Vehicle risk level assessment method, device, computer equipment and storage medium

Abstract

The present application relates to the field of insurance pricing, and provides a vehicle risk level assessment method, device, equipment and computer storage medium, the method comprising: obtaining first vehicle data of a fuel vehicle, the first vehicle data comprising first driving data and first risk assessment data obtained by risk assessment based on the first driving data, filtering the first data by a first filter according to the first driving data, and filtering the second data by a second filter according to the first risk assessment data, and determining transfer learning data that can be used for training an electric vehicle risk assessment model after two rounds of screening. Training an electric vehicle risk assessment model by transferring learning data solves the problem of insufficient electric vehicle data in the prior art and improves the accuracy of electric vehicle risk assessment.

Description

Technical Field

[0001] The present application relates to the field of insurance pricing, and in particular to a vehicle risk level assessment method, device, computer equipment and storage medium. Background Art

[0002] With the development of new energy vehicles, more and more vehicles are starting to use hybrid power or pure electricity as a power source. Although the risk assessment for new energy vehicle insured users has certain similarities with that for traditional fuel vehicles, since the development of new energy vehicles is in its infancy, safety performance needs to be improved, and the driving habits of drivers also need to gradually adapt to the electronic control system, the risk assessment of new energy vehicles is still different from that of traditional fuel vehicles. However, there is currently little risk assessment data related to new energy vehicle insurance, and it is difficult to train a risk assessment model with a large sample size based solely on data related to new energy vehicles. The accuracy of risk assessment for new energy vehicles is low, and how to effectively conduct risk assessment on new energy vehicles has become an urgent problem to be solved. Summary of the invention

[0003] The main purpose of this application is to provide a vehicle risk level assessment method, device, equipment and computer storage medium, which aims to train a risk assessment model for electric vehicles and perform risk level assessment on electric vehicles when data samples are insufficient.

[0004] In a first aspect, the present application provides a vehicle risk level assessment method, the vehicle risk level assessment method comprising the following steps:

[0005] Acquire first vehicle data of a fuel vehicle, the first vehicle data including first driving data and first risk assessment data obtained by performing risk assessment based on the first driving data;

[0006] Using a preset first filter to perform a first data screening on the first driving data based on standard driving data of the electric vehicle, so as to determine first vehicle data whose first data feature matching degree between the first driving data and the standard driving data exceeds a first threshold as candidate sample data;

[0007] Using a preset second filter to perform a second data screening on the candidate sample data based on the standard risk assessment data of the electric vehicle, so as to obtain candidate sample data whose matching degree between the first risk assessment data and the second data feature of the standard risk assessment data exceeds a second threshold as transfer learning data;

[0008] Training the risk assessment model of electric vehicles according to the transfer learning data to obtain a target risk assessment model;

[0009] The risk level of the target driving data of the target electric vehicle is evaluated based on the target risk assessment model to obtain risk assessment data of the target electric vehicle.

[0010] In a second aspect, the present application further provides a vehicle risk level assessment device, the vehicle risk level assessment device comprising:

[0011] A data acquisition module, used for acquiring first vehicle data of a fuel vehicle, wherein the first vehicle data includes first driving data and first risk assessment data obtained by performing risk assessment based on the first driving data;

[0012] A first screening module, configured to perform a first data screening on the first driving data based on the standard driving data of the electric vehicle using a preset first filter, so as to determine first vehicle data whose first data feature matching degree between the first driving data and the standard driving data exceeds a first threshold as candidate sample data;

[0013] A second screening module is used to perform a second data screening on the candidate sample data based on the standard risk assessment data of the electric vehicle using a preset second filter to obtain candidate sample data whose matching degree between the first risk assessment data and the second data feature of the standard risk assessment data exceeds a second threshold as transfer learning data;

[0014] A model training module, used to train the risk assessment model of electric vehicles according to the transfer learning data to obtain a target risk assessment model;

[0015] The risk assessment module is used to assess the risk level of the target driving data of the target electric vehicle based on the target risk assessment model to obtain the risk assessment data of the target electric vehicle.

[0016] In a third aspect, the present application also provides a computer device, comprising a processor, a memory, and a computer program stored in the memory and executable by the processor, wherein when the computer program is executed by the processor, the vehicle risk level assessment method as described above is implemented.

[0017] In a fourth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored, wherein when the computer program is executed by a processor, the vehicle risk level assessment method as described above is implemented.

[0018] The present application provides a vehicle risk level assessment method, device, equipment and computer storage medium. The present application obtains first vehicle data of a fuel vehicle, wherein the first vehicle data includes first driving data and first risk assessment data obtained by risk assessment based on the first driving data; uses a preset first filter to perform first data screening on the first driving data based on standard driving data of an electric vehicle, so as to determine first vehicle data whose first data feature matching degree between the first driving data and the standard driving data exceeds a first threshold as candidate sample data; uses a preset second filter to perform second data screening on the candidate sample data based on standard risk assessment data of an electric vehicle, so as to obtain candidate sample data whose second data feature matching degree between the first risk assessment data and the standard risk assessment data exceeds a second threshold as transfer learning data; trains a risk assessment model of an electric vehicle according to the transfer learning data to obtain a target risk assessment model; and evaluates the risk level of the target driving data of a target electric vehicle based on the target risk assessment model to obtain the risk assessment data of the target electric vehicle. By screening out transfer learning data from the first vehicle data of a fuel vehicle and training the risk assessment model of an electric vehicle through transfer learning data, the problem of insufficient sample data of electric vehicles is solved, and the accuracy of vehicle risk level assessment of electric vehicles is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0020] Figure 1 A flow chart of a vehicle risk level assessment method provided in one embodiment of the present application;

[0021] Figure 2 A diagram of a use scenario of a vehicle risk level assessment method provided in one embodiment of the present application;

[0022] Figure 3 A schematic block diagram of a vehicle risk level assessment device provided in one embodiment of the present application;

[0023] Figure 4 The present invention is a block diagram showing the structure of a computer device according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0025] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0026] Embodiments of the present application provide a vehicle risk level assessment method, apparatus, computer equipment, and computer-readable storage medium.

[0027] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0028] Please refer to Figure 1 , Figure 1 A flow chart of a vehicle risk level assessment method provided for an embodiment of the present application. The vehicle risk level assessment method can be used in a terminal or a server to achieve an accurate assessment of the vehicle risk of an electric vehicle. The terminal can be an electronic device such as a mobile phone, a tablet computer, a laptop computer, a desktop computer, a personal digital assistant, and a wearable device; the server can be an independent server, a server cluster, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0029] Please refer to Figure 2 , Figure 2 This is a usage scenario diagram provided by an embodiment of the present application. Figure 2 As shown, the first vehicle data of the fuel vehicle are screened in two rounds by using a first filter and a second filter, so as to determine the transfer learning data that can be used to train the electric vehicle risk assessment model from a large amount of first vehicle data, and the risk assessment model is trained by using the transfer learning data to train the target driving data that can be used for the risk assessment of the electric vehicle.

[0030] like Figure 1 As shown, the vehicle risk level assessment method includes steps S101 to S104.

[0031] Step S101: Acquire first vehicle data of a fuel vehicle, where the first vehicle data includes first driving data and first risk assessment data obtained by performing risk assessment based on the first driving data.

[0032] Due to the advantages of fuel vehicles in development history and market share, fuel vehicles have accumulated a large amount of data resources in a long development process. In contrast, new energy vehicles such as electric vehicles have a shorter development history and relatively less data accumulation. For insurance companies, when underwriting electric vehicles, it is also necessary to conduct risk assessments on the vehicles in order to price them according to vehicle risks. However, due to the limited initial data of electric vehicles in the prior art, it is difficult to train a risk assessment model for electric vehicles through a large amount of sample data. Therefore, the vehicle risk level assessment method provided in the embodiment of the present application is based on transfer learning based on the data of fuel vehicles to train a risk assessment model for electric vehicles.

[0033] Exemplarily, the risk assessment of a vehicle is to determine the risk level of the vehicle or the driver based on data related to the vehicle's driving process. Therefore, the first vehicle data of a fuel vehicle includes first driving data and first risk assessment data obtained by performing risk assessment based on the first driving data. Specifically, data related to the driving mode of a fuel vehicle can be obtained, such as behavioral data such as the driver's operating habits, sudden acceleration, sudden braking, sharp turns, and number of violations; or real-time data on the vehicle status can be obtained through sensors and monitoring systems equipped on modern vehicles, such as obtaining monitoring data such as tire pressure and engine temperature through sensors of fuel vehicles. Of course, it is not limited to this and is not limited here.

[0034] Step S102: Using a preset first filter, perform a first data screening on the first driving data based on standard driving data of the electric vehicle to determine first vehicle data whose first data feature matching degree between the first driving data and the standard driving data exceeds a first threshold as candidate sample data.

[0035] Exemplarily, standard vehicle data for evaluating the first vehicle data is obtained, wherein the standard vehicle data includes standard driving data and standard risk assessment data obtained by risk assessment based on the standard driving data. It can be understood that the amount of standard vehicle data is relatively small, and the standard risk assessment data can be manually determined based on the standard driving data. Among them, the standard driving data is driving data of the same type as the first driving data collected on the electric vehicle. For example, if the first driving data includes driving mode data of a fuel vehicle, the standard driving data may include driving mode data of the electric vehicle.

[0036] Exemplarily, the distribution probability of data can reflect the characteristics of the data, so the first driving data with a similar distribution probability to the standard driving data can be screened out, and the first vehicle data corresponding to the screened out first driving data is used as candidate sample data for further screening. Among them, since the first risk assessment data is determined based on the first driving data, the probability distribution of the first driving data is also called the prior probability of the first vehicle data, and the candidate sample data is screened out based on the prior probability of the first vehicle data of the fuel vehicle and the standard vehicle data of the electric vehicle.

[0037] In some embodiments, the first data screening of the first driving data by using a preset first filter based on standard driving data of the electric vehicle to determine first vehicle data whose first data feature matching degree between the first driving data and the standard driving data exceeds a first threshold as candidate sample data includes:

[0038] Inputting the first driving data into a preset logistic regression model, and determining first target driving data in the first driving data according to a matching probability between the first driving data and the standard driving data output by the logistic regression model;

[0039] The first target driving data is verified, and when a first data feature matching degree between the first target driving data and the standard driving data exceeds a first threshold, first vehicle data corresponding to the first target driving data is determined as the candidate sample data.

[0040] Exemplarily, the logistic regression model can learn the probability distribution of standard driving data, map the input first driving data to the probability space through a logical function, and thus predict the probability that the first driving data belongs to the standard driving data set, thereby determining the first driving data with a probability greater than a preset probability as the first target driving data.

[0041] Exemplarily, in order to ensure the accuracy of the candidate sample data and avoid classification errors of the logistic regression model, after the first target driving data is determined, the first target driving data is verified, and then the verified first target driving data is determined as the candidate sample data.

[0042] In some implementations, the verifying the first target driving data, and determining the first vehicle data corresponding to the first target driving data as the candidate sample data when the first data feature matching degree between the first target driving data and the standard driving data exceeds a first threshold, includes:

[0043] Acquire a data set to be verified consisting of the first target driving data and the standard driving data;

[0044] Performing a classification operation on the data set to be verified based on a preset classifier, and drawing a data characteristic curve corresponding to the data set to be verified according to a result of the classification operation, and determining an area under the curve corresponding to the data characteristic curve;

[0045] When the area under the curve is less than a preset area threshold, the first vehicle data corresponding to the first target driving data is determined as the candidate sample data.

[0046] Exemplarily, a classifier is set to classify the data set to be verified obtained by mixing the first target driving data and the standard driving data, wherein the classifier and the logistic regression model form an adversarial network. If the classifier has difficulty in distinguishing the first target driving data and the standard driving data in the data set to be verified, it means that the first target driving data has passed the verification. Conversely, if the classifier can distinguish the first target driving data and the standard driving data in the data set to be verified, it means that there is still a significant difference between the first target driving data and the standard driving data, and the logistic regression model needs to be further adjusted.

[0047] For example, the classification accuracy of the classifier can be quantified by the area under the curve (AUC). The data characteristic curve is the receiver operating characteristic curve (ROC) obtained by the classifier for the data set to be verified. If the value of AUC is closer to 1, it means that the classification effect of the classifier for the data set to be verified is better, that is, the difference between the first target driving data and the standard driving data is more significant; on the contrary, if the value of AUC is closer to 0.5, it means that the classifier cannot accurately distinguish the first target driving data and the standard driving data in the data set to be verified, and the first target driving data is more similar to the standard driving data. Therefore, when AUC is less than the preset area threshold, it is determined that the first target driving data has passed the verification and the first target driving data is used as the candidate sample data.

[0048] Step S103: Use a preset second filter to perform a second data screening on the candidate sample data based on the standard risk assessment data of electric vehicles to obtain candidate sample data whose matching degree between the first risk assessment data and the second data feature of the standard risk assessment data exceeds a second threshold as transfer learning data.

[0049] Exemplarily, since the first risk assessment data is determined based on the first driving data, the probability distribution of the first risk assessment data is also called the posterior probability of the first vehicle data. The candidate sample data is further screened out based on the prior probability of the candidate sample data of fuel vehicles and the standard vehicle data of electric vehicles, so as to determine the transfer learning data that matches the standard vehicle data in both prior probability and posterior probability, thereby improving the matching degree of the transfer learning data with the standard vehicle data.

[0050] In some embodiments, the second data screening of the candidate sample data by using a preset second filter based on the standard risk assessment data of the electric vehicle to obtain the candidate sample data whose matching degree between the first risk assessment data and the second data feature of the standard risk assessment data exceeds a second threshold as the transfer learning data includes:

[0051] Inputting the first driving data corresponding to the candidate sample data and the standard driving data into a preset random forest model respectively;

[0052] Based on at least one preset path of the random forest model, predicting first risk assessment data in the candidate sample data and standard risk assessment data corresponding to the standard driving data according to the first driving data and the standard driving data, respectively, to obtain a first prediction result and a second prediction result;

[0053] The transfer learning data is determined from the candidate sample data according to a second data feature matching degree between the first prediction result and the second prediction result.

[0054] Exemplarily, the random forest includes multiple preset paths, and the posterior probabilities of the candidate sample data and the standard vehicle data can be predicted by the random forest model, that is, the first driving data and the standard driving data corresponding to the candidate sample data are input into each preset path of the random forest, so that each preset path predicts the first risk assessment data and the standard risk assessment data according to the first driving data and the standard driving data, respectively, and obtains the first prediction result corresponding to the first risk assessment data and the second prediction result corresponding to the standard risk assessment data. It can be understood that if there is a preset path with similar prediction results for fuel vehicles and electric vehicles, that is, the preset path outputs similar first prediction results and second prediction results, it means that the prediction index used by the preset path is an index that is more important for both the candidate sample data and the standard sample data, and the candidate sample data can be screened according to the preset path.

[0055] In some implementations, determining the transfer learning data from the candidate sample data according to a second data feature matching degree between the first prediction result and the second prediction result includes:

[0056] Calculating result difference information between the first prediction result and the second prediction result output by each of the preset paths;

[0057] When the result difference information is less than a preset difference threshold, determining the preset path as the target path;

[0058] Based on the target path, the candidate sample data are screened according to the first risk assessment data to obtain transfer learning data in the candidate sample data that can be used as the transfer learning data.

[0059] For example, the accuracy of the first prediction result and the second prediction result can be counted. If the accuracy is similar, it means that the probability distributions of the two are relatively similar. The second data feature matching degree between the first prediction result and the second prediction result can be the difference between the accuracy of the first prediction result and the accuracy of the second prediction result, but it is not limited to this and is not limited here.

[0060] In some implementations, the step of screening the candidate sample data based on the target path and the first risk assessment data to obtain transfer learning data from the candidate sample data that can be used as the transfer learning data includes:

[0061] Determining target features among the classification features according to the classification features of each node in the target path;

[0062] Acquire a first target feature of the candidate sample data and a second target feature of standard vehicle data of an electric vehicle;

[0063] The transfer learning data is determined from the candidate sample data according to the similarity between the first target feature and the second target feature.

[0064] For example, the importance of each classification feature to the prediction result can be measured according to the mean impurity reduction (MDI) or mean decrease in accuracy (MDA) of the classification features of each node in the target path of the random forest, and one or more classification features that have the greatest impact on the prediction result are determined as target features, and the similarity between the target features of the candidate sample data and the standard vehicle data is calculated, for example, the cosine distance between the first target feature and the second target feature is calculated, and the candidate sample data with greater similarity is determined as the transfer learning data. Among them, the standard vehicle data includes standard driving data and standard risk assessment data.

[0065] Step S104: training the risk assessment model of electric vehicles according to the transfer learning data to obtain a target risk assessment model.

[0066] Exemplarily, the risk assessment model of electric vehicles can be iterated based on transfer learning data and standard vehicle data until the value of the model's loss function no longer decreases significantly, thereby obtaining a target risk assessment model that can assess the risk of electric vehicles based on the electric vehicle's driving data.

[0067] In some implementations, the vehicle risk level assessment method further includes:

[0068] During the training of the risk assessment model, the first filter and the second filter are adjusted according to the gradient norm of the transfer learning data.

[0069] Exemplarily, the first filter and the second filter respectively perform screening by setting thresholds for the first data feature matching degree and the second data feature matching degree. It is understandable that in order to obtain more source domain data that can be used for model training, the corresponding threshold can be set to a relatively small value. If it is found during the model training process that the screened transfer learning data cannot meet the requirements of model training, for example, the gradient norm of the transfer learning data cannot meet the requirements, the first filter and the second filter can be adjusted to increase the thresholds set for the first data feature matching degree and the second data feature matching degree.

[0070] Step S105 : evaluating the risk level of the target driving data of the target electric vehicle based on the target risk assessment model to obtain risk assessment data of the target electric vehicle.

[0071] Exemplarily, after obtaining the target risk assessment model, the target risk assessment model can be used to assess the risk level of a specific electric vehicle. The risk level of the target electric vehicle is assessed based on the target driving data collected on the specific target electric vehicle to obtain risk assessment data of the target electric vehicle. For example, the risk level of the target electric vehicle can be low, medium, high or other levels, so that the risk assessment data of the target electric vehicle can be used as an influencing factor for auto insurance pricing.

[0072] The vehicle risk level assessment method provided in the above embodiment obtains the first vehicle data of a fuel vehicle, wherein the first vehicle data includes first driving data and first risk assessment data obtained by risk assessment based on the first driving data; uses a preset first filter to perform a first data screening on the first driving data based on the standard driving data of an electric vehicle, so as to determine the first vehicle data whose first data feature matching degree between the first driving data and the standard driving data exceeds a first threshold as candidate sample data; uses a preset second filter to perform a second data screening on the candidate sample data based on the standard risk assessment data of the electric vehicle, so as to obtain the candidate sample data whose second data feature matching degree between the first risk assessment data and the standard risk assessment data exceeds a second threshold as transfer learning data; trains the risk assessment model of the electric vehicle according to the transfer learning data to obtain a target risk assessment model; and evaluates the risk level of the target driving data of the target electric vehicle based on the target risk assessment model to obtain the risk assessment data of the target electric vehicle. By screening the transfer learning data from the first vehicle data of the fuel vehicle and training the risk assessment model of the electric vehicle through the transfer learning data, the problem of insufficient sample data of the electric vehicle is solved, and the accuracy of the vehicle risk level assessment of the electric vehicle is improved.

[0073] See also Figure 3 , Figure 3 It is a schematic diagram of a vehicle risk level assessment device provided in one embodiment of the present application. The vehicle risk level assessment device can be configured in a server or a terminal to execute the aforementioned vehicle risk level assessment method.

[0074] like Figure 3 As shown, the vehicle risk level assessment device includes: a data acquisition module 110, a first screening module 120, a second screening module 130, a model training module 140, and a risk assessment module 150.

[0075] The data acquisition module 110 is used to acquire first vehicle data of a fuel vehicle, where the first vehicle data includes first driving data and first risk assessment data obtained by performing risk assessment based on the first driving data;

[0076] A first screening module 120, configured to perform a first data screening on the first driving data based on the standard driving data of the electric vehicle using a preset first filter, so as to determine first vehicle data whose first data feature matching degree between the first driving data and the standard driving data exceeds a first threshold as candidate sample data;

[0077] A second screening module 130 is used to perform a second data screening on the candidate sample data based on the standard risk assessment data of the electric vehicle using a preset second filter to obtain candidate sample data whose matching degree between the first risk assessment data and the second data feature of the standard risk assessment data exceeds a second threshold as transfer learning data;

[0078] A model training module 140 is used to train the risk assessment model of electric vehicles according to the transfer learning data to obtain a target risk assessment model;

[0079] The risk assessment module 150 is used to assess the risk level of the target driving data of the target electric vehicle based on the target risk assessment model to obtain risk assessment data of the target electric vehicle.

[0080] In some embodiments, the first screening module 120, in the process of performing the first data screening on the first driving data based on the standard driving data of the electric vehicle using the preset first filter to determine the first vehicle data whose first data feature matching degree between the first driving data and the standard driving data exceeds the first threshold as the candidate sample data, is used to implement:

[0081] Inputting the first driving data into a preset logistic regression model, and determining first target driving data in the first driving data according to a matching probability between the first driving data and the standard driving data output by the logistic regression model;

[0082] The first target driving data is verified, and when a first data feature matching degree between the first target driving data and the standard driving data exceeds a first threshold, first vehicle data corresponding to the first target driving data is determined as the candidate sample data.

[0083] In some implementations, the first screening module 120, in the process of implementing the verification of the first target driving data and determining the first vehicle data corresponding to the first target driving data as the candidate sample data when the first data feature matching degree between the first target driving data and the standard driving data exceeds a first threshold, is used to implement:

[0084] Acquire a data set to be verified consisting of the first target driving data and the standard driving data;

[0085] Performing a classification operation on the data set to be verified based on a preset classifier, and drawing a data characteristic curve corresponding to the data set to be verified according to a result of the classification operation, and determining an area under the curve corresponding to the data characteristic curve;

[0086] When the area under the curve is less than a preset area threshold, the first vehicle data corresponding to the first target driving data is determined as the candidate sample data.

[0087] In some embodiments, the second screening module 130 is used to implement the process of performing a second data screening on the candidate sample data based on the standard risk assessment data of the electric vehicle using the preset second filter to obtain the candidate sample data whose matching degree between the first risk assessment data and the second data feature of the standard risk assessment data exceeds a second threshold as the transfer learning data, in which:

[0088] Inputting the first driving data corresponding to the candidate sample data and the standard driving data into a preset random forest model respectively;

[0089] Based on at least one preset path of the random forest model, predicting first risk assessment data in the candidate sample data and standard risk assessment data corresponding to the standard driving data according to the first driving data and the standard driving data, respectively, to obtain a first prediction result and a second prediction result;

[0090] The transfer learning data is determined from the candidate sample data according to a second data feature matching degree between the first prediction result and the second prediction result.

[0091] In some implementations, the second screening module 130, in the process of determining the transfer learning data from the candidate sample data according to the second data feature matching degree between the first prediction result and the second prediction result, is used to implement:

[0092] Calculating result difference information between the first prediction result and the second prediction result output by each of the preset paths;

[0093] When the result difference information is less than a preset difference threshold, determining the preset path as the target path;

[0094] Based on the target path, the candidate sample data are screened according to the first risk assessment data to obtain transfer learning data in the candidate sample data that can be used as the transfer learning data.

[0095] In some implementations, the second screening module 130, in the process of screening the candidate sample data based on the target path and the first risk assessment data to obtain the transfer learning data that can be used as the transfer learning data in the candidate sample data, is used to implement:

[0096] Determining target features among the classification features according to the classification features of each node in the target path;

[0097] Acquire a first target feature of the candidate sample data and a second target feature of standard vehicle data of an electric vehicle;

[0098] The transfer learning data is determined from the candidate sample data according to the similarity between the first target feature and the second target feature.

[0099] In some implementations, the vehicle risk level assessment device, in the process of implementing the vehicle risk level assessment method, is used to implement:

[0100] During the training of the risk assessment model, the first filter and the second filter are adjusted according to the gradient norm of the transfer learning data.

[0101] It should be noted that those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices and modules and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0102] The method and apparatus of the present application can be used in many general or special computing system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0103] Exemplarily, the above method and apparatus may be implemented in the form of a computer program. The computer program may be implemented in Figure 4 Runs on the computer device shown.

[0104] See also Figure 4 , Figure 4 A schematic block diagram of the structure of a computer device provided in an embodiment of the present application. The computer device may be a server or a terminal.

[0105] like Figure 4As shown, the computer device includes a processor, a memory, and a network interface connected via a system bus, wherein the memory may include a storage medium and an internal memory.

[0106] The storage medium can store an operating system and a computer program. The computer program includes program instructions, and when the program instructions are executed, the processor can execute any vehicle risk level assessment method.

[0107] The processor is used to provide computing and control capabilities and support the operation of the entire computer equipment.

[0108] The internal memory provides an environment for the operation of the computer program in the storage medium. When the computer program is executed by the processor, the processor can execute any vehicle risk level assessment method.

[0109] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0110] It should be understood that the processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0111] In one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:

[0112] Acquire first vehicle data of a fuel vehicle, the first vehicle data including first driving data and first risk assessment data obtained by performing risk assessment based on the first driving data;

[0113] Using a preset first filter to perform a first data screening on the first driving data based on standard driving data of the electric vehicle, so as to determine first vehicle data whose first data feature matching degree between the first driving data and the standard driving data exceeds a first threshold as candidate sample data;

[0114] Using a preset second filter to perform a second data screening on the candidate sample data based on the standard risk assessment data of the electric vehicle, so as to obtain candidate sample data whose matching degree between the first risk assessment data and the second data feature of the standard risk assessment data exceeds a second threshold as transfer learning data;

[0115] Training the risk assessment model of electric vehicles according to the transfer learning data to obtain a target risk assessment model;

[0116] The risk level of the target driving data of the target electric vehicle is evaluated based on the target risk assessment model to obtain risk assessment data of the target electric vehicle.

[0117] In some embodiments, the processor, in the process of implementing the first data screening of the first driving data based on the standard driving data of the electric vehicle by using the preset first filter to determine the first vehicle data whose first data feature matching degree between the first driving data and the standard driving data exceeds a first threshold as candidate sample data, is used to implement:

[0118] Inputting the first driving data into a preset logistic regression model, and determining first target driving data in the first driving data according to a matching probability between the first driving data and the standard driving data output by the logistic regression model;

[0119] The first target driving data is verified, and when a first data feature matching degree between the first target driving data and the standard driving data exceeds a first threshold, first vehicle data corresponding to the first target driving data is determined as the candidate sample data.

[0120] In some implementations, the processor, in implementing the verification of the first target driving data, determines the first vehicle data corresponding to the first target driving data as the candidate sample data when the first data feature matching degree between the first target driving data and the standard driving data exceeds a first threshold, is used to implement:

[0121] Acquire a data set to be verified consisting of the first target driving data and the standard driving data;

[0122] Performing a classification operation on the data set to be verified based on a preset classifier, and drawing a data characteristic curve corresponding to the data set to be verified according to a result of the classification operation, and determining an area under the curve corresponding to the data characteristic curve;

[0123] When the area under the curve is less than a preset area threshold, the first vehicle data corresponding to the first target driving data is determined as the candidate sample data.

[0124] In some embodiments, the processor, in the process of implementing the second data screening of the candidate sample data based on the standard risk assessment data of the electric vehicle using the preset second filter to obtain the candidate sample data whose matching degree between the first risk assessment data and the second data feature of the standard risk assessment data exceeds a second threshold as the transfer learning data, is used to implement:

[0125] Inputting the first driving data corresponding to the candidate sample data and the standard driving data into a preset random forest model respectively;

[0126] Based on at least one preset path of the random forest model, predicting first risk assessment data in the candidate sample data and standard risk assessment data corresponding to the standard driving data according to the first driving data and the standard driving data, respectively, to obtain a first prediction result and a second prediction result;

[0127] The transfer learning data is determined from the candidate sample data according to a second data feature matching degree between the first prediction result and the second prediction result.

[0128] In some implementations, the processor, in the process of determining the transfer learning data from the candidate sample data according to the second data feature matching degree between the first prediction result and the second prediction result, is configured to implement:

[0129] Calculating result difference information between the first prediction result and the second prediction result output by each of the preset paths;

[0130] When the result difference information is less than a preset difference threshold, determining the preset path as the target path;

[0131] Based on the target path, the candidate sample data are screened according to the first risk assessment data to obtain transfer learning data in the candidate sample data that can be used as the transfer learning data.

[0132] In some implementations, the processor, in implementing the process of screening the candidate sample data based on the target path and according to the first risk assessment data to obtain the transfer learning data in the candidate sample data that can be used as the transfer learning data, is used to implement:

[0133] Determining target features among the classification features according to the classification features of each node in the target path;

[0134] Acquire a first target feature of the candidate sample data and a second target feature of standard vehicle data of an electric vehicle;

[0135] The transfer learning data is determined from the candidate sample data according to the similarity between the first target feature and the second target feature.

[0136] In some implementations, the processor, in the process of implementing the vehicle risk level assessment method, is further used to implement:

[0137] During the training of the risk assessment model, the first filter and the second filter are adjusted according to the gradient norm of the transfer learning data.

[0138] It should be noted that technicians in the relevant field can clearly understand that for the convenience and brevity of description, the specific working process of the vehicle risk level assessment described above can refer to the corresponding process in the aforementioned vehicle risk level assessment control method embodiment, and will not be repeated here.

[0139] An embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored. The computer program includes program instructions. The method implemented when the program instructions are executed can refer to the various embodiments of the vehicle risk level assessment method of the present application.

[0140] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiment, such as a hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a smart memory card (SmartMedia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc., equipped on the computer device.

[0141] It should be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0142] It should also be understood that the term "and / or" used in the specification of this application and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, including these combinations. It should be noted that, in this article, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "including a..." does not exclude the presence of other identical elements in the process, method, article or system including the element.

[0143] The serial numbers of the embodiments of the present application are for description only and do not represent the advantages and disadvantages of the embodiments. The above description is only a specific implementation mode of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A vehicle risk level assessment method, characterized in that: The method comprises: Acquire first vehicle data of a fuel vehicle, the first vehicle data including first driving data and first risk assessment data obtained by performing risk assessment based on the first driving data; Using a preset first filter to perform a first data screening on the first driving data based on standard driving data of the electric vehicle, so as to determine first vehicle data whose first data feature matching degree between the first driving data and the standard driving data exceeds a first threshold as candidate sample data; Using a preset second filter to perform a second data screening on the candidate sample data based on the standard risk assessment data of the electric vehicle, so as to obtain candidate sample data whose matching degree between the first risk assessment data and the second data feature of the standard risk assessment data exceeds a second threshold as transfer learning data; Training the risk assessment model of electric vehicles according to the transfer learning data to obtain a target risk assessment model; Evaluate the risk level of target driving data of the target electric vehicle based on the target risk assessment model to obtain risk assessment data of the target electric vehicle; The method of using a preset second filter to perform a second data screening on the candidate sample data based on the standard risk assessment data of the electric vehicle to obtain candidate sample data whose matching degree between the first risk assessment data and the second data feature of the standard risk assessment data exceeds a second threshold as transfer learning data includes: Inputting the first driving data corresponding to the candidate sample data and the standard driving data into a preset random forest model respectively; Based on at least one preset path of the random forest model, predicting first risk assessment data in the candidate sample data and standard risk assessment data corresponding to the standard driving data according to the first driving data and the standard driving data, respectively, to obtain a first prediction result and a second prediction result; The transfer learning data is determined from the candidate sample data according to a second data feature matching degree between the first prediction result and the second prediction result.

2. The vehicle risk level assessment method according to claim 1, characterized in that: The method of using a preset first filter to perform a first data screening on the first driving data based on the standard driving data of the electric vehicle to determine the first vehicle data whose first data feature matching degree between the first driving data and the standard driving data exceeds a first threshold as candidate sample data includes: Inputting the first driving data into a preset logistic regression model, and determining first target driving data in the first driving data according to a matching probability between the first driving data and the standard driving data output by the logistic regression model; The first target driving data is verified, and when a first data feature matching degree between the first target driving data and the standard driving data exceeds a first threshold, first vehicle data corresponding to the first target driving data is determined as the candidate sample data.

3. The vehicle risk level assessment method according to claim 2, characterized in that: The verifying the first target driving data, and determining the first vehicle data corresponding to the first target driving data as the candidate sample data when the first data feature matching degree between the first target driving data and the standard driving data exceeds a first threshold, includes: Acquire a data set to be verified consisting of the first target driving data and the standard driving data; Performing a classification operation on the to-be-verified data set based on a preset classifier, and drawing a data characteristic curve corresponding to the to-be-verified data set according to a result of the classification operation, and determining an area under the curve corresponding to the data characteristic curve; When the area under the curve is less than a preset area threshold, the first vehicle data corresponding to the first target driving data is determined as the candidate sample data.

4. The vehicle risk level assessment method according to claim 1, characterized in that: The determining the transfer learning data from the candidate sample data according to the second data feature matching degree between the first prediction result and the second prediction result includes: Calculating result difference information between the first prediction result and the second prediction result output by each of the preset paths; When the result difference information is less than a preset difference threshold, determining the preset path as the target path; Based on the target path, the candidate sample data are screened according to the first risk assessment data to obtain transfer learning data in the candidate sample data that can be used as the transfer learning data.

5. The vehicle risk level assessment method according to claim 4, characterized in that: The step of screening the candidate sample data based on the target path and the first risk assessment data to obtain transfer learning data from the candidate sample data that can be used as the transfer learning data includes: Determining target features among the classification features according to the classification features of each node in the target path; Acquire a first target feature of the candidate sample data and a second target feature of standard vehicle data of an electric vehicle; The transfer learning data is determined from the candidate sample data according to the similarity between the first target feature and the second target feature.

6. The vehicle risk level assessment method according to claim 1, characterized in that: The vehicle risk level assessment method further includes: During the training of the risk assessment model, the first filter and the second filter are adjusted according to the gradient norm of the transfer learning data.

7. A vehicle risk level assessment device, characterized in that: The vehicle risk level assessment device comprises: A data acquisition module, used for acquiring first vehicle data of a fuel vehicle, wherein the first vehicle data includes first driving data and first risk assessment data obtained by performing risk assessment based on the first driving data; A first screening module, configured to perform a first data screening on the first driving data based on the standard driving data of the electric vehicle using a preset first filter, so as to determine first vehicle data whose first data feature matching degree between the first driving data and the standard driving data exceeds a first threshold as candidate sample data; A second screening module is used to perform a second data screening on the candidate sample data based on the standard risk assessment data of the electric vehicle using a preset second filter to obtain candidate sample data whose matching degree between the first risk assessment data and the second data feature of the standard risk assessment data exceeds a second threshold as transfer learning data; A model training module, used to train the risk assessment model of electric vehicles according to the transfer learning data to obtain a target risk assessment model; A risk assessment module, used to assess the risk level of target driving data of a target electric vehicle based on the target risk assessment model to obtain risk assessment data of the target electric vehicle; The second screening module is used for performing a second data screening on the candidate sample data based on the standard risk assessment data of the electric vehicle using the preset second filter to obtain the candidate sample data whose matching degree between the first risk assessment data and the second data feature of the standard risk assessment data exceeds a second threshold as the transfer learning data, and is also used for: Inputting the first driving data corresponding to the candidate sample data and the standard driving data into a preset random forest model respectively; Based on at least one preset path of the random forest model, predicting first risk assessment data in the candidate sample data and standard risk assessment data corresponding to the standard driving data according to the first driving data and the standard driving data, respectively, to obtain a first prediction result and a second prediction result; The transfer learning data is determined from the candidate sample data according to a second data feature matching degree between the first prediction result and the second prediction result.

8. A computer device, characterized in that: The computer device includes a processor, a memory, and a computer program stored in the memory and executable by the processor, wherein when the computer program is executed by the processor, the steps of the vehicle risk level assessment method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the steps of the vehicle risk level assessment method according to any one of claims 1 to 6 are implemented.

Images