Battery usage behavior-based insurance pricing method, device, equipment and medium

Abstract

The present application belongs to the technical field of Internet of Vehicles insurance, and provides an insurance pricing method, device, equipment and medium based on battery usage behavior to solve the problem that the existing technology does not consider battery usage behavior and health status, resulting in deviation between battery compensation and actual situation, and the need to bear a large compensation risk. The method comprises: constructing a first data set and a second data set according to vehicle records, wherein the first data set comprises battery usage behavior data, and the second data set comprises battery attenuation identifiers; obtaining a target data set according to the first data set and the second data set; constructing a battery attenuation occurrence probability prediction model according to the target data set; obtaining the battery attenuation occurrence probability of a target vehicle according to the battery attenuation occurrence probability prediction model; and obtaining the risk cost of the target vehicle according to the battery attenuation occurrence probability. The present application predicts the battery attenuation occurrence probability based on the battery usage behavior, and helps insurance companies to accurately price insurance.

Description

Technical Field

[0001] This invention relates to the field of vehicle-to-everything (V2X) insurance technology, and in particular to an insurance pricing method, apparatus, device, and medium based on battery usage behavior. Background Technology

[0002] With increasing global emphasis on environmental protection and sustainable development, new energy vehicles (NEVs) have become an important development direction in the modern transportation sector. NEVs include various types, such as pure electric vehicles, range-extended electric vehicles, hybrid electric vehicles, fuel cell electric vehicles, and hydrogen engine vehicles. Among these, pure electric vehicles rely entirely on their battery systems as their primary power source. For other types of NEVs, the performance, lifespan, and health of the battery also have a significant impact on the overall performance and lifespan of the vehicle.

[0003] In the auto insurance sector, traditional pricing methods primarily rely on factors such as vehicle usage, driver experience, and vehicle damage history. However, for new energy vehicles, especially electric vehicles, battery health and degradation rates have become a new and crucial risk factor. Battery degradation can not only reduce a vehicle's driving range but also affect its overall performance and safety.

[0004] Currently, most insurance companies estimate battery compensation costs based on empirical data when providing insurance for new energy vehicles. This method often relies on historical data and statistical methods, rather than truly considering the actual usage and health of the battery. The main problem with this approach is that it can lead to significant deviations in the estimation of battery compensation costs, resulting in low accuracy in quantifying compensation costs, thus impacting the profits of insurance companies and the insurance premiums for new energy vehicle users. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide an insurance pricing method, apparatus, device and medium based on battery usage behavior, in order to solve the problem that the prior art does not consider battery usage behavior and health status, resulting in discrepancies between battery compensation and reality, and requiring the assumption of greater compensation risk.

[0006] In a first aspect, embodiments of the present invention provide an insurance pricing method based on battery usage behavior, the method comprising:

[0007] Based on the vehicle records of new energy vehicles, a first dataset and a second dataset are constructed. The first dataset includes battery usage behavior data, which includes battery dynamic parameters and battery static parameters. The second dataset includes battery degradation indicators, whereby the battery static parameters include battery type, rated battery capacity, battery energy density, and battery nominal voltage.

[0008] Based on the first dataset and the second dataset, obtain the target dataset;

[0009] Based on the target dataset, construct a battery degradation probability prediction model;

[0010] Based on the battery degradation probability prediction model, the probability of battery degradation occurring in the target vehicle is obtained.

[0011] The risk cost of the target vehicle is obtained based on the probability of battery degradation.

[0012] Preferably, the step of constructing the first dataset and the second dataset based on the vehicle records of new energy vehicles includes:

[0013] The new energy vehicles are screened to obtain new energy vehicles that meet the first preset conditions, wherein the first preset conditions include being within the warranty conditions and having a battery capacity retention rate higher than a first threshold.

[0014] According to the warranty conditions, a number of first vehicle records of new energy vehicles are obtained, wherein the warranty conditions include a first year threshold and / or a first mileage threshold, and the difference between the first mileage data of the first vehicle record and the first mileage threshold is the smallest or the difference between the first year data and the first year threshold is the smallest.

[0015] Based on the first vehicle record, obtain the first dataset;

[0016] According to the target extended warranty conditions, several second vehicle records of new energy vehicles are obtained, wherein the target extended warranty conditions include a second year threshold and / or a second mileage threshold, and the difference between the second mileage data of the second vehicle records and the second mileage threshold is the smallest or the difference between the second year data and the second year threshold is the smallest.

[0017] Obtain the second dataset based on the second vehicle record.

[0018] Preferably, the battery dynamic parameters include battery capacity retention rate, and the target dataset includes a first target dataset, a second target dataset, and a third target dataset. The step of obtaining the target dataset based on the first dataset and the second dataset includes:

[0019] The first dataset is divided according to the first preset battery capacity, the second preset battery capacity, and the third preset battery capacity to obtain a first initial dataset, a second initial dataset, and a third initial dataset. In the first initial dataset, the battery capacity retention rate of each data point is greater than the first preset battery capacity. In the second initial dataset, the battery capacity retention rate of each data point is between the first preset battery capacity and the second preset battery capacity. In the third initial dataset, the battery capacity retention rate of each data point is between the second preset battery capacity and the third preset battery capacity. The first preset battery capacity is greater than the second preset battery capacity, and the second preset battery capacity is greater than the third preset battery capacity.

[0020] Based on the second dataset, a fourth initial dataset, a fifth initial dataset, and a sixth initial dataset are obtained. The battery degradation indicator of the fourth initial dataset includes whether the battery capacity retention rate is lower than the first preset battery capacity. The battery degradation indicator of the fifth initial dataset includes whether the battery capacity retention rate is lower than the second preset battery capacity. The battery degradation indicator of the sixth initial dataset includes whether the battery capacity retention rate is lower than the third preset battery capacity.

[0021] Based on the unique identifier of the new energy vehicle, the first initial dataset and the fourth initial dataset are combined to obtain the first target dataset, the second initial dataset and the fifth initial dataset are combined to obtain the second target dataset, and the third initial dataset and the sixth initial dataset are combined to obtain the third target dataset.

[0022] Preferably, the step of constructing a battery degradation probability prediction model based on the target dataset includes:

[0023] The target dataset is preprocessed to obtain a modeling dataset, wherein the modeling dataset includes a first modeling dataset, a second modeling dataset, and a third modeling dataset;

[0024] Based on the modeling dataset, a battery degradation probability prediction model is constructed, wherein the battery degradation probability prediction model includes a first initial prediction model, a second initial prediction model, and a third initial prediction model. The first initial prediction model, the second initial prediction model, and the third initial prediction model all include a logistic regression model, a random forest classification model, and a limit gradient boosting tree classification model.

[0025] Based on the evaluation results of the battery degradation probability prediction model, the optimal first initial prediction model, second initial prediction model, and third initial prediction model are obtained.

[0026] Based on the evaluation results, the parameters of the optimal first initial prediction model, second initial prediction model and third initial prediction model are tuned to obtain the first target prediction model, second target prediction model and third target prediction model.

[0027] Preferably, the step of preprocessing the target dataset to obtain the modeling dataset includes:

[0028] Fill in the empty values ​​in the target dataset according to the usage type of the new energy vehicle, wherein the usage type includes private car, taxi, ride-hailing, and bus;

[0029] Based on the usage type, abnormal data in the target dataset is corrected, wherein the abnormal data is data that exceeds the corresponding parameter threshold;

[0030] The modeling dataset is obtained by performing maximum and minimum value checks on the limited indicator data in the target dataset.

[0031] Preferably, the step of obtaining the optimal battery degradation probability prediction model based on the evaluation results of the battery degradation probability prediction model includes:

[0032] Based on the first actual predicted value and the first theoretical predicted value output by the logistic regression model, a first confusion matrix is ​​constructed.

[0033] Based on the second confusion matrix of the second actual predicted value and the second theoretical predicted value of the random forest tree classification model;

[0034] The third confusion matrix is ​​based on the third actual predicted value and the third theoretical predicted value output by the extreme gradient boosting tree classification model.

[0035] Based on the first confusion matrix, the second confusion matrix, and the third confusion matrix, obtain the optimal first initial prediction model, the second initial prediction model, and the third initial prediction model.

[0036] Preferably, the step of obtaining the probability of battery degradation of the target vehicle based on the battery degradation probability prediction model includes:

[0037] Obtain the current battery capacity retention rate of the target vehicle;

[0038] When the current battery capacity retention rate is greater than the first preset battery capacity, based on the first target prediction model and the trained logistic regression model, the first probability of the current battery capacity retention rate of the target vehicle decaying to a level lower than the first preset battery capacity, the second probability of the current battery capacity retention rate decaying to a level lower than the second preset battery capacity, and the third probability of the current battery capacity decaying to a level lower than the third preset battery capacity are obtained.

[0039] When the current battery capacity retention rate is between the first preset battery capacity and the second preset battery capacity, according to the second target prediction model, the fourth probability of the current battery capacity retention rate of the target vehicle decaying to below the second preset battery capacity and the fifth probability of the current battery capacity decaying to the third preset battery capacity are obtained.

[0040] When the current battery capacity retention rate is between the second preset battery capacity and the third preset battery capacity, the sixth probability of battery capacity degradation of the target vehicle to below the third preset battery capacity is obtained according to the third target prediction model.

[0041] Secondly, embodiments of the present invention provide an insurance pricing device based on battery usage behavior, the device comprising:

[0042] The first dataset construction module is used to construct a first dataset and a second dataset based on the vehicle records of new energy vehicles. The first dataset includes battery usage behavior data, which includes battery dynamic parameters and battery static parameters. The second dataset includes battery degradation indicators, whereby the battery static parameters include battery type, rated battery capacity, battery energy density, and battery nominal voltage.

[0043] The second dataset acquisition module is used to acquire the target dataset based on the first dataset and the second dataset;

[0044] The model building module is used to build a battery degradation probability prediction model based on the target dataset.

[0045] The degradation occurrence rate acquisition module is used to acquire the battery degradation occurrence probability of the target vehicle based on the battery degradation occurrence probability prediction model.

[0046] The risk cost calculation module is used to obtain the risk cost of the target vehicle based on the probability of battery degradation.

[0047] Thirdly, embodiments of the present invention provide an electronic device, including: at least one processor, at least one memory, and computer program instructions stored in the memory, which, when executed by the processor, implement the method of the first aspect described above.

[0048] Fourthly, embodiments of the present invention provide a storage medium storing computer program instructions, which, when executed by a processor, implement the method of the first aspect described above.

[0049] In summary, the beneficial effects of the present invention are as follows:

[0050] This invention provides an insurance pricing method based on battery usage behavior. It constructs a first dataset and a second dataset based on vehicle records of new energy vehicles. The first dataset includes battery usage behavior data, including dynamic and static parameters. The second dataset includes battery degradation indicators. The dynamic and static parameters provide the model with information on the battery's performance and health status under different usage conditions. This helps to more accurately predict battery degradation, thereby improving the accuracy of risk assessment. The battery degradation indicators provide the model with direct indicators of battery health status. This helps the model to more accurately understand and predict the battery's future performance, thus improving the accuracy of risk assessment. Based on the first and second datasets, a target dataset is obtained. By integrating the two datasets, this step ensures the integrity and consistency of the model training data. This helps the model better understand the relationship between battery usage behavior and its degradation, thereby improving prediction accuracy. Based on the target dataset, a battery degradation probability prediction model is constructed, ensuring that the model is trained on the most relevant and accurate data. This helps improve the model's predictive accuracy and robustness; based on the battery degradation probability prediction model, the probability of battery degradation of the target vehicle is obtained, and based on the probability of battery degradation, the risk cost of the target vehicle is obtained. By estimating the risk cost through specific and quantifiable battery degradation risk indicators, insurance companies can price more accurately. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments of the present invention will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, and these are all within the protection scope of the present invention.

[0052] Figure 1 This is a flowchart illustrating the insurance pricing method based on battery usage behavior according to an embodiment of the present invention.

[0053] Figure 2 This is a schematic diagram of the process for obtaining the target dataset according to an embodiment of the present invention.

[0054] Figure 3 This is a schematic diagram of the structure of an insurance pricing device based on battery usage behavior according to an embodiment of the present invention.

[0055] Figure 4 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0056] The features and exemplary embodiments of various aspects of the present invention will now be described in detail. To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only configured to explain the present invention and are not configured to limit the present invention. For those skilled in the art, the present invention can be practiced without some of these specific details. The following description of the embodiments is merely intended to provide a better understanding of the present invention by illustrating examples of the invention.

[0057] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0058] Please see Figure 1 This invention provides an insurance pricing method based on battery usage behavior, the method comprising:

[0059] S1. Based on the vehicle records of new energy vehicles, construct a first dataset and a second dataset, wherein the first dataset includes battery usage behavior data, which includes battery dynamic parameters and battery static parameters; the second dataset includes battery degradation indicators, wherein the battery static parameters include battery type, rated battery capacity, battery energy density, and battery nominal voltage.

[0060] During the operation of new energy vehicles, the onboard system records a large amount of data related to battery performance and health. To more accurately assess the risk of battery degradation, it is necessary to extract and integrate relevant parameters from this data. This step aims to construct two datasets: the first dataset focuses on battery usage behavior, while the second dataset focuses on battery degradation. Battery behavior data, including dynamic and static parameters, is crucial for predicting battery performance degradation and lifespan. This is because battery performance and lifespan are not only related to their inherent physical and chemical properties but also closely related to their behavioral patterns during actual use.

[0061] Specifically, the battery dynamic parameters refer to parameters that change with battery use. These parameters include charge cycles, battery capacity retention, battery temperature, and state of charge. Charge cycles reflect the number of charge-discharge cycles the battery has undergone; frequent charge-discharge cycles accelerate battery aging and lead to capacity reduction. Battery capacity retention represents the ratio of the battery's current capacity to its initial capacity; a decrease in capacity retention indicates a decline in the battery's energy storage capacity, which is an important indicator of battery aging. Battery temperature reflects the temperature state of the battery during operation. Temperature directly affects the battery's chemical reactions; extreme temperatures accelerate battery aging and affect performance and safety.

[0062] The battery static parameters include battery type, rated battery capacity, battery energy density, and battery nominal voltage. These parameters determine the basic performance and application range of the battery. The battery degradation indicator indicates whether the battery capacity has degraded. When the battery capacity is lower than a preset capacity threshold, the battery degradation indicator is marked as 1; otherwise, it is marked as 0. The preset capacity threshold is set according to the actual situation. In a specific embodiment, the preset capacity threshold is 85%.

[0063] Without considering battery behavior data, predictive models may ignore changes in battery state and the impact of environmental factors during actual use, leading to inaccurate predictions and causing confusion for insurance companies and owners of new energy vehicles. For example, suppose there are two electric vehicles with identical battery static parameters, but one is mainly driven for short distances in the city with frequent start-stop operations and high charging frequency; the other is mainly used for long-distance driving with fewer charging frequency but larger charging amounts per charge. The battery behavior data of these two vehicles differ significantly. Therefore, even with identical static parameters, their battery performance degradation and lifespan will differ significantly. Without considering this behavioral data, it is impossible to accurately predict the battery health status of each vehicle and thus obtain an accurate probability of battery degradation.

[0064] As an optional embodiment of the present invention, see Figure 2 The steps of constructing the first dataset and the second dataset based on the vehicle records of new energy vehicles include:

[0065] S11. Screen the new energy vehicles to obtain new energy vehicles that meet the first preset conditions, wherein the first preset conditions include being within the warranty conditions and having a battery capacity retention rate higher than a first threshold.

[0066] Among numerous new energy vehicles, not every vehicle is suitable for battery degradation risk assessment. To ensure data quality and accuracy, vehicles meeting specific conditions need to be selected. This selection process excludes data that might interfere with model predictions, such as vehicles still under warranty or those with significantly degraded battery capacity retention. This ensures dataset quality and improves model prediction accuracy. Warranty conditions include time or mileage limits. Most automakers typically offer warranties of 3-8 years for their electric vehicles, depending on the brand and model. Besides time limits, warranty conditions may also include mileage limits; for example, some automakers may offer a 5-year or 100,000-kilometer warranty, whichever comes first. In another embodiment, battery health and capacity degradation are key considerations for electric vehicles. Therefore, many manufacturers provide specific capacity guarantees for batteries, such as ensuring that the battery capacity does not fall below 70% or 80% of its original capacity during the warranty period. The first threshold is set according to actual conditions; in one specific embodiment, the first threshold is 85%.

[0067] S12. According to the warranty conditions, obtain a number of first vehicle records of new energy vehicles, wherein the warranty conditions include a first year threshold and / or a first mileage threshold, and the difference between the first mileage data of the first vehicle record and the first mileage threshold is the smallest or the difference between the first year data and the first year threshold is the smallest.

[0068] To assess the battery's performance and health status during the warranty period, it is necessary to obtain records for each vehicle as it approaches the warranty period. This ensures that the data reflects the battery's actual usage and health status during the warranty period, providing a more accurate and realistic data basis for assessing the risk of battery degradation.

[0069] The first vehicle record closest to the warranty conditions refers to the record among all vehicle records where the usage status or parameters of a particular vehicle are closest to the definition of the warranty conditions. The warranty conditions include a first year threshold and / or a first mileage threshold. If the warranty conditions are based on time, such as "3 years", then the closest vehicle record may be the record when the vehicle is close to 3 years after purchase. If the warranty conditions are based on mileage, such as "100,000 kilometers", then the closest vehicle record may be the record when the vehicle has traveled close to 100,000 kilometers.

[0070] S13. Obtain the first dataset based on the first vehicle record;

[0071] Based on the selected vehicle records, a first dataset is constructed, which includes multiple first vehicle records, each corresponding to a new energy vehicle. This dataset mainly focuses on battery usage behavior, as battery usage behavior is one of the main factors affecting its degradation, and therefore requires detailed recording and analysis.

[0072] S14. According to the target extended warranty conditions, obtain several second vehicle records of the new energy vehicle, wherein the target extended warranty conditions include a second year threshold and / or a second mileage threshold, and the difference between the second mileage data of the second vehicle record and the second mileage threshold is the smallest or the difference between the second year data and the second year threshold is the smallest.

[0073] To assess the battery's performance and health status during the extended warranty period, it is necessary to obtain records for each vehicle as it approaches the extended warranty conditions. This ensures that the data reflects the battery's actual usage and health status during the extended warranty period, providing a more accurate and reliable data basis for assessing the risk of battery degradation.

[0074] Similar to the warranty conditions, the extended warranty conditions also include time and / or mileage limits. However, the time or mileage limits included in the extended warranty conditions differ from those in the warranty conditions. The second vehicle record closest to the extended warranty conditions refers to the record among all vehicle records where the vehicle's usage status or parameters are closest to the definition of the extended warranty conditions. If the warranty conditions are based on time, such as "4 years," then the closest vehicle record might be the record when the vehicle is close to 4 years old after purchase. If the warranty conditions are based on mileage, such as "150,000 kilometers," then the closest vehicle record might be the record when the vehicle has traveled close to 150,000 kilometers.

[0075] S15. Obtain the second dataset based on the second vehicle record.

[0076] Based on the selected vehicle records, a second dataset is constructed, which includes multiple second vehicle records. Each first vehicle record corresponds to a new energy vehicle. This dataset mainly focuses on battery degradation, which is a key indicator for assessing risk. Therefore, detailed recording and analysis are required to provide detailed and accurate data for subsequent model training and ensure the accuracy of model predictions.

[0077] S2. Obtain the target dataset based on the first dataset and the second dataset;

[0078] By integrating the first dataset (battery usage behavior data) and the second dataset (battery degradation identifiers) into a unified target dataset, the target dataset will provide complete and structured data for subsequent battery degradation probability prediction models. Merging the two datasets into a single target dataset ensures that we have complete information about each vehicle, including its battery usage behavior and battery degradation status. Having a unified dataset makes subsequent data analysis and model training simpler and more direct, eliminating the need to switch between multiple data sources.

[0079] As an optional embodiment of the present invention, the battery dynamic parameters include battery capacity retention rate, the target dataset includes a first target dataset, a second target dataset, and a third target dataset, and the step of obtaining the target dataset based on the first dataset and the second dataset includes:

[0080] S21. Divide the first dataset according to the first preset battery capacity, the second preset battery capacity, and the third preset battery capacity to obtain a first initial dataset, a second initial dataset, and a third initial dataset. In the first initial dataset, the battery capacity retention rate of each data point is greater than the first preset battery capacity. In the second initial dataset, the battery capacity retention rate of each data point is between the first preset battery capacity and the second preset battery capacity. In the third initial dataset, the battery capacity retention rate of each data point is between the second preset battery capacity and the third preset battery capacity. The first preset battery capacity is greater than the second preset battery capacity, and the second preset battery capacity is greater than the third preset battery capacity.

[0081] In this step, in order to better predict the performance and degradation of new energy vehicle batteries under different capacity thresholds, the first dataset is divided according to different preset battery capacities. The first preset battery capacity is greater than the second preset battery capacity, the second preset battery capacity is greater than the third preset battery capacity. In a specific embodiment, the first preset battery capacity is 95%, the second preset battery capacity is 90%, and the third preset battery capacity is 85%. The battery capacity retention rate in the first initial dataset is greater than 95%, the battery capacity retention rate in the second initial dataset is between 90% and 95%, and the battery capacity retention rate in the third initial dataset is between 85% and 90%.

[0082] S22. Based on the second dataset, obtain a fourth initial dataset, a fifth initial dataset, and a sixth initial dataset, wherein the battery degradation indicator of the fourth initial dataset includes whether the battery capacity retention rate is lower than the first preset battery capacity, the battery degradation indicator of the fifth initial dataset includes whether the battery capacity retention rate is lower than the second preset battery capacity, and the battery degradation indicator of the sixth initial dataset includes whether the battery capacity retention rate is lower than the third preset battery capacity.

[0083] In this step, the second dataset is used to generate three new initial datasets: the fourth initial dataset, the fifth initial dataset, and the sixth initial dataset. The battery degradation indicator of the fourth initial dataset includes whether the battery capacity retention rate is lower than the first preset battery capacity. The battery degradation indicator of the fifth initial dataset includes whether the battery capacity retention rate is lower than the second preset battery capacity. The battery degradation indicator of the sixth initial dataset includes whether the battery capacity retention rate is lower than the third preset battery capacity. The first, second, and third initial datasets are features or input variables of the model. The fourth, fifth, and sixth initial datasets are target variables or labels of the prediction model, representing the result that the prediction model wants to predict. The features or input variables are the data used to predict the target variables or labels. In this embodiment, the dynamic and static parameters of the battery are used as features to predict the battery degradation.

[0084] S23. Based on the unique identifier of the new energy vehicle, the first initial dataset and the fourth initial dataset are combined to obtain the first target dataset, the second initial dataset and the fifth initial dataset are combined to obtain the second target dataset, and the third initial dataset and the sixth initial dataset are combined to obtain the third target dataset.

[0085] In the above steps, different initial datasets are combined according to the unique identifier of new energy vehicles to obtain different target datasets. The purpose is to combine battery usage behavior data divided according to battery capacity retention rate with battery degradation identifier in order to build a battery degradation probability prediction model.

[0086] Specifically, using the unique identifier of new energy vehicles, such as VIN number or other proprietary identifier code, as a key field, the first initial dataset and the fourth initial dataset are combined to obtain the first target dataset, the second initial dataset and the fifth initial dataset are combined to obtain the second target dataset, and the third initial dataset and the sixth initial dataset are combined to obtain the third target dataset, thus constructing three complete target datasets. Each new energy vehicle has a unique identifier, which ensures the accuracy of data matching.

[0087] For example, and not limited to, the first initial dataset includes battery behavior data of vehicle A and vehicle B obtained through the first vehicle record;

[0088] Battery behavior data for vehicle A: Unique identifier A, battery capacity retention rate 95%, battery type 1, rated battery capacity 100Ah;

[0089] Battery behavior data for vehicle B: Unique identifier B, battery capacity retention rate 92%, battery type 2, rated battery capacity 120A;

[0090] The fourth initial dataset includes battery degradation indicators obtained from the second vehicle record;

[0091] Battery degradation indicator for vehicle A: Unique identifier A, whether the battery capacity retention rate is lower than the first preset battery capacity (assuming 95%) - No;

[0092] Battery degradation indicator for vehicle B: Is the battery capacity retention rate lower than the first preset battery capacity (assuming 95%)? - Yes;

[0093] By using the unique identifiers of the vehicles, the first initial dataset and the fourth initial dataset can be matched and combined to obtain the following first target dataset:

[0094] Vehicle A: Unique identifier A, battery capacity retention rate 95%, battery type 1, rated battery capacity 100Ah, battery degradation indicator - no.

[0095] Vehicle B: Unique identifier B, battery capacity retention rate 92%, battery type 2, rated battery capacity 120Ah, battery degradation indicator - Yes;

[0096] Using unique identifiers ensures that the information extracted from both datasets is about the same vehicle. In the specific data matching process, database queries or data processing tools are typically used to merge the battery usage behavior data in the first dataset with the battery degradation identifier in the second dataset based on the unique identifier of each vehicle. Accurately matched data provides a solid foundation for subsequent analysis and model training, thereby improving the accuracy of the entire analysis process.

[0097] S3. Based on the target dataset, construct a battery degradation probability prediction model;

[0098] In this step, the integrated target dataset is used to build a predictive model of the probability of battery degradation. This model aims to predict whether the battery will degrade at some point in the future or during the usage phase. Building a predictive model of the probability of battery degradation can help predict the battery's lifespan and maintenance costs more accurately. Through the predictive model, insurance companies or car owners can manage the risk of battery degradation more effectively, for example, by adjusting insurance premiums or taking preventative maintenance.

[0099] As an optional embodiment of the present invention, the step of constructing a battery degradation probability prediction model based on the target dataset includes:

[0100] S31. The target dataset is preprocessed to obtain a modeling dataset, wherein the modeling dataset includes a first modeling dataset, a second modeling dataset, and a third modeling dataset;

[0101] Preprocessing is a crucial step in data analysis, ensuring the quality and applicability of the dataset. This includes handling missing values, outliers, data standardization, and encoding categorical variables. Preprocessing can improve the accuracy and stability of the model, ensuring that the model is trained on high-quality data. In this step, the first target dataset, the second target dataset, and the third target dataset are preprocessed respectively to obtain the first modeling dataset, the second modeling dataset, and the third modeling dataset.

[0102] As an optional embodiment of the present invention, the step of preprocessing the target dataset to obtain the modeling dataset includes:

[0103] S311. Fill in the empty values ​​in the target dataset according to the usage type of the new energy vehicle;

[0104] Specifically, depending on the vehicle's usage type, such as private passenger cars or buses, specific strategies are used to fill in missing values. Continuous variables are filled with the median, while categorical variables are filled with the mode. For example, rather than limiting it, for private passenger cars, if annualized mileage data is missing, 8000km is used as the filler value; if power type data is missing, "pure electric" is used as the filler value. Missing values ​​may lead to instability in model training. Using the median and mode for filling is a common strategy because it does not have a significant impact on the data distribution.

[0105] S312. According to the usage type, correct the abnormal data in the target dataset, wherein the abnormal data is data that exceeds the corresponding parameter threshold;

[0106] Abnormal data may be caused by data entry errors or other reasons. This data may have an adverse effect on model training. In order to ensure that all data in the dataset is within a reasonable range, thereby improving the accuracy of the model, data that exceeds a predetermined threshold is identified and corrected. For example, for certain percentage data, their values ​​should be between 0 and 1. If the percentage of a data point is greater than 1, it is corrected to 1; if it is less than 0, it is corrected to 0.

[0107] S313. Perform maximum and minimum value verification on the limited indicator data in the target dataset to obtain the modeling dataset.

[0108] Specifically, extreme values ​​can distort the distribution of data, thus affecting model performance. The Laida criterion is a common method for identifying and handling these extreme values. To remove or correct extreme values ​​in a dataset and ensure a normal data distribution, the Laida criterion is used to check for extreme values. For normally or approximately normally distributed data, the data should be within the range (μ-3σ, μ+3σ), where μ is the mean and σ is the standard deviation. If a data point's value exceeds the range (μ-3σ, μ+3σ), it is considered an extreme value and is removed or corrected.

[0109] S32. Based on the modeling dataset, construct the battery degradation probability prediction model, wherein the battery degradation probability prediction model includes a first initial prediction model, a second initial prediction model, and a third initial prediction model, and the first initial prediction model, the second initial prediction model, and the third initial prediction model all include a logistic regression model, a random forest classification model, and a limit gradient boosting tree classification model.

[0110] Specifically, in this step, various machine learning algorithms such as logistic regression, random forest, and extreme gradient boosting tree are used to construct prediction models. Specifically, the battery degradation probability prediction model is trained using the first, second, and third modeling datasets respectively, resulting in a first initial prediction model, a second initial prediction model, and a third initial prediction model. Each of these initial prediction models includes a logistic regression model, a random forest classification model, and an extreme gradient boosting tree classification model. Each model has its own characteristics: logistic regression is a statistical method used to predict the probability of binary classification problems; random forest classification is based on an ensemble method of decision trees, classifying data through voting from multiple decision trees; and extreme gradient boosting tree classification is an efficient gradient boosting algorithm used for both classification and regression tasks.

[0111] The first initial prediction model trained on the first modeling dataset is used to predict the probability that a target vehicle with a current battery capacity retention rate greater than a first preset battery capacity is lower than the first preset battery capacity. The second initial prediction model trained on the second modeling dataset is used to predict the probability that a target vehicle with a current battery capacity retention rate between the first preset battery capacity and the second preset battery capacity is lower than the second preset battery capacity. The third initial prediction model trained on the third modeling dataset is used to predict the probability that a target vehicle with a current battery capacity retention rate between the second preset battery capacity and the third preset battery capacity is lower than the third preset battery capacity.

[0112] S33. Based on the evaluation results of the battery degradation probability prediction model, obtain the optimal first initial prediction model, second initial prediction model, and third initial prediction model.

[0113] In this step, evaluation metrics are used to assess the performance of each of the first, second, and third initial prediction models, and the model with the best performance is selected as the optimal model. These evaluation metrics include accuracy, recall, and F1 score. Selecting the optimal model ensures the accuracy and stability of the predictions and provides reliable data support for subsequent decision-making.

[0114] As an optional embodiment of the present invention, the step of determining the optimal first initial prediction model, second initial prediction model, and third initial prediction model based on the evaluation results of the battery degradation probability prediction model includes:

[0115] S331. Construct a first confusion matrix based on the first actual predicted value and the first theoretical predicted value output by the logistic regression model;

[0116] S332. A second confusion matrix based on the second actual predicted value and the second theoretical predicted value of the random forest tree classification model;

[0117] S333, The third confusion matrix between the third actual predicted value and the third theoretical predicted value output by the extreme gradient boosting tree classification model;

[0118] Specifically, the confusion matrix is ​​a table used to describe the correctness and error of a model's predictions. It includes true positives, false positives, true negatives, and false negatives; the actual predicted values ​​are the predictions made by the model based on the input data, while the theoretical predicted values ​​are the true labels or results; the confusion matrix provides an intuitive way to evaluate the model's performance, especially its precision, recall, and F1 score.

[0119] S334. Based on the first confusion matrix, the second confusion matrix, and the third confusion matrix, obtain the optimal first initial prediction model, the second initial prediction model, and the third initial prediction model.

[0120] Confusion matrices provide a visual method for evaluating the performance of classification models. To obtain the optimal battery degradation probability prediction model based on the confusion matrix, the following key metrics can be calculated from each confusion matrix: precision (the ratio of correctly predicted samples to the total number of samples); recall (the ratio of true positives to actual positives); accuracy (the ratio of true positives to samples predicted as positives); and the F1 score (the harmonic mean of precision and recall). For each model, these metrics can be compared. For example, if avoiding false positives about battery degradation is a greater concern, a model with high precision might be more important. If ensuring that all possible battery degradation is captured is a greater priority, a model with high recall might be more important.

[0121] In this embodiment, confusion matrices are constructed for the logistic regression model, random forest classification model, and extreme gradient boosting tree classification model in the first initial prediction model, the second initial prediction model, and the third initial prediction model, respectively. The optimal first initial prediction model, the optimal second initial prediction model, and the optimal third initial prediction model are obtained based on the confusion matrices.

[0122] Using the optimal model ensures that the predicted results are based on the best estimate of currently available data and technology, which improves the accuracy and reliability of the predictions. Once the optimal model is determined, we do not need to consider other models. This simplifies the prediction process, making it more efficient and stable. By providing predictions of the probability of battery degradation for target vehicles, all parties (such as car owners, insurance companies, maintenance service providers, etc.) can obtain valuable information and make more informed decisions.

[0123] S34. Based on the evaluation results, the parameters of the optimal first initial prediction model, second initial prediction model and third initial prediction model are tuned to obtain the first target prediction model, second target prediction model and third target prediction model.

[0124] This step involves parameter tuning for the optimal first initial prediction model, the optimal second initial prediction model, and the optimal third initial prediction model. Parameter tuning is a key step in machine learning, with the aim of finding the optimal parameter settings for the model. Parameter tuning can be achieved through grid search or random search, as well as cross-validation to evaluate the performance of different parameter combinations.

[0125] First, it's necessary to determine which parameters are adjustable. For example, for a random forest, possible parameters include the number of trees, the maximum depth of each tree, and the number of features. These can be determined using either grid search or random search. Grid search is a brute-force approach that tries all possible combinations of parameters to find the optimal one. For instance, if you want to adjust the number of trees and the maximum depth of each tree in a random forest, you can set a range for each parameter and then try all possible combinations. Unlike grid search, random search doesn't try all possible parameter combinations; instead, it randomly selects combinations within a range of parameters. This method is generally faster than grid search and may be more efficient in some cases.

[0126] To ensure that the tuning results do not lead to overfitting, cross-validation is typically used. This means that the data is split into several parts, one part for training and another part for validation. This process is repeated multiple times, each time using a different part for validation.

[0127] After each parameter adjustment, the model's performance needs to be evaluated. This can be done by comparing the accuracy, recall, F1 score, or other metrics of different models. Once the optimal combination of parameters is determined, the model is finally validated using a separate test dataset to ensure its performance in real-world applications.

[0128] S4. Based on the battery degradation probability prediction model, obtain the battery degradation probability of the target vehicle.

[0129] The target vehicle is a specific vehicle whose battery degradation probability needs to be predicted. The battery degradation probability prediction model predicts the rate of degradation based on input data, including battery dynamic and static parameters. Using this prediction model, relevant data for the target vehicle is input, and the model outputs a predicted value representing the likelihood or probability of battery degradation. Each vehicle's usage, driving habits, and maintenance conditions may differ, therefore, the battery degradation situation may also vary from vehicle to vehicle. Through this step, personalized battery degradation probability predictions can be provided for each target vehicle. For insurance companies or vehicle maintenance service providers, understanding the risk of battery degradation helps them better manage risk, such as developing more appropriate insurance pricing strategies or providing preventative maintenance services. For car owners, understanding the battery degradation risk of their vehicles helps them better maintain and use their vehicles, thereby extending battery life and improving vehicle performance.

[0130] As an optional embodiment of the present invention, the step of obtaining the probability of battery degradation of the target vehicle based on the battery degradation probability prediction model includes:

[0131] S41. Obtain the current battery capacity retention rate of the target vehicle;

[0132] First, obtain the current battery capacity retention rate of the target vehicle. The current battery capacity retention rate is the key to determining which prediction model to use. Different battery capacity ranges may be affected by different influencing factors, so different models are needed for prediction.

[0133] S42. When the current battery capacity retention rate is greater than the first preset battery capacity, according to the first target prediction model and the trained logistic regression model, obtain the first probability of the current battery capacity retention rate of the target vehicle decaying to a level lower than the first preset battery capacity, the second probability of the current battery capacity retention rate decaying to a level lower than the second preset battery capacity, and the third probability of the current battery capacity decaying to a level lower than the third preset battery capacity.

[0134] When the current battery capacity retention rate is greater than the first preset battery capacity, a first target prediction model is used for prediction. First, the static and dynamic parameters of the target vehicle are input into the first target prediction model to obtain a first probability of the current battery capacity retention rate decaying to below the first preset battery capacity. Then, the first probability of decay is input into a trained first logistic regression model to obtain a second probability of decay. The second probability of decay is equal to the first probability of decay multiplied by the model coefficient of the first logistic regression model. Similarly, the second probability of decay is input into a trained second logistic regression model to obtain a third probability of decay. The third probability of decay is equal to the second probability of decay multiplied by the model coefficient of the second logistic regression model. In the logistic regression model, the model coefficients are part of the model and are used to determine the degree of influence of the input features on the prediction result.

[0135] S43. When the current battery capacity retention rate is between the first preset battery capacity and the second preset battery capacity, according to the second target prediction model, obtain the fourth probability of the current battery capacity retention rate of the target vehicle decaying to below the second preset battery capacity and the fifth probability of the current battery capacity decaying to the third preset battery capacity.

[0136] When the current battery capacity retention rate is between the first preset battery capacity and the second preset battery capacity, a second target prediction model is used for prediction. First, the static and dynamic parameters of the target vehicle are input into the second target prediction model to obtain the fourth probability of the current battery capacity retention rate decaying to below the second preset battery capacity. Then, the fourth probability of decay is input into the trained third logistic regression model to obtain the fifth probability of decay. The fifth probability of decay is equal to the fourth probability of decay multiplied by the model coefficient of the third logistic regression model.

[0137] S44. When the current battery capacity retention rate is between the second preset battery capacity and the third preset battery capacity, according to the third target prediction model, obtain the sixth probability of battery capacity degradation of the target vehicle to below the third preset battery capacity.

[0138] When the current battery capacity retention rate is between the second preset battery capacity and the third preset battery capacity, the third target prediction model is used for prediction. The static and dynamic parameters of the target vehicle are input into the third target prediction model to obtain the sixth probability of the current battery capacity retention rate decaying to below the third preset battery capacity.

[0139] Specifically, the first target prediction model, the second target prediction model, and the third target prediction model are all constructed based on datasets divided into the first dataset and the second dataset according to different thresholds. The second dataset is composed of vehicle records that are closest to the target extended warranty conditions. This means that the data in the second dataset reflects the battery degradation state of the vehicle when it is close to the extended warranty conditions. Therefore, the period of battery degradation probability predicted by the first target prediction model, the second target prediction model, and the third target prediction model should be consistent with this target extended warranty conditions.

[0140] Specifically, if the target extended warranty conditions include a time limit threshold and a mileage threshold, such as a second time limit threshold and a second mileage threshold, then the predicted period for the probability of battery degradation should be the time point when this time limit threshold or mileage threshold is reached.

[0141] For example, if the target extended warranty condition is 5 years of vehicle use or 100,000 kilometers of driving, then the predicted probability of battery degradation occurs over the period of those 5 years or 100,000 kilometers. Within this period, by analyzing information from both the first and second datasets, the model predicts the probability of battery degradation, thereby helping to determine the risk cost of the insurance.

[0142] Logistic regression estimates probabilities using a logistic function that maps any value to between 0 and 1; the sigmoid function is commonly used. The first logistic regression model takes the prediction result from the first target prediction model as input and outputs the probability that the current battery capacity retention rate has decreased to below a second preset battery capacity. The training process uses the first modeling dataset, where the dependent variable is whether the battery has decreased to the first preset battery capacity, and the independent variable is the prediction result from the first target prediction model. Similarly, the second logistic regression model takes the product of the probabilities that the current battery capacity retention rate has decreased to below the second preset battery capacity as input and outputs the probability that the current battery capacity retention rate has decreased to below a third preset battery capacity. The probability of battery degradation is calculated by taking the prediction results of the second target prediction model as input to the third model and outputting the probability that the current battery capacity retention rate will decay to below the third preset battery capacity. The training process uses the second modeling dataset, where the dependent variable is whether the battery has decayed to the third preset battery capacity, and the independent variable is the prediction results of the second target prediction model. By using the logistic regression model, battery degradation can be continuously predicted, providing users with a continuous overview of battery health. This allows for the prediction of degradation probabilities for a single vehicle at multiple battery capacity thresholds, providing decision-makers with more information and enabling them to better understand the risks of the vehicle under different usage conditions. Since the predictions of each subsequent stage are based on the prediction results of the previous stage, this ensures logical consistency in the predictions. Specifically, as the threshold decreases, the degradation rate monotonically decreases. This is because if the battery capacity has already decayed to a lower threshold, then it has naturally also decayed to a higher threshold.

[0143] By dividing battery degradation prediction into multiple stages, the model can more accurately predict the degradation probability at different capacity thresholds. Each stage's prediction considers the results of the previous stage, providing more contextual information and increasing prediction accuracy. Traditional prediction methods may result in contradictory predictions for the same vehicle at different thresholds. However, this staged prediction method avoids such conflicts because each stage's prediction considers the results of the previous stage. Insurance companies can better understand the battery degradation risk at different capacities, leading to more reasonable insurance pricing for customers. Knowing which vehicles are more likely to experience degradation at specific capacity thresholds helps insurance companies, repair shops, and manufacturers allocate resources more effectively, such as spare parts inventory, repair personnel, and other relevant resources.

[0144] For example, rather than limiting, if the first preset battery capacity is 95%, the second preset battery capacity is 90%, the third preset battery capacity is 85%, and the current battery capacity retention rate of the target vehicle is 93%, and a phased prediction method is used, since the target's battery capacity retention rate of 93% is between the first preset battery capacity of 95% and the second preset battery capacity of 90%, using the second target prediction model, the probability of its capacity decaying to below 90% is 30%, and the probability of its capacity decaying to below 85% is 10%. Based on the results of the phased analysis, a basic premium can be set for the target vehicle, and the premium can be adjusted according to the decay probability of different stages to achieve flexible pricing. If a uniform model is used for prediction, without distinguishing between different stages of battery capacity, the probability of the target vehicle's battery decaying to 85% is directly predicted to be 15%. Based on the results of the uniform analysis, only a fixed premium can be set for the target, without considering the risk of battery capacity decay at different stages. Regardless of whether the target's battery capacity drops below 90%, the premium remains unchanged; only when it drops below 85% will the premium be increased.

[0145] It can be seen that by making phased predictions, we can more accurately understand the decay risk of the target under different capacity thresholds, thereby achieving more accurate risk assessment and pricing. In contrast, the pricing scheme of uniform prediction is relatively fixed and cannot adjust the premium according to the actual decay situation at different stages, resulting in poor adaptability.

[0146] For car owners, pricing schemes obtained through phased forecasts may be fairer and more reasonable because they reflect the actual deterioration risk of the vehicle, while a uniform forecast pricing scheme may cause some car owners to pay for high-risk vehicles.

[0147] S5. Obtain the risk cost of the target vehicle based on the probability of battery degradation.

[0148] Specifically, the risk cost can be calculated using the following formula:

[0149] Risk cost = Probability of battery degradation × Base cost

[0150] The baseline cost is determined based on different repair options, including replacing battery cells and replacing the battery pack.

[0151] When the repair solution involves replacing the battery cells, the baseline cost is calculated using the following formula:

[0152] Base cost = Battery repair capacity * Cell cost per kilowatt-hour for the corresponding battery type * (1 + Risk buffer)

[0153] The battery repair capacity is (target repair battery capacity - current actual battery capacity) * system nominal energy;

[0154] When the repair solution involves replacing the battery pack, the baseline cost is calculated using the following formula:

[0155] Base cost = Battery pack capacity * Pack cost per kilowatt-hour for corresponding battery type * (1 + risk buffer)

[0156] Taking into account potential additional costs during battery repair, such as over-repair, fluctuations in battery market prices, and labor costs for battery repair, a risk buffer has been added to the calculation.

[0157] These two options offer different battery repair methods, each with its own specific cost calculation method. The choice of which method to use depends on the specific circumstances, such as the extent of battery damage, the complexity of the repair, and cost-benefit analysis. These two options provide car owners, insurance companies, and maintenance service providers with more choices and flexibility, helping them make more informed decisions.

[0158] Example 2

[0159] Please see Figure 3 The invention provides an insurance pricing device based on battery usage behavior, the device comprising:

[0160] The first dataset construction module is used to construct a first dataset and a second dataset based on the vehicle records of new energy vehicles. The first dataset includes battery usage behavior data, which includes battery dynamic parameters and battery static parameters. The second dataset includes battery degradation indicators, whereby the battery static parameters include battery type, rated battery capacity, battery energy density, and battery nominal voltage.

[0161] The second dataset acquisition module is used to acquire the target dataset based on the first dataset and the second dataset;

[0162] The model building module is used to build a battery degradation probability prediction model based on the target dataset.

[0163] The degradation occurrence rate acquisition module is used to acquire the battery degradation occurrence probability of the target vehicle based on the battery degradation occurrence probability prediction model.

[0164] The risk cost calculation module is used to obtain the risk cost of the target vehicle based on the probability of battery degradation.

[0165] It should be noted that each module and unit in the insurance pricing device based on battery usage behavior in this embodiment corresponds one-to-one with each step in the insurance pricing method based on battery usage behavior in the aforementioned embodiment. Therefore, the specific implementation of this embodiment can refer to the implementation of the aforementioned insurance pricing method based on battery usage behavior, and will not be repeated here.

[0166] In addition, combined Figure 4 The insurance pricing method based on battery usage behavior in this embodiment of the invention can be implemented by an electronic device. Figure 4 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present invention is shown.

[0167] The electronic device may include a processor 301 and a memory 302 storing computer program instructions.

[0168] Electronic devices mainly refer to devices that perform related calculations and analyses, while new energy vehicles are the source of data and the object of insurance services. Electronic devices are mainly responsible for data processing, model building and risk cost calculation, while new energy vehicles are entities that provide battery usage behavior data and battery degradation indicators.

[0169] Specifically, the processor may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement embodiments of the present invention.

[0170] The memory may include a large-capacity storage device for data or instructions. For example, and not limitingly, the memory may include a hard disk drive (HDD), a floppy disk drive, flash memory, an optical disk drive, a magneto-optical disk drive, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory may include removable or non-removable (or fixed) media. Where appropriate, the memory may be internal or external to a data processing device. In a particular embodiment, the memory is a non-volatile solid-state memory. In a particular embodiment, the memory includes a read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0171] The processor reads and executes computer program instructions stored in memory to implement any of the battery usage behavior-based insurance pricing methods in the above embodiments.

[0172] In one example, the electronic device may also include a communication interface 303 and a bus 310. For example, Figure 4 As shown, the processor, memory, and communication interface are connected via a bus and communicate with each other.

[0173] The communication interface is mainly used to enable communication between various modules, devices, units and / or equipment in the embodiments of the present invention.

[0174] A bus, including hardware, software, or both, couples components of an electronic device together. For example, and not limitingly, a bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, a bus may include one or more buses. While specific buses are described and illustrated in embodiments of the invention, the invention contemplates any suitable bus or interconnect.

[0175] Furthermore, in conjunction with the battery usage behavior-based insurance pricing method in the above embodiments, this invention can be implemented using a computer-readable storage medium. This computer-readable storage medium stores computer program instructions; when executed by a processor, these computer program instructions implement any of the battery usage behavior-based insurance pricing methods in the above embodiments.

[0176] In summary, the insurance pricing method, apparatus, device, and storage medium based on battery usage behavior provided in this invention constructs a first dataset and a second dataset based on vehicle records of new energy vehicles. The first dataset includes battery usage behavior data, which includes battery dynamic parameters and battery static parameters. The second dataset includes battery degradation indicators. The battery's dynamic and static parameters provide the model with information on the battery's performance and health status under different usage conditions. This helps to more accurately predict battery degradation, thereby improving the accuracy of risk assessment. The battery degradation indicators provide the model with direct indicators of battery health status. This helps the model to more accurately understand and predict the battery's future performance, thereby improving the accuracy of risk assessment. Based on the first and second datasets, a target dataset is obtained. By integrating the two datasets, this step ensures the integrity and consistency of the model training data. This helps the model better understand the relationship between battery usage behavior and its degradation, thereby improving prediction accuracy. Based on the target dataset, a battery degradation probability prediction model is constructed, ensuring that the model is trained based on the most relevant and accurate data. This helps improve the model's predictive accuracy and robustness; based on the battery degradation probability prediction model, the probability of battery degradation of the target vehicle is obtained, and based on the probability of battery degradation, the risk cost of the target vehicle is obtained. By estimating the risk cost through specific and quantifiable battery degradation risk indicators, insurance companies can price more accurately.

[0177] It should be clarified that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of the present invention.

[0178] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this invention are programs or code segments used to perform the required tasks. The programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0179] It should also be noted that the exemplary embodiments mentioned in this invention describe methods or systems based on a series of steps or apparatus. However, this invention is not limited to the order of the steps described above; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0180] The above description is merely a specific embodiment of the present invention. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the protection scope of the present invention.

Claims

1. An insurance pricing method based on battery usage behavior, characterized in that, The method includes: Based on the vehicle records of new energy vehicles, a first dataset and a second dataset are constructed. The first dataset includes battery usage behavior data, which includes battery dynamic parameters and battery static parameters. The second dataset includes battery degradation indicators, whereby the battery static parameters include battery type, rated battery capacity, battery energy density, and battery nominal voltage. Based on the first dataset and the second dataset, obtain the target dataset; Based on the target dataset, construct a battery degradation probability prediction model; Based on the battery degradation probability prediction model, the probability of battery degradation occurring in the target vehicle is obtained. The risk cost of the target vehicle is obtained based on the probability of battery degradation. The step of constructing the first dataset and the second dataset based on the vehicle records of new energy vehicles includes: The new energy vehicles are screened to obtain new energy vehicles that meet the first preset conditions, wherein the first preset conditions include being within the warranty conditions and having a battery capacity retention rate higher than a first threshold. According to the warranty conditions, a number of first vehicle records of new energy vehicles are obtained, wherein the warranty conditions include a first year threshold and / or a first mileage threshold, and the difference between the first mileage data of the first vehicle record and the first mileage threshold is the smallest or the difference between the first year data and the first year threshold is the smallest. Based on the first vehicle record, obtain the first dataset; According to the target extended warranty conditions, several second vehicle records of new energy vehicles are obtained, wherein the target extended warranty conditions include a second year threshold and / or a second mileage threshold, and the difference between the second mileage data of the second vehicle records and the second mileage threshold is the smallest or the difference between the second year data and the second year threshold is the smallest. Obtain the second dataset based on the second vehicle record.

2. The insurance pricing method based on battery usage behavior according to claim 1, characterized in that, The battery dynamic parameters include battery capacity retention rate, and the target dataset includes a first target dataset, a second target dataset, and a third target dataset. The step of obtaining the target dataset based on the first dataset and the second dataset includes: The first dataset is divided according to the first preset battery capacity, the second preset battery capacity, and the third preset battery capacity to obtain a first initial dataset, a second initial dataset, and a third initial dataset. In the first initial dataset, the battery capacity retention rate of each data point is greater than the first preset battery capacity. In the second initial dataset, the battery capacity retention rate of each data point is between the first preset battery capacity and the second preset battery capacity. In the third initial dataset, the battery capacity retention rate of each data point is between the second preset battery capacity and the third preset battery capacity. The first preset battery capacity is greater than the second preset battery capacity, and the second preset battery capacity is greater than the third preset battery capacity. Based on the second dataset, a fourth initial dataset, a fifth initial dataset, and a sixth initial dataset are obtained. The battery degradation indicator of the fourth initial dataset includes whether the battery capacity retention rate is lower than the first preset battery capacity. The battery degradation indicator of the fifth initial dataset includes whether the battery capacity retention rate is lower than the second preset battery capacity. The battery degradation indicator of the sixth initial dataset includes whether the battery capacity retention rate is lower than the third preset battery capacity. Based on the unique identifier of the new energy vehicle, the first initial dataset and the fourth initial dataset are combined to obtain the first target dataset, the second initial dataset and the fifth initial dataset are combined to obtain the second target dataset, and the third initial dataset and the sixth initial dataset are combined to obtain the third target dataset.

3. The insurance pricing method based on battery usage behavior according to claim 2, characterized in that, The step of constructing a battery degradation probability prediction model based on the target dataset includes: The target dataset is preprocessed to obtain a modeling dataset, wherein the modeling dataset includes a first modeling dataset, a second modeling dataset, and a third modeling dataset; Based on the modeling dataset, a battery degradation probability prediction model is constructed, wherein the battery degradation probability prediction model includes a first initial prediction model, a second initial prediction model, and a third initial prediction model. The first initial prediction model, the second initial prediction model, and the third initial prediction model all include a logistic regression model, a random forest classification model, and a limit gradient boosting tree classification model. Based on the evaluation results of the battery degradation probability prediction model, the optimal first initial prediction model, second initial prediction model, and third initial prediction model are obtained. Based on the evaluation results, the parameters of the optimal first initial prediction model, second initial prediction model and third initial prediction model are tuned to obtain the first target prediction model, second target prediction model and third target prediction model.

4. The insurance pricing method based on battery usage behavior according to claim 3, characterized in that, The step of preprocessing the target dataset to obtain the modeling dataset includes: Fill in the empty values ​​in the target dataset according to the usage type of the new energy vehicle, wherein the usage type includes private car, taxi, ride-hailing, and bus; Based on the usage type, abnormal data in the target dataset is corrected, wherein the abnormal data is data that exceeds the corresponding parameter threshold; The modeling dataset is obtained by performing maximum and minimum value checks on the limited indicator data in the target dataset.

5. The insurance pricing method based on battery usage behavior according to claim 3, characterized in that, The step of obtaining the optimal battery degradation probability prediction model based on the evaluation results of the battery degradation probability prediction model includes: Based on the first actual predicted value and the first theoretical predicted value output by the logistic regression model, a first confusion matrix is ​​constructed. Based on the second confusion matrix of the second actual predicted value and the second theoretical predicted value of the random forest classification model; The third confusion matrix is ​​based on the third actual predicted value and the third theoretical predicted value output by the extreme gradient boosting tree classification model. Based on the first confusion matrix, the second confusion matrix, and the third confusion matrix, obtain the optimal first initial prediction model, the second initial prediction model, and the third initial prediction model.

6. The insurance pricing method based on battery usage behavior according to claim 3, characterized in that, The step of obtaining the probability of battery degradation of the target vehicle based on the battery degradation probability prediction model includes: Obtain the current battery capacity retention rate of the target vehicle; When the current battery capacity retention rate is greater than the first preset battery capacity, based on the first target prediction model and the trained logistic regression model, the first probability of the current battery capacity retention rate of the target vehicle decaying to a level lower than the first preset battery capacity, the second probability of the current battery capacity retention rate decaying to a level lower than the second preset battery capacity, and the third probability of the current battery capacity decaying to a level lower than the third preset battery capacity are obtained. When the current battery capacity retention rate is between the first preset battery capacity and the second preset battery capacity, according to the second target prediction model, the fourth probability of the current battery capacity retention rate of the target vehicle decaying to below the second preset battery capacity and the fifth probability of the current battery capacity decaying to the third preset battery capacity are obtained. When the current battery capacity retention rate is between the second preset battery capacity and the third preset battery capacity, the sixth probability of battery capacity degradation of the target vehicle to below the third preset battery capacity is obtained according to the third target prediction model.

7. An insurance pricing device based on battery usage behavior, characterized in that, The device includes: The first dataset construction module is used to construct a first dataset and a second dataset based on the vehicle records of new energy vehicles. The first dataset includes battery usage behavior data, which includes battery dynamic parameters and battery static parameters. The second dataset includes battery degradation indicators, whereby the battery static parameters include battery type, rated battery capacity, battery energy density, and battery nominal voltage. The second dataset acquisition module is used to acquire the target dataset based on the first dataset and the second dataset; The model building module is used to build a battery degradation probability prediction model based on the target dataset. The degradation occurrence rate acquisition module is used to acquire the battery degradation occurrence probability of the target vehicle based on the battery degradation occurrence probability prediction model. The risk cost calculation module is used to obtain the risk cost of the target vehicle based on the probability of battery degradation. The step of constructing the first dataset and the second dataset based on the vehicle records of new energy vehicles includes: The new energy vehicles are screened to obtain new energy vehicles that meet the first preset conditions, wherein the first preset conditions include being within the warranty conditions and having a battery capacity retention rate higher than a first threshold. According to the warranty conditions, a number of first vehicle records of new energy vehicles are obtained, wherein the warranty conditions include a first year threshold and / or a first mileage threshold, and the difference between the first mileage data of the first vehicle record and the first mileage threshold is the smallest or the difference between the first year data and the first year threshold is the smallest. Based on the first vehicle record, obtain the first dataset; According to the target extended warranty conditions, several second vehicle records of new energy vehicles are obtained, wherein the target extended warranty conditions include a second year threshold and / or a second mileage threshold, and the difference between the second mileage data of the second vehicle records and the second mileage threshold is the smallest or the difference between the second year data and the second year threshold is the smallest. Obtain the second dataset based on the second vehicle record.

8. An electronic device, characterized in that, include: At least one processor, at least one memory, and computer program instructions stored in the memory, which, when executed by the processor, implement the method as described in any one of claims 1-6.

9. A storage medium storing computer program instructions thereon, characterized in that, The method as described in any one of claims 1-6 is implemented when the computer program instructions are executed by the processor.

Images