations of traditional separate input of risk data and behavioral data, improving the accuracy and real-time performance of risk prediction. Traditional models only use risk areas as a single label input, without deep integration with vehicle dwell point data, which can easily lead to deviations such as pricing vehicles parked at the edge of the grid as high-risk. This embodiment accurately captures the correlation strength between vehicles and risk areas through feature-level fusion (such as weighted correlation between dwell point coordinates and grid risk values).
[0052] Step S120: Input the environmental dwelling point data and the external risk map into the first prediction model for spatiotemporal risk analysis to generate the environmental risk level and risk prediction probability of the target object.
[0053] In some embodiments of the present invention, such as Figure 4 As shown, step S120 includes the following steps.
[0054] Step S121: Extract the historical location sequence and dwell time pattern of the target object from the environmental dwell point data.
[0055] Specifically, in this step, the high-frequency vehicle dwelling location sequence (such as residential area (18:00-8:00 the next day) - commercial area (9:00-18:00 the next day)) and dwelling time pattern (such as nighttime dwelling time averaging 3 hours per week, and frequently stopping in high-risk grids) can be extracted from the environmental dwelling point data.
[0056] Step S122: Obtain an external risk map associated with the historical residence location sequence, wherein the external risk map is constructed based on the spatiotemporal distribution of historical risk events within a geographic grid.
[0057] Specifically, in this step, the location sequence of the outposts can be mapped to the grid cells of the external risk map to obtain the risk level and real-time risk density corresponding to each outpost.
[0058] Step S123: Input the historical location sequence of the dwelling point, the dwelling time pattern and the external risk map features into the sequence prediction model built based on LSTM, and output the risk prediction probability of the target object in the future preset time period and the environmental risk level of the target object's location.
[0059] Specifically, in this step, the historical location sequence of the vehicle's dwell time, dwell time pattern, and risk map features (grid risk level, density value) can be input into the sequence prediction model built based on LSTM. The model outputs the predicted probability of theft of the vehicle in the next 30 days (e.g., the predicted probability of a vehicle frequently parked in a high-risk area is P=0.85) and the environmental risk level (high / medium / low).
[0060] Understandably, the above embodiment overcomes the pain point of traditional pricing based solely on fixed regional risks and ignoring dynamic user behavior, achieving precise quantification of three-dimensional risks: behavior, space, and time. Traditional theft insurance pricing only roughly divides regional risks (such as city / suburbs), without considering the impact of behavioral characteristics such as the type of point of interest (POI) where the vehicle is frequently parked (commercial area / urban village) and the duration of nighttime stays on risk. This embodiment captures temporal patterns (such as long-term nighttime parking in high-risk areas) through a Long Short-Term Memory (LSTM) network model, combined with spatial risk maps, to accurately predict the probability of future theft.
[0061] In some embodiments of the present invention, step S120 includes: The real-time risk density grid is input into the first prediction model, and the real-time risk density network is used as a dynamic input feature layer of the first prediction model to perform feature-level fusion with the environmental dwell point data to generate the environmental risk level and risk prediction probability.
[0062] Specifically, in this embodiment, the real-time risk density grid is used as a dynamic input feature layer and fused with environmental dwell point data at the feature level (such as the weighted correlation between dwell point coordinates and grid risk values) to improve the risk prediction accuracy of the LSTM model.
[0063] Step S130: Input the internal feature data and the configuration benchmark data into the second evaluation model for value analysis to generate the internal value coefficient of the target object.
[0064] In some embodiments of the present invention, such as Figure 5 As shown, step S130 includes the following steps.
[0065] Step S131: Preprocess the image data in the internal feature data.
[0066] Specifically, in this step, the images of the interior and high-value items uploaded by the car owner can be normalized in size (to a uniform 800 pixels). (600 pixels) and grayscale processing are used to remove image noise and improve recognition accuracy.
[0067] Step S132: Input the processed image data into the image recognition model built on CNN to identify the internal component categories of the target object and the state attributes of the internal component categories.
[0068] Specifically, the preprocessed images can be input into an image recognition model built on a convolutional neural network (CNN) to identify the categories of internal components (such as leather seats, smart cockpits, and ordinary fabric seats) and their status attributes (such as newness and wear level); and to distinguish between battery components of new energy vehicles and engine components of fuel vehicles.
[0069] Step S133: Based on the internal component category, state attributes, and the baseline configuration information of the target object obtained from the configuration baseline data source, calculate the internal value coefficient of the target object using a value decay algorithm.
[0070] The internal value coefficient is proportional to the value of the category weight factor of the internal component category set in the value decay algorithm, and the internal value coefficient is proportional to the value of the time depreciation factor determined based on the state attribute.
[0071] Specifically, in this step, the internal value coefficient = the base value of all internal components. Category weight factor The total value of the time depreciation factor is the sum of the values of the components, where the basic value of the components can be referenced from the price of the OEM's optional packages (e.g., basic value of intelligent cockpit = 15,000 yuan, basic value of ordinary central control = 3,000 yuan); the category weight factor is set according to the component value (intelligent cockpit 0.2, valuable items in the vehicle 0.3, intelligent control of doors and windows 0.2); the time depreciation factor can be calculated according to the vehicle's service life, with the formula (1-0.15t) (t is the service life), and the depreciation factor for battery components of new energy vehicles is set separately (1-0.2t).
[0072] As a specific example, for a 3-year-old new energy vehicle with a smart cockpit (category weight 0.2) and valuable items inside the vehicle (category weight 0.3), the internal value coefficient is calculated to be 0.45.
[0073] Understandably, the above embodiments address the core pain points of traditional interior value reliance on post-declaration and a single pricing dimension, achieving automated, accurate, and dynamic assessment of internal value. Traditional pricing does not differentiate between the interiors of new energy vehicles and gasoline vehicles, nor can it verify high-value items inside the vehicle, simply pricing based on the vehicle's manufacturer's suggested retail price. This embodiment automatically extracts interior configurations through CNN image recognition, combined with a value decay algorithm, to accurately calculate the internal value coefficient.
[0074] Step S140: When the risk prediction probability meets the preset threshold condition, the environmental risk level and the internal value coefficient are input into the preset dynamic adjustment algorithm to calculate and generate dynamic resource allocation parameters for the target object.
[0075] The dynamic resource allocation parameters are used to guide resource allocation decisions for the target object.
[0076] Understandably, traditional pricing relies solely on the vehicle's selling price, resulting in an extremely limited data dimension. This invention's solution lays the foundation for refined pricing by acquiring three types of core data. The multimodal state data in this invention covers environmental location data and internal characteristic data, directly capturing two key pieces of information neglected by traditional pricing: the usage environment and added value. The external risk map integrates risk data such as theft and collisions in areas where vehicles are frequently parked, quantifying differences in environmental risks. The configuration benchmark data is aligned with OEM configuration information and industry-standard premiums to ensure the accuracy of basic value assessment.
[0077] Furthermore, traditional pricing ignores the risk differences in vehicle usage environments. This embodiment achieves precise risk quantification through a first prediction model. The model input combines vehicle dwell point data with external risk maps. The model output generates environmental risk levels and risk prediction probabilities, transforming intangible environmental risks into quantifiable pricing criteria. This enables reasonable upward pricing for vehicles in high-risk environments and more favorable pricing for vehicles in low-risk environments, avoiding the problem of insurance companies losing money due to low-price insurance for high-risk vehicles, and ensuring that pricing matches actual risk.
[0078] Furthermore, traditional pricing only considers the vehicle's selling price, ignoring the added value differences brought by the owner's accessories and optional configurations. The solution of this invention achieves accurate quantification of added value through a second evaluation model. The model analyzes internal characteristic data and configuration benchmarks through input, and generates internal value coefficients through output, transforming intangible added value into a quantifiable pricing basis. This allows for a reasonable price increase for vehicles with high added value, ensuring that subsequent compensation can cover the actual value. Vehicles with low added value do not need to bear additional premiums, avoiding the situation where low-priced vehicles pay for high-priced vehicles, and ensuring that the pricing matches the actual value of the vehicle.
[0079] Finally, traditional pricing is a static model where a fixed selling price corresponds to a fixed premium. The solution of this invention achieves multi-dimensional integrated pricing of basic value, added value, and environmental risk through dynamic adjustment algorithms, replacing the traditional single selling price pricing. This ensures that when pricing car insurance, various factors affecting car insurance are fully considered, thereby ensuring sufficient resource allocation for car insurance in the end.
[0080] In some embodiments of the present invention, the dynamic adjustment algorithm is as follows: in, Parameters for allocating the dynamic resources, These are the baseline parameters determined based on the baseline attributes of the target object. The internal value coefficient is... The risk adjustment factor is determined based on the environmental risk level, and the risk adjustment factor has a positive correlation with the environmental risk level.
[0081] As a specific example, the preset risk prediction probability threshold can be 0.8. When the prediction probability is greater than or equal to 0.8, dynamic premium calculation is triggered; if the prediction probability is less than 0.8, the basic premium can be used for pricing. As a specific example, for a 3-year-old new energy vehicle, the basic insurance premium B = 300 yuan. =0.45, high environmental risk level ( =1.5), then the dynamic premium =300 (1+0.45) 1.5 = 652.5 yuan / year.
[0082] Understandably, the above embodiment breaks away from the traditional pricing model of fixed compensation ratios and disconnect between risk and value, achieving personalized pricing driven by both risk and value. Traditional theft insurance uses a uniform compensation ratio regardless of the luxury of the interior or the level of regional risk, resulting in insufficient compensation for vehicles with luxurious interiors and excessively high premiums for low-risk users. This embodiment uses an algorithm to directly link the vehicle's basic value, the added value of the interior, and the risk level of the scenario, forming a dynamic premium. This covers the risks associated with high-value interiors while ensuring that low-risk users pay lower premiums than those for high-risk scenarios, balancing the insurance company's costs with user fairness.
[0083] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention. Software tools or components not belonging to this company that appear in the embodiments of this application are merely illustrative examples and do not represent actual use.
[0084] In one embodiment, a resource allocation device is provided, which corresponds one-to-one with the resource allocation method in the above embodiments. For example... Figure 6 As shown, the processing device includes a data acquisition module 610, a first analysis module 620, a second analysis module 630, and a resource allocation module 640. Detailed descriptions of each functional module are as follows: The data acquisition module 610 is used to acquire multimodal state data of the target object, an external risk map faced by the target object, and configuration baseline data corresponding to the target object. The multimodal state data includes at least the environmental dwell point data and internal feature data. The first analysis module 620 is used to input the environmental dwelling point data and the external risk map into the first prediction model to perform spatiotemporal risk analysis and generate the environmental risk level and risk prediction probability of the target object. The second analysis module 630 is used to input the internal feature data and the configuration benchmark data into the second evaluation model for value analysis and generate the internal value coefficient of the target object. The resource allocation module 640 is used to input the environmental risk level and the internal value coefficient into a preset dynamic adjustment algorithm when the risk prediction probability meets a preset threshold condition, and calculate and generate dynamic resource allocation parameters for the target object; wherein, the dynamic resource allocation parameters are used to guide resource allocation decisions for the target object.
[0085] In one embodiment, the first analysis module 620 is specifically used for: Extract the historical location sequence and dwell time pattern of the target object from the environmental dwell point data; Obtain an external risk map associated with the historical location sequence of the outposts, wherein the external risk map is constructed based on the spatiotemporal distribution of historical risk events within a geographic grid; The historical location sequence of the dwelling point, the dwelling time pattern, and the external risk map features are input into a sequence prediction model based on LSTM, and the output is the risk prediction probability of the target object in the future preset time period and the environmental risk level of the target object's location.
[0086] In one embodiment, the second analysis module 630 is specifically used for: The image data in the internal feature data is preprocessed; The processed image data is input into an image recognition model built on CNN to identify the internal component categories of the target object and the state attributes of the internal component categories. Based on the internal component categories, state attributes, and baseline configuration information of the target object obtained from the configuration baseline data source, the internal value coefficient of the target object is calculated using a value decay algorithm. The internal value coefficient is proportional to the value of the category weight factor of the internal component category set in the value decay algorithm, and the internal value coefficient is proportional to the value of the time depreciation factor determined based on the state attributes.
[0087] In one embodiment, the dynamic adjustment algorithm is: in, Parameters for allocating the dynamic resources, These are the baseline parameters determined based on the baseline attributes of the target object. The internal value coefficient is... The risk adjustment factor is determined based on the environmental risk level, and the risk adjustment factor has a positive correlation with the environmental risk level.
[0088] In one embodiment, the data acquisition module 610 is further configured to: Historical risk event data within a target geographic area is periodically acquired from multiple heterogeneous data sources; Perform spatiotemporal alignment and standardization on the acquired historical risk event data; The kernel density estimation algorithm is used to calculate the spatial distribution density of historical risk event data in each grid cell of the target geographic area. The processed historical risk event data is calculated using a time decay function to obtain the weight of the historical risk event data, wherein the weight of the historical risk event data decays faster over time; The weights and spatial distribution density of the historical risk event data are associated with the corresponding grid cells to form a real-time risk density grid. The external risk map is formed by integrating the real-time risk density network corresponding to all grid cells in the target geographical area.
[0089] In one embodiment, the first analysis module 620 is specifically used for: The real-time risk density grid is input into the first prediction model, and the real-time risk density network is used as a dynamic input feature layer of the first prediction model to perform feature-level fusion with the environmental dwell point data to generate the environmental risk level and risk prediction probability.
[0090] In one embodiment, the data acquisition module 610 is further configured to: A unified event severity quantification scoring model is established for historical risk event data from different data sources to generate a severity score for each historical risk event data. Based on the data update frequency and authority level of each data source, different confidence weights are assigned to the historical risk event data from different sources, wherein the data update frequency is directly proportional to the confidence weight value, and the authority level is directly proportional to the confidence weight value. For the historical risk event data in each grid cell, the historical risk event data is weighted based on the confidence weight corresponding to each historical risk event data to obtain the real-time risk corresponding to each grid cell; The weights of the real-time risk and the historical risk event data, as well as the spatial distribution density of the historical risk event data, are associated with the corresponding grid cells to form a real-time risk density network corresponding to the grid cells.
[0091] Understandably, traditional pricing relies solely on the vehicle's selling price, resulting in an extremely limited data dimension. This invention's solution lays the foundation for refined pricing by acquiring three types of core data. The multimodal state data in this invention covers environmental location data and internal characteristic data, directly capturing two key pieces of information neglected by traditional pricing: the usage environment and added value. The external risk map integrates risk data such as theft and collisions in areas where vehicles are frequently parked, quantifying differences in environmental risks. The configuration benchmark data is aligned with OEM configuration information and industry-standard premiums to ensure the accuracy of basic value assessment.
[0092] Furthermore, traditional pricing ignores the risk differences in vehicle usage environments. This embodiment achieves precise risk quantification through a first prediction model. The model input combines vehicle dwell point data with external risk maps. The model output generates environmental risk levels and risk prediction probabilities, transforming intangible environmental risks into quantifiable pricing criteria. This enables reasonable upward pricing for vehicles in high-risk environments and more favorable pricing for vehicles in low-risk environments, avoiding the problem of insurance companies losing money due to low-price insurance for high-risk vehicles, and ensuring that pricing matches actual risk.
[0093] Furthermore, traditional pricing only considers the vehicle's selling price, ignoring the added value differences brought by the owner's accessories and optional configurations. The solution of this invention achieves accurate quantification of added value through a second evaluation model. The model analyzes internal characteristic data and configuration benchmarks through input, and generates internal value coefficients through output, transforming intangible added value into a quantifiable pricing basis. This allows for a reasonable price increase for vehicles with high added value, ensuring that subsequent compensation can cover the actual value. Vehicles with low added value do not need to bear additional premiums, avoiding the situation where low-priced vehicles pay for high-priced vehicles, and ensuring that the pricing matches the actual value of the vehicle.
[0094] Finally, traditional pricing is a static model where a fixed selling price corresponds to a fixed premium. The solution of this invention achieves multi-dimensional integrated pricing of basic value, added value, and environmental risk through dynamic adjustment algorithms, replacing the traditional single selling price pricing. This ensures that when pricing car insurance, various factors affecting car insurance are fully considered, thereby ensuring sufficient resource allocation for car insurance in the end.
[0095] Specific limitations regarding the resource allocation device can be found in the limitations of the resource allocation method above, and will not be repeated here. Each module in the aforementioned resource allocation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in the computer device in hardware form, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to each module.
[0096] Based on the above resource allocation methods, such as Figure 7 As shown in the diagram, this embodiment of the invention also provides a structural schematic of an apparatus for a resource allocation method, the apparatus including a processor 71 and a memory 72 coupled to the processor 71. The memory 72 stores a computer program, which, when executed by the processor 71, causes the processor 71 to perform the steps of the resource allocation method in the above embodiment.
[0097] For further details regarding the implementation of the above technical solution by the processor 71 in the device for the above resource allocation method steps, please refer to the description of the resource allocation method provided in the above embodiments of the invention, which will not be repeated here.
[0098] The processor 71 can also be called a CPU (Central Processing Unit). The processor 71 may be an integrated circuit chip with signal processing capabilities. The processor 71 can also be a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor can be a microprocessor, or the processor 71 can be any conventional processor.
[0099] like Figure 8 As shown in the diagram, this embodiment of the invention also provides a schematic diagram of a computer-readable storage medium, on which a readable computer program 81 is stored. The computer program 81 can be stored in the storage medium in the form of a software product, including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods described in various embodiments of the invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks or optical disks, ROM (Read-Only Memory), RAM (Random Access Memory), or terminal devices such as computers, servers, mobile phones, and tablets.
[0100] In the embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, apparatuses, or modules, and may be electrical, mechanical, or other forms.
[0101] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.
[0102] Furthermore, the functional modules in the various embodiments of this invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium.
[0103] In the above embodiments, the implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, in the form of a computer program product.
[0104] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium, or a semiconductor medium (e.g., SSD (solid state disk)).
[0105] The technical solution provided by the present invention has been described in detail above. Specific examples have been used to illustrate the principle and implementation of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core idea of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation and application scope based on the idea of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.
[0106] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, etc.) containing computer-usable program code.
[0107] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0108] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0109] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0110] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.
Claims
1. A resource allocation method, characterized in that, include: Acquire multimodal state data of the target object, an external risk map faced by the target object, and configuration baseline data corresponding to the target object. The multimodal state data includes at least environmental dwell point data and internal feature data. The environmental relocation point data and the external risk map are input into the first prediction model for spatiotemporal risk analysis to generate the environmental risk level and risk prediction probability of the target object. The internal feature data and the configuration baseline data are input into the second evaluation model for value analysis to generate the internal value coefficient of the target object. When the predicted risk probability meets a preset threshold condition, the environmental risk level and the internal value coefficient are input into a preset dynamic adjustment algorithm to calculate and generate dynamic resource allocation parameters for the target object; wherein, the dynamic resource allocation parameters are used to guide resource allocation decisions for the target object.
2. The resource allocation method according to claim 1, characterized in that, The step of inputting the environmental residency point data and the external risk map into the first prediction model for spatiotemporal risk analysis to generate the environmental risk level and risk prediction probability of the target object includes: Extract the historical location sequence and dwell time pattern of the target object from the environmental dwell point data; Obtain an external risk map associated with the historical location sequence of the outposts, wherein the external risk map is constructed based on the spatiotemporal distribution of historical risk events within a geographic grid; The historical location sequence of the dwelling point, the dwelling time pattern, and the external risk map features are input into a sequence prediction model based on LSTM, and the output is the risk prediction probability of the target object in the future preset time period and the environmental risk level of the target object's location.
3. The resource allocation method according to claim 1, characterized in that, The step of inputting the internal feature data and the configuration baseline data into the second evaluation model for value analysis to generate the internal value coefficient of the target object includes: The image data in the internal feature data is preprocessed; The processed image data is input into an image recognition model built on CNN to identify the internal component categories of the target object and the state attributes of the internal component categories. Based on the internal component categories, state attributes, and baseline configuration information of the target object obtained from the configuration baseline data source, the internal value coefficient of the target object is calculated using a value decay algorithm. The internal value coefficient is proportional to the value of the category weight factor of the internal component category set in the value decay algorithm, and the internal value coefficient is proportional to the value of the time depreciation factor determined based on the state attributes.
4. The resource allocation method according to claim 1, characterized in that, The dynamic adjustment algorithm is as follows: in, Parameters for allocating the dynamic resources, These are the baseline parameters determined based on the baseline attributes of the target object. The internal value coefficient is... The risk adjustment factor is determined based on the environmental risk level, and the risk adjustment factor has a positive correlation with the environmental risk level.
5. The resource allocation method according to claim 1 or 2, characterized in that, Obtaining the external risk map faced by the target object includes: Historical risk event data within a target geographic area is periodically acquired from multiple heterogeneous data sources; Perform spatiotemporal alignment and standardization on the acquired historical risk event data; The kernel density estimation algorithm is used to calculate the spatial distribution density of historical risk event data in each grid cell of the target geographic area. The processed historical risk event data is calculated using a time decay function to obtain the weight of the historical risk event data, wherein the weight of the historical risk event data decays faster over time; The weights and spatial distribution density of the historical risk event data are associated with the corresponding grid cells to form a real-time risk density grid. The external risk map is formed by combining the real-time risk density grids corresponding to all grid cells in the target geographical area.
6. The resource allocation method according to claim 5, characterized in that, The step of inputting the environmental residency point data and the external risk map into the first prediction model for spatiotemporal risk analysis to generate the environmental risk level and risk prediction probability of the target object includes: The real-time risk density grid is input into the first prediction model, and the real-time risk density network is used as a dynamic input feature layer of the first prediction model to perform feature-level fusion with the environmental dwell point data to generate the environmental risk level and risk prediction probability.
7. The resource allocation method according to claim 5, characterized in that, The process of associating the weights of the historical risk event data and the spatial distribution density of the historical risk event data with corresponding grid cells to form a real-time risk density grid includes: A unified event severity quantification scoring model is established for historical risk event data from different data sources to generate a severity score for each historical risk event data. Based on the data update frequency and authority level of each data source, different confidence weights are assigned to the historical risk event data from different sources, wherein the data update frequency is directly proportional to the confidence weight value, and the authority level is directly proportional to the confidence weight value. For the historical risk event data in each grid cell, the historical risk event data is weighted based on the confidence weight corresponding to each historical risk event data to obtain the real-time risk corresponding to each grid cell; The weights of the real-time risk and the historical risk event data, as well as the spatial distribution density of the historical risk event data, are associated with the corresponding grid cells to form a real-time risk density network corresponding to the grid cells.
8. A resource allocation device, characterized in that, include: The data acquisition module is used to acquire multimodal state data of the target object, the external risk map faced by the target object, and the configuration baseline data corresponding to the target object. The multimodal state data includes at least environmental dwell point data and internal feature data. The first analysis module is used to input the environmental residence point data and the external risk map into the first prediction model to perform spatiotemporal risk analysis and generate the environmental risk level and risk prediction probability of the target object. The second analysis module is used to input the internal feature data and the configuration benchmark data into the second evaluation model for value analysis and generate the internal value coefficient of the target object. The resource allocation module is used to input the environmental risk level and the internal value coefficient into a preset dynamic adjustment algorithm when the risk prediction probability meets a preset threshold condition, and calculate and generate dynamic resource allocation parameters for the target object; wherein, the dynamic resource allocation parameters are used to guide resource allocation decisions for the target object.
9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the resource allocation method as described in any one of claims 1 to 7.
10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the resource allocation method as described in any one of claims 1 to 7.