Car insurance premium pricing method and device based on data fusion, equipment and medium

Abstract

The invention relates to the technical field of artificial intelligence and financial science and technology, and discloses a vehicle insurance premium pricing method, device, equipment and medium based on data fusion, and the method comprises the steps: obtaining the external environment data, vehicle attribute data and customer evaluation data of an unmanned vehicle, carrying out the coding of a driving scene of the unmanned vehicle according to the external environment data, and obtaining a vehicle insurance premium pricing result; obtaining scene coding data; calculating performance data of the unmanned vehicle according to the vehicle attribute data, and performing code conversion on the customer evaluation data to obtain evaluation code data; performing data fusion on the scene coding data, the performance data and the evaluation coding data to obtain multi-source fusion data; analyzing the multi-source fusion data by using a pre-constructed risk evaluation model to obtain a risk regression coefficient corresponding to each section of driving service; and determining the vehicle insurance pricing premium of the unmanned vehicle according to the pre-acquired reference premium, the corrected rate adjustment coefficient and the risk regression coefficient. According to the invention, the accuracy of insurance premium pricing can be improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence and financial technology, and in particular to a method, device, equipment and medium for pricing automobile insurance premiums based on data fusion. Background Art

[0002] With the maturity and popularization of autonomous driving technology, some cities have launched unmanned vehicle travel services on autonomous driving travel service platforms such as Luobo Kuaipao. With the development of 5G technology and vehicle networking technology, unmanned driving services will become more and more popular in the future. The driving performance of unmanned vehicles is determined by the interactive decision-making of the vehicle's autonomous driving algorithm and various sensor data.

[0003] The traditional auto insurance premium pricing model mainly uses differentiated pricing based on factors such as the purchase price of a new car, mileage, and historical claims. More advanced pricing methods also use user historical driving behavior data, but the basis of pricing is basically a premium pricing model based on human driving. The traditional auto insurance premium pricing model is not suitable for situations where the vehicle's autonomous driving algorithm interacts with various sensor data for decision-making. More complex and diverse data can also be obtained through the on-board terminal system, resulting in lower accuracy in auto insurance premium pricing. Summary of the invention

[0004] The present invention provides a method, device, equipment and medium for pricing automobile insurance premiums based on data fusion, so as to solve the technical problem of low accuracy in pricing automobile insurance premiums.

[0005] In the first aspect, a method for pricing auto insurance premiums based on data fusion is provided, including:

[0006] Acquire external environment data, vehicle attribute data, and customer evaluation data of the driverless vehicle, and encode the driving scene of the driverless vehicle according to the external environment data to obtain scene encoding data;

[0007] Calculating performance data of the driverless vehicle based on the vehicle attribute data;

[0008] Performing encoding conversion on the customer evaluation data to obtain evaluation encoding data;

[0009] fusing the scene coding data, the performance data and the evaluation coding data to obtain multi-source fused data;

[0010] Analyze the multi-source fusion data using a pre-built risk assessment model to obtain a risk regression coefficient corresponding to each driving service of the driverless vehicle;

[0011] The car insurance pricing premium for the driverless car is determined based on the pre-acquired base premium, the modified rate adjustment coefficient and the risk regression coefficient.

[0012] In a second aspect, a vehicle insurance premium pricing device based on data fusion is provided, comprising:

[0013] A driving scene encoding module, used to obtain external environment data, vehicle attribute data and customer evaluation data of the driverless car, and encode the driving scene of the driverless car according to the external environment data to obtain scene encoding data;

[0014] A performance data calculation module, used to calculate the performance data of the driverless vehicle according to the vehicle attribute data;

[0015] A coding conversion module, used for coding the customer evaluation data to obtain evaluation coding data;

[0016] A multi-source data fusion module, used for fusing the scene coding data, the performance data and the evaluation coding data to obtain multi-source fusion data;

[0017] A risk regression coefficient calculation module is used to analyze the multi-source fusion data using a pre-built risk assessment model to obtain a risk regression coefficient corresponding to each driving service of the driverless vehicle;

[0018] The motor vehicle insurance pricing premium determination module is used to determine the motor vehicle insurance pricing premium of the driverless car based on the pre-acquired benchmark premium, the modified rate adjustment coefficient and the risk regression coefficient.

[0019] In a third aspect, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the above-mentioned method for pricing automobile insurance premiums based on data fusion are implemented.

[0020] In a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned method for pricing automobile insurance premiums based on data fusion are implemented.

[0021] The embodiment of the present invention is based on the data from the external and multi-source data of the unmanned vehicle terminal, and uses the Transformer network mechanism to perform heterogeneous data fusion processing, and then combines it with the deep learning network for training and analysis to obtain the future travel risk of the unmanned service vehicle. Based on the travel risk, the specific service city environment of the unmanned vehicle can be considered, and the specific data of each trip is fully utilized for evaluation. The premium pricing model based on risk assessment takes into account both the vehicle and service characteristics of the unmanned vehicle and the dynamic personalized pricing characteristics of using a single vehicle based on actual operating risks. Therefore, the auto insurance premium pricing method, device, equipment and medium based on data fusion proposed by the present invention can solve the problem of low accuracy in auto insurance premium pricing. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0023] Figure 1 is a schematic diagram of an application environment of a method for pricing automobile insurance premiums based on data fusion in one embodiment of the present invention;

[0024] Figure 2 It is a flow chart of a method for pricing automobile insurance premium based on data fusion in one embodiment of the present invention;

[0025] Figure 3 yes Figure 1 A schematic flow chart of a specific implementation of step S2;

[0026] Figure 4 yes Figure 1 A schematic flow chart of a specific implementation of step S3;

[0027] Figure 5 It is a structural schematic diagram of a vehicle insurance premium pricing device based on data fusion in one embodiment of the present invention;

[0028] Figure 6 is a schematic diagram of a structure of a computer device in one embodiment of the present invention;

[0029] Figure 7 It is another structural schematic diagram of a computer device in one embodiment of the present invention. DETAILED DESCRIPTION

[0030] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0031] The car insurance premium pricing method based on data fusion provided by the embodiment of the present invention can be applied in Figure 1 In an application environment, the client communicates with the server through a network. The server can obtain the external environment data, vehicle attribute data and customer evaluation data of the driverless car through the client, encode the driving scene of the driverless car according to the external environment data to obtain scene encoding data; calculate the performance data of the driverless car according to the vehicle attribute data, encode and convert the customer evaluation data to obtain evaluation encoding data; fuse the scene encoding data, performance data and evaluation encoding data to obtain multi-source fusion data; use a pre-built risk assessment model to analyze the multi-source fusion data to obtain the risk regression coefficient corresponding to each driving service of the driverless car; according to the pre-obtained benchmark premium, modified rate adjustment coefficient and risk The regression coefficient determines the auto insurance pricing premium of the driverless car, and the auto insurance pricing premium is returned to the client. In the present invention, based on the data from the external and multi-source data of the unmanned vehicle terminal, the Transformer network mechanism is used to perform heterogeneous data fusion processing, and then combined with the deep learning network for training and analysis, the future travel risk of the unmanned service vehicle is obtained. Based on the travel risk, the specific service city environment of the unmanned vehicle can be considered, and the specific data of each trip is fully utilized for evaluation. The premium pricing model based on risk assessment takes into account both the vehicle and service characteristics of the unmanned vehicle, and the dynamic personalized pricing characteristics of using a single vehicle based on actual operating risks. Among them, the client can be but is not limited to various personal computers, laptops, smart phones, tablets and portable wearable devices. The server can be implemented with an independent server or a server cluster consisting of multiple servers. The present invention is described in detail below through specific embodiments.

[0032] See also Figure 2 As shown, Figure 2 A flow chart of a method for pricing automobile insurance premiums based on data fusion provided by an embodiment of the present invention includes the following steps:

[0033] S1. Obtain external environment data, vehicle attribute data, and customer evaluation data of the driverless vehicle, and encode the driving scene of the driverless vehicle according to the external environment data to obtain scene encoding data.

[0034] In the embodiment of the present invention, the external environment data refers to basic data such as the geographical location, weather, road structure and road conditions of the unmanned vehicle service area, which mainly reflects the basic travel risks of the unmanned vehicle service city and region; the vehicle attribute data includes the vehicle's own data and the vehicle's dynamic driving sensor data, and the vehicle's own data includes the configuration information and fatigue status of the vehicle's core hardware. The core hardware data of the unmanned vehicle includes hardware information such as the sensor photosensitivity of the vehicle body, the calculation speed of the algorithm system, the engine fatigue life and the brake pad braking distance, which mainly reflects the failure and error risks of the unmanned vehicle. The various dynamic driving sensor data of the unmanned vehicle under the control of the automatic driving algorithm will reflect the driving behavior habits and judgment ability of similar driving vehicles. According to the behavioral data such as the unmanned driving speed, acceleration, braking frequency, vehicle spacing and traffic rules compliance, it can well reflect the various braking and strain capabilities of the unmanned vehicle algorithm under actual road conditions, which are used to evaluate the actual driving risk of the automatic driving algorithm; the customer evaluation data refers to the user experience and evaluation data of each unmanned driving ride service, such as the score and natural language evaluation data of the trip, reflecting the actual driving risk of the unmanned vehicle.

[0035] In detail, external environmental data can be collected based on local urban geographic location information, such as urban road structure, weather and road conditions, which can be collected based on GPS navigation systems and officially published data such as the Transportation Bureau or Meteorological Bureau. Road structure and road condition information can be obtained based on common navigation apps, such as congested sections, sections prone to accidents, or one-way streets and other regional road layout information; the vehicle's own data can be collected based on market automobile manufacturers or professional automobile testing organizations or apps such as Autohome. The vehicle's own driving data comes from body sensors and cameras, and is generally available in the vehicle's EDR system; customer evaluation data can be obtained from a pre-stored storage area through computer statements with data capture functions (such as Java statements, Python statements, etc.), where the storage area includes but is not limited to databases and blockchains.

[0036] Furthermore, by investigating and collecting basic environmental data of cities served by driverless cars, and dividing and coding scenes according to the environmental data of the cities served, it can help the driverless system to quickly identify and adapt to different scenarios and improve its ability to respond to potential dangers.

[0037] In the embodiment of the present invention, the scene coding data refers to a method of systematically representing the characteristics of the external environment information and the driving scene. It converts different environmental characteristics (such as weather, road conditions, time, etc.) into a coding form that can be processed by the algorithm through a specific format. This coding enables the unmanned driving system to quickly identify and adapt to different driving environments.

[0038] In the embodiment of the present invention, encoding the driving scene of the driverless vehicle according to the external environment data to obtain scene encoding data includes:

[0039] Extracting environmental features from the external environmental data;

[0040] identifying a driving scenario of the driverless vehicle according to the environmental features;

[0041] Dividing scene features in the driving scene to obtain scene category features;

[0042] The scene category features are feature spliced ​​to obtain scene coding data corresponding to the driving scene.

[0043] In detail, the sources of environmental data include meteorological information, road condition monitoring, traffic signals and surrounding obstacles, etc., extracting environmental features from external environmental data, such as sunny, cloudy, light rain, heavy rain, snow, etc. in weather conditions; dry, slippery, icy, construction, etc. in road conditions; congestion, smooth traffic, accidents, etc. in traffic conditions; daytime and night in time period information, and then identifying the driving scenarios of the driverless car based on the environmental features, such as determining the current driving scenario based on the city’s geographical location and driving environment information, e.g., heavy rain in a city in summer, sunny weather in a city in winter, etc.

[0044] Specifically, the scene features in the driving scene are divided to obtain scene category features, such as identifying the city, season, weather, road condition, etc. in the driving scene, and the identified scene category features are spliced ​​to form a unified code. It is assumed that the coding format is city (2 digits), season (1 digit), weather (2 digits), road condition (a combination of n letters and numbers). For example, the rainy weather in Wuhan in summer is a scene, which can be encoded as a data expression of city + season + weather + road condition (010202A). In this way, all scenes can be encoded and classified.

[0045] Furthermore, driverless vehicles are equipped with a variety of sensors, such as cameras, radars, and GPS, which can significantly improve their safety, reliability, and efficiency. In order to reflect the failure and error risks of driverless vehicles, it is necessary to analyze the various performance indicators of driverless vehicles.

[0046] S2. Calculate the performance data of the driverless vehicle according to the vehicle attribute data.

[0047] In the embodiment of the present invention, the performance data is various indicators used to evaluate and measure the operating performance of the driverless car under different conditions, such as perception performance, power performance, and safety performance, among which perception performance such as sensor accuracy (such as the detection accuracy of cameras, radars and LiDAR), environmental recognition ability (the ability to recognize traffic signs, pedestrians, obstacles, etc.); power performance such as acceleration ability (such as the time required for 0-100km / h), braking distance and response time; safety performance such as collision avoidance ability (the ability to react and avoid in various situations).

[0048] In the embodiment of the present invention, refer to Figure 3 As shown, the calculating the performance data of the driverless vehicle according to the vehicle attribute data includes:

[0049] S31, dividing the vehicle attribute data into vehicle core hardware data and vehicle behavior data;

[0050] S32, performing curve calculation on the core hardware equipment of the driverless vehicle according to the core hardware data of the vehicle to obtain performance data of the driverless vehicle;

[0051] S33, performing multi-matrix fusion on the vehicle behavior data to obtain behavior fusion data of the driverless vehicle;

[0052] S34. Determine the performance data of the driverless vehicle according to the efficiency data and the behavior fusion data.

[0053] Specifically, the core hardware data includes the sensor photosensitivity of the vehicle body, the calculation speed of the algorithm system, the engine fatigue life and the braking distance of the brake pads, etc. The vehicle behavior data includes the acceleration, braking, steering angle velocity, driving speed, etc. in driving behavior, and the road type, traffic density, weather conditions, obstacle location, etc. in environmental information.

[0054] Specifically, using methods such as multivariate regression analysis, we can build a performance model based on the performance parameters of different hardware. Curve processing is often used to fit experimental data to find the best mathematical model to describe the trend of the data. For example, we can build a relationship curve between processor performance and its response time, and calculate the performance indicators of each core hardware, such as processing delay, perception accuracy, energy efficiency, etc. For example, if we calculate the fatigue and performance of the core hardware equipment of the driverless car based on the test results of professional manufacturers, different components have corresponding fatigue calculation formulas and measurement indicators. For example, the formula for the fatigue test of the vehicle brake pad is S=V 2 / gμS, this formula is used to calculate the braking distance of a car, where S represents the braking distance, V is the speed of the vehicle, g is the acceleration of gravity, and μ is the friction coefficient between the tire and the road. This formula reveals the relationship between the braking distance and the vehicle speed, the acceleration of gravity, and the friction coefficient; the vehicle behavior data is collected by cameras, radars, and GPS. By performing multi-matrix fusion on the vehicle behavior data and integrating various data sources, that is, standardizing different types of behavior data into the same dimension, and constructing a behavior data matrix, including driving status, environmental information, and task information, the data from different sources are integrated into a comprehensive behavior data set. The performance data and behavior fusion data are combined to calculate the comprehensive performance data of the driverless car. The performance data of the driverless car can be extracted from the vehicle's core hardware data and behavior data.

[0055] Furthermore, the image data of the external environment collected on the vehicle driving section is processed by cropping, scaling and normalization using computer vision and graphics technology, and the customer evaluation service data of each section of unmanned driving service is encoded and converted using natural language processing technology.

[0056] S3. Convert the customer evaluation data into code to obtain evaluation code data.

[0057] In an embodiment of the present invention, the evaluation encoding data refers to converting the customer's evaluation information into data in a numerical or vector format, converting each evaluation into a numerical vector of a fixed dimension, and including sentiment classification information such as positive, negative or neutral during the encoding process, which helps to understand the customer's feedback emotions.

[0058] In the embodiment of the present invention, the step of converting the customer evaluation data into code to obtain evaluation code data includes:

[0059] Dividing the customer evaluation data into evaluation data corresponding to each segment of unmanned driving service;

[0060] Identifying customer emotions corresponding to the evaluation data, and classifying the evaluation data according to the customer emotions to obtain evaluation data categories;

[0061] Vector encoding is performed on each evaluation in the evaluation data category to obtain evaluation encoding data.

[0062] In detail, the customer evaluation data is divided according to each segment of unmanned driving service to obtain a corresponding evaluation data set, which can be segmented by timestamp, service identifier and other information, such as service 1 corresponds to user feedback 1, service 2 corresponds to user feedback 2, etc., and then natural language processing (NLP) is used to identify customer emotions in each segment of evaluation data. For example, if the customer evaluation data is that the service is very helpful, the identified emotion is positive emotion, and if the customer evaluation is somewhat dissatisfied, the identified emotion is negative emotion. According to the identified customer emotions, the evaluation data is classified, such as positive evaluation, neutral evaluation, and negative evaluation. For example, positive evaluation is user feedback 1 and user feedback 3, and message evaluation is user feedback 3. Then, each evaluation is vector encoded, wherein each evaluation in the evaluation data category can be vector encoded by a vector conversion model to obtain evaluation encoding data. The vector conversion model includes but is not limited to Word2Vec and BERT models, and then the vector encodings of all evaluations are summarized to obtain the final evaluation encoding data set, so that the customer evaluation data can be effectively encoded and converted into evaluation encoding data that can be used for data analysis and model training.

[0063] Specifically, the customer's comments are classified according to each section of the driverless service to ensure that each comment can be associated with a specific service time period, such as collecting customer evaluation records during the use of driverless services, and dividing the comments into different sections according to timestamps or service IDs. For example, a user may leave feedback after completing a trip, and these feedbacks need to match the specific time period of the trip. Then, based on the results of sentiment analysis, the evaluation data is divided into different categories, positive comments: such as "very satisfied", "good experience", etc., negative comments: such as "poor service", "problems", etc., neutral comments: such as "normal service", "no special feeling", etc. The classified evaluation data is converted into a numerical form that can be processed by computers for subsequent analysis and model training.

[0064] Furthermore, multi-source data collected by driverless cars can reduce the deviation caused by a single data source, thereby improving the accuracy and reliability of analysis results. By combining scenario, performance and customer evaluation data, potential patterns and trends can be discovered to help identify key factors affecting customer satisfaction.

[0065] S4. Fusing the scene coding data, the performance data and the evaluation coding data to obtain multi-source fused data.

[0066] In the embodiment of the present invention, the multi-source fusion data refers to the generation of a comprehensive and unified data set by integrating data from different sources. Such fusion can improve the overall quality and information content of the data, thereby better supporting analysis and decision-making.

[0067] In the embodiment of the present invention, referring to Figure 4 As shown, the scene coding data, the performance data and the evaluation coding data are fused to obtain multi-source fused data, including:

[0068] S41, extracting scene features corresponding to the scene coding data, extracting performance features corresponding to the performance data, and determining evaluation features according to the evaluation coding data;

[0069] S42, spatially aligning the scene feature, the performance feature, and the evaluation feature to obtain a spatial alignment feature;

[0070] S43, using a preset Transformer network unit to perform data fusion on the spatial alignment features to obtain multi-source fusion data.

[0071] In detail, considering that the driving behavior data of each section of the journey is continuous and multi-dimensional, the measurement methods and encoding meanings of data from multiple sources have different effects on the final risk assessment. Considering the use of deep learning modeling to predict each driving service, in the current offline risk prediction based on driving behavior, the Transformer network unit can better handle the spatiotemporal sequence data model and multi-source, multi-modal data input, and is a key network structure for multi-source data fusion processing, that is, standardize and normalize different types of data to ensure that they are on the same scale, generate time series features, and extract spatial features, including road section type, traffic density, etc. The Transformer network unit architecture includes an input layer: encode multi-source data into a format suitable for input Transformer, and usually use an embedding layer to map data of different modalities into vectors of the same dimension; multi-head self-attention mechanism: through the multi-head attention mechanism, the model can capture long-range dependencies and important features in the data, so as to better understand the relationship between driving behavior and risk; position encoding: Since the Transformer lacks the order information of time series data, position encoding needs to be added to retain time information.

[0072] Specifically, key features are extracted from the scene coding data, such as scene type, environmental conditions, user behavior, etc., and performance indicators related to the product or service are extracted from the performance data, such as response time, stability, efficiency, etc. The scene features and performance features are converted into vector representations, and the vectors corresponding to the evaluation coding data are represented as evaluation features, so that the scene features, performance features and evaluation features after vector conversion are spatially aligned so that they are expressed in the same space to ensure that the features can be effectively compared and combined. Common methods include normalization, standardization or the use of embedding technology to map data of different dimensions to the same feature space, and then use the preset Transformer network model to fuse the spatially aligned features. Transformer is good at processing sequence data and can capture the complex relationship between features through the self-attention mechanism. The spatially aligned features are input into the Transformer network to output multi-source fused data, which contains integrated feature representations that can better reflect the scene, performance and evaluation information.

[0073] Furthermore, based on the fused multi-source fusion data, it is necessary to analyze the risks of autonomous driving, which helps to identify potential safety hazards and then improve the safety of autonomous vehicles through effective risk management, thereby enhancing consumers' trust and acceptance of the technology.

[0074] S5. Analyze the multi-source fusion data using a pre-built risk assessment model to obtain a risk regression coefficient corresponding to each driving service of the driverless vehicle.

[0075] In the embodiment of the present invention, the risk regression coefficient refers to a parameter that reflects the degree of influence of a certain characteristic variable on the target variable (such as risk score) in regression analysis, indicating the magnitude and direction of the influence of each unit change of the characteristic variable on the risk score. For example, if the regression coefficient of a certain characteristic is a positive value, it means that an increase in the characteristic value will lead to an increase in the risk score; if it is a negative value, it means that an increase in the characteristic value will lead to a decrease in the risk score.

[0076] In the embodiment of the present invention, the multi-source fusion data is analyzed by using a pre-built risk assessment model to obtain a risk regression coefficient corresponding to each driving service of the driverless vehicle, including:

[0077] Dividing the multi-source fusion data to obtain fusion data corresponding to each driving service segment;

[0078] Extracting fusion features corresponding to the fusion data one by one;

[0079] Calculating the risk score corresponding to the fusion feature using the risk assessment model;

[0080] The risk regression coefficient corresponding to each driving service segment is extracted according to the risk score.

[0081] In detail, the collected multi-source data are labeled with risks for each driving experience according to expert systems, simulation, augmented reality and other methods, and then the Transformer network is used for multi-source data ingestion and feature extraction processing, combined with deep learning neural network training to obtain a reliable risk assessment model, that is, data is collected from different sensors (such as cameras, lidar, GPS, etc.), integrated into a unified data format, and risk labeled through an expert system, that is, the knowledge of domain experts is used to conduct risk assessment on each driving experience, and potential risk events (such as collisions, sudden braking, etc.) are labeled. The simulation mode refers to the use of a driving simulator to simulate different driving scenarios, record and label risk conditions, and augmented reality is the real-time labeling of risk information through augmented reality technology in a real environment to improve the accuracy of labeling, and then features are extracted through the Transformer network, that is, multi-source data is input into Tra The Transformer network uses the self-attention mechanism to extract important features, so that the model can focus on the key parts of the data, generate feature representations for each driving experience, capture complex time series and spatial relationships, and input the features extracted by the Transformer into the deep learning model. It uses labeled risk labels for supervised learning, selects appropriate loss functions (such as cross entropy loss, mean square error, etc.), and optimizes model parameters to obtain a trained risk assessment model. The output of the model can be a risk score, which represents the potential risk of this driving service. The trained model is deployed to the actual system to achieve real-time risk monitoring and early warning, helping drivers make safe decisions, and then using the Transformer network to perform risk prediction on multi-source driving behavior data, which can effectively integrate different data sources, capture complex spatiotemporal relationships, and improve the accuracy and reliability of risk prediction.

[0082] Specifically, the data is divided into multiple driving service segments according to timestamp or distance to ensure the integrity and coherence of each segment of data. Key features are extracted from the data of each driving service segment, and the extracted fusion features are input into the pre-built risk assessment model to calculate the risk score of each service segment. Based on the calculated risk score, regression analysis methods (such as linear regression, logistic regression, etc.) are used to extract the risk regression coefficient. Each regression coefficient represents the degree of influence of the corresponding feature on the overall risk score.

[0083] Furthermore, by using the risk regression coefficient, personalized pricing can be performed based on the characteristics and risk levels of different driverless vehicles, so that premiums can better reflect actual risks. Accurate pricing strategies enable insurance companies to adapt to market changes more quickly and enhance their competitiveness in the field of driverless car insurance.

[0084] S6. Determine the car insurance pricing premium for the driverless car based on the pre-acquired base premium, the modified rate adjustment coefficient and the risk regression coefficient.

[0085] In the embodiment of the present invention, the automobile insurance pricing premium refers to the insurance fee that the car owner needs to pay, calculated by the insurance company based on a specific risk assessment and pricing model.

[0086] In the embodiment of the present invention, the method of determining the car insurance pricing premium of the driverless car based on the pre-acquired base premium, the modified rate adjustment coefficient and the risk regression coefficient includes:

[0087] Multiplying the base premium, the modified rate adjustment coefficient and the risk regression coefficient to obtain an initial pricing premium for the driverless car;

[0088] Determine travel risk based on multi-source fusion data corresponding to each driving service of the driverless car;

[0089] The initial pricing premium is dynamically adjusted according to the travel risk to obtain a pricing premium for the auto insurance of the driverless car.

[0090] In detail, the traditional auto insurance premium pricing formula is roughly as follows: Commercial auto insurance premium = base premium × rate adjustment coefficient. Based on the vehicle and service characteristics of driverless cars, the driver's pricing influence is removed and the underwriter adjusts the rate adjustment coefficient. After each trip service of the driverless car, multi-source data is used to evaluate the risk of the service trip and the premium pricing is dynamically adjusted. Then, the premium calculation formula based on the specific performance data of each driverless car driving service can be adjusted to: Commercial auto insurance premium = base premium × modified rate adjustment coefficient × predicted risk regression coefficient.

[0091] Specifically, we will continue to collect and analyze multi-source data in order to accurately assess travel risks, establish a system to monitor driverless vehicles in real time, promptly update risk assessments and premium pricing, and continuously optimize risk assessment models and pricing strategies based on user feedback and accident situations. This will enable scientific and dynamic auto insurance pricing for driverless vehicles, and provide insurance companies and car owners with more reasonable and effective insurance solutions.

[0092] It can be seen that in the above scheme, based on the data from multiple sources outside the unmanned vehicle terminal and itself, the Transformer network structure is used to perform heterogeneous data fusion processing, and then combined with the deep learning network for training and analysis, the future travel risk of the unmanned driving service vehicle is obtained. Based on the travel risk, the specific service city environment of the unmanned vehicle can be considered, and the specific data of each trip is fully utilized for evaluation. The premium pricing model based on risk assessment takes into account both the vehicle and service characteristics of the unmanned vehicle, and the dynamic personalized pricing characteristics of using a single vehicle based on actual operating risks.

[0093] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean 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 embodiment of the present invention.

[0094] In one embodiment, a vehicle insurance premium pricing device based on data fusion is provided, and the vehicle insurance premium pricing device based on data fusion corresponds to the vehicle insurance premium pricing method based on data fusion in the above embodiment. Figure 5 As shown, the vehicle insurance premium pricing device based on data fusion includes a driving scene encoding module 101, a performance data calculation module 102, a coding conversion module 103, a multi-source data fusion module 104, a risk regression coefficient calculation module 105 and a vehicle insurance pricing premium determination module 106. The functional modules are described in detail as follows:

[0095] The driving scene encoding module 101 is used to obtain the external environment data, vehicle attribute data and customer evaluation data of the driverless car, and encode the driving scene of the driverless car according to the external environment data to obtain scene encoding data;

[0096] A performance data calculation module 102, used to calculate the performance data of the driverless vehicle according to the vehicle attribute data;

[0097] The encoding conversion module 103 is used to perform encoding conversion on the customer evaluation data to obtain evaluation encoding data;

[0098] A multi-source data fusion module 104 is used to fuse the scene coding data, the performance data and the evaluation coding data to obtain multi-source fusion data;

[0099] The risk regression coefficient calculation module 105 is used to analyze the multi-source fusion data using a pre-built risk assessment model to obtain a risk regression coefficient corresponding to each driving service of the driverless vehicle;

[0100] The auto insurance pricing premium determination module 106 is used to determine the auto insurance pricing premium of the driverless car based on the pre-acquired benchmark premium, the modified rate adjustment coefficient and the risk regression coefficient.

[0101] In one embodiment, the driving scene encoding module 101, when encoding the driving scene of the driverless vehicle according to the external environment data to obtain the scene encoding data, is used to:

[0102] Extracting environmental features from the external environmental data;

[0103] identifying a driving scenario of the driverless vehicle according to the environmental features;

[0104] Dividing scene features in the driving scene to obtain scene category features;

[0105] The scene category features are feature spliced ​​to obtain scene coding data corresponding to the driving scene.

[0106] In one embodiment, the performance data calculation module 102, when calculating the performance data of the driverless vehicle according to the vehicle attribute data, is used to:

[0107] Dividing the vehicle attribute data into vehicle core hardware data and vehicle behavior data;

[0108] Performing curve calculation on the core hardware equipment of the driverless vehicle according to the core hardware data of the vehicle to obtain the performance data of the driverless vehicle;

[0109] Performing multi-matrix fusion on the vehicle behavior data to obtain behavior fusion data of the driverless vehicle;

[0110] The performance data of the driverless vehicle is determined according to the efficiency data and the behavior fusion data.

[0111] In one embodiment, the encoding conversion module 103, when performing encoding conversion on the customer evaluation data to obtain the evaluation encoding data, is used to:

[0112] Dividing the customer evaluation data into evaluation data corresponding to each segment of unmanned driving service;

[0113] identifying customer sentiment corresponding to the evaluation data;

[0114] Classifying the evaluation data according to the customer sentiment to obtain evaluation data categories;

[0115] Vector encoding is performed on each evaluation in the evaluation data category to obtain evaluation encoding data.

[0116] In one embodiment, the multi-source data fusion module 104, when fusing the scene coding data, the performance data and the evaluation coding data to obtain the multi-source fusion data, is used to:

[0117] Extracting scene features corresponding to the scene coding data, extracting performance features corresponding to the performance data, and determining evaluation features according to the evaluation coding data;

[0118] Spatially aligning the scene feature, the performance feature, and the evaluation feature to obtain a spatial alignment feature;

[0119] The spatial alignment features are fused using a preset Transformer network unit to obtain multi-source fused data.

[0120] In one embodiment, the risk regression coefficient calculation module 105, when analyzing the multi-source fusion data using a pre-built risk assessment model to obtain the risk regression coefficient corresponding to each driving service of the driverless vehicle, is used to:

[0121] Dividing the multi-source fusion data to obtain fusion data corresponding to each driving service segment;

[0122] Extracting fusion features corresponding to the fusion data one by one;

[0123] Calculating the risk score corresponding to the fusion feature using the risk assessment model;

[0124] The risk regression coefficient corresponding to each driving service segment is extracted according to the risk score.

[0125] In one embodiment, the auto insurance pricing premium determination module 106, when determining the auto insurance pricing premium of the driverless car according to the pre-acquired base premium, the modified rate adjustment coefficient and the risk regression coefficient, is used to:

[0126] Multiplying the base premium, the modified rate adjustment coefficient and the risk regression coefficient to obtain an initial pricing premium for the driverless car;

[0127] Determine travel risk based on multi-source fusion data corresponding to each driving service of the driverless car;

[0128] The initial pricing premium is dynamically adjusted according to the travel risk to obtain a pricing premium for the auto insurance of the driverless car.

[0129] The present invention provides a vehicle insurance premium pricing device based on data fusion. Based on the data from multiple sources outside and within the unmanned vehicle terminal, the device uses a Transformer network mechanism to perform heterogeneous data fusion processing, and then combines it with a deep learning network for training and analysis to obtain the future travel risk of the unmanned service vehicle. Based on the travel risk, the specific service city environment of the unmanned vehicle is considered, and the specific data of each trip is fully utilized for evaluation. The premium pricing model based on risk assessment takes into account both the vehicle and service characteristics of the unmanned vehicle and the dynamic personalized pricing characteristics of a single vehicle based on actual operating risks.

[0130] For the specific definition of the auto insurance premium pricing device based on data fusion, please refer to the definition of the intelligent question and answer method above, which will not be repeated here. Each module in the above-mentioned auto insurance premium pricing device based on data fusion can be implemented in whole or in part through software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0131] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 6 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. Among them, 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 and 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 an external client through a network connection. When the computer program is executed by the processor, it realizes the functions or steps on the service side of a motor insurance premium pricing method based on data fusion.

[0132] In one embodiment, a computer device is provided. The computer device may be a client, and its internal structure diagram may be as follows: Figure 7 As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, 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 an external server through a network connection. When the computer program is executed by the processor, it realizes the functions or steps on the client side of a method for pricing automobile insurance premiums based on data fusion

[0133] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program:

[0134] Acquire external environment data, vehicle attribute data, and customer evaluation data of the driverless vehicle, and encode the driving scene of the driverless vehicle according to the external environment data to obtain scene encoding data;

[0135] Calculating performance data of the driverless vehicle based on the vehicle attribute data;

[0136] Performing encoding conversion on the customer evaluation data to obtain evaluation encoding data;

[0137] fusing the scene coding data, the performance data and the evaluation coding data to obtain multi-source fused data;

[0138] Analyze the multi-source fusion data using a pre-built risk assessment model to obtain a risk regression coefficient corresponding to each driving service of the driverless vehicle;

[0139] The car insurance pricing premium for the driverless car is determined based on the pre-acquired base premium, the modified rate adjustment coefficient and the risk regression coefficient.

[0140] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0141] Acquire external environment data, vehicle attribute data, and customer evaluation data of the driverless vehicle, and encode the driving scene of the driverless vehicle according to the external environment data to obtain scene encoding data;

[0142] Calculating performance data of the driverless vehicle based on the vehicle attribute data;

[0143] Performing encoding conversion on the customer evaluation data to obtain evaluation encoding data;

[0144] fusing the scene coding data, the performance data and the evaluation coding data to obtain multi-source fused data;

[0145] Analyze the multi-source fusion data using a pre-built risk assessment model to obtain a risk regression coefficient corresponding to each driving service of the driverless vehicle;

[0146] The car insurance pricing premium for the driverless car is determined based on the pre-acquired base premium, the modified rate adjustment coefficient and the risk regression coefficient.

[0147] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can refer to the relevant descriptions on the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0148] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0149] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0150] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit it. If software tools or components other than those of the company appear in the application embodiments, they are only used for illustrative purposes and do not represent actual use. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person of ordinary skill in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A method for pricing automobile insurance premiums based on data fusion, characterized in that: include: Acquire external environment data, vehicle attribute data, and customer evaluation data of the driverless vehicle, and encode the driving scene of the driverless vehicle according to the external environment data to obtain scene encoding data; Calculating performance data of the driverless vehicle based on the vehicle attribute data; Performing encoding conversion on the customer evaluation data to obtain evaluation encoding data; fusing the scene coding data, the performance data and the evaluation coding data to obtain multi-source fused data; Analyze the multi-source fusion data using a pre-built risk assessment model to obtain a risk regression coefficient corresponding to each driving service of the driverless vehicle; The car insurance pricing premium for the driverless car is determined based on the pre-acquired base premium, the modified rate adjustment coefficient and the risk regression coefficient.

2. The method for pricing automobile insurance premiums based on data fusion according to claim 1, characterized in that: The step of encoding the driving scene of the driverless vehicle according to the external environment data to obtain scene encoding data includes: Extracting environmental features from the external environmental data; identifying a driving scenario of the driverless vehicle according to the environmental features; Dividing scene features in the driving scene to obtain scene category features; The scene category features are feature spliced ​​to obtain scene coding data corresponding to the driving scene.

3. The method for pricing automobile insurance premiums based on data fusion according to claim 1, characterized in that: The calculating the performance data of the driverless vehicle according to the vehicle attribute data includes: Dividing the vehicle attribute data into vehicle core hardware data and vehicle behavior data; Performing curve calculation on the core hardware equipment of the driverless vehicle according to the core hardware data of the vehicle to obtain the performance data of the driverless vehicle; Performing multi-matrix fusion on the vehicle behavior data to obtain behavior fusion data of the driverless vehicle; The performance data of the driverless vehicle is determined according to the efficiency data and the behavior fusion data.

4. The method for pricing automobile insurance premiums based on data fusion according to claim 1, characterized in that: The step of converting the customer evaluation data into code to obtain evaluation code data includes: Dividing the customer evaluation data into evaluation data corresponding to each segment of unmanned driving service; Identifying customer emotions corresponding to the evaluation data, and classifying the evaluation data according to the customer emotions to obtain evaluation data categories; Vector encoding is performed on each evaluation in the evaluation data category to obtain evaluation encoding data.

5. The method for pricing automobile insurance premiums based on data fusion according to claim 1, characterized in that: The step of fusing the scene coding data, the performance data and the evaluation coding data to obtain multi-source fused data includes: Extracting scene features corresponding to the scene coding data, extracting performance features corresponding to the performance data, and determining evaluation features according to the evaluation coding data; Spatially aligning the scene feature, the performance feature, and the evaluation feature to obtain a spatial alignment feature; The spatial alignment features are fused using a preset Transformer network unit to obtain multi-source fused data.

6. The method for pricing automobile insurance premiums based on data fusion according to claim 1, characterized in that: The multi-source fusion data is analyzed by using the pre-built risk assessment model to obtain the risk regression coefficient corresponding to each driving service of the driverless car, including: Dividing the multi-source fusion data to obtain fusion data corresponding to each driving service segment; Extracting fusion features corresponding to the fusion data one by one; Calculating the risk score corresponding to the fusion feature using the risk assessment model; The risk regression coefficient corresponding to each driving service segment is extracted according to the risk score.

7. The method for pricing automobile insurance premiums based on data fusion according to claim 1, characterized in that: The method of determining the auto insurance pricing premium for the driverless car based on the pre-acquired base premium, the modified rate adjustment coefficient and the risk regression coefficient includes: Multiplying the base premium, the modified rate adjustment coefficient and the risk regression coefficient to obtain an initial pricing premium for the driverless car; Determine travel risk based on multi-source fusion data corresponding to each driving service of the driverless car; The initial pricing premium is dynamically adjusted according to the travel risk to obtain a pricing premium for the auto insurance of the driverless car.

8. A car insurance premium pricing device based on data fusion, characterized in that: include: A driving scene encoding module, used to obtain external environment data, vehicle attribute data and customer evaluation data of the driverless car, and encode the driving scene of the driverless car according to the external environment data to obtain scene encoding data; A performance data calculation module, used to calculate the performance data of the driverless vehicle according to the vehicle attribute data; A coding conversion module, used for coding the customer evaluation data to obtain evaluation coding data; A multi-source data fusion module, used for fusing the scene coding data, the performance data and the evaluation coding data to obtain multi-source fusion data; A risk regression coefficient calculation module is used to analyze the multi-source fusion data using a pre-built risk assessment model to obtain a risk regression coefficient corresponding to each driving service of the driverless vehicle; The motor vehicle insurance pricing premium determination module is used to determine the motor vehicle insurance pricing premium of the driverless car based on the pre-acquired benchmark premium, the modified rate adjustment coefficient and the risk regression coefficient.

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, the steps of the method for pricing automobile insurance premium based on data fusion as claimed in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for pricing automobile insurance premium based on data fusion as claimed in any one of claims 1 to 7 are implemented.

Images