Vehicle tracing analysis method and system in hot event and electronic equipment

Abstract

The invention relates to the technical field of vehicle tracing, and discloses a vehicle tracing analysis method and system in a hot event and electronic equipment. The method comprises the following steps: acquiring a hot event, identifying a target vehicle in an image or video associated with the hot event, and determining a target feature; searching based on the published information of the hotspot event, and determining the occurrence time period and the location area of the hotspot event; inputting the target feature, the time period and the position area into a cloud for joint retrieval, and determining a vehicle identification code to which the target vehicle belongs; vehicle information corresponding to the target vehicle in front and back time periods is called according to the vehicle identification code, data analysis is carried out on the vehicle information, an analysis report is determined, and the vehicle information including instrument data, central control screen display data and appearance data is stored in the cloud. Through cloud joint retrieval, accurate identification of the target vehicle is realized, rapid and accurate public gateway response and forward maintenance of the vehicle brand value are facilitated, and the event processing efficiency is improved at the same time.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle tracing, and in particular to a vehicle tracing analysis method, system and electronic equipment in hot events. Background Art

[0002] With the continuous development and progress of technology, the automobile industry has undergone tremendous changes. It no longer relies solely on gasoline and diesel as fuel as in the past, but is moving towards new energy.

[0003] In related technologies, new energy vehicles have thousands of sensors distributed throughout the vehicle, which upload sensor data to cloud servers in real time, forming a huge data stream that is becoming more and more networked and intelligent. If a car accident is detected, the cloud server can clearly know it. However, for hot events on the Internet caused by accidents, it is impossible to quickly locate the relevant information of the vehicle, which easily leads to untimely and inadequate event handling and public relations handling, which is not only not conducive to vehicle brand building, but also reduces the user experience.

[0004] In view of this, there is an urgent need for a vehicle tracing and analysis solution for quickly locating hot events. Summary of the invention

[0005] In order to provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. The summary is not an extensive review, nor is it intended to identify key / critical components or delineate the scope of protection of these embodiments, but rather serves as a prelude to the detailed description that follows.

[0006] In view of the above-mentioned shortcomings of the prior art, the present application provides a method, system and electronic device for tracing and analyzing vehicles in hot events to solve the problem of being unable to quickly and accurately locate vehicle information in hot events.

[0007] A first aspect of the present application provides a method for tracing and analyzing vehicles in hot events, including: acquiring a hot event, identifying a target vehicle in an image or video associated with the hot event, and determining target features, wherein the target features include appearance data including vehicle model, paint color, and abnormal parts, wherein the abnormal parts are parts of the target vehicle damaged due to an accident; searching based on the release information of the hot event to determine the time period and location area where the hot event occurred; inputting the target features, the time period, and the location area into the cloud for joint retrieval to determine the vehicle identification code to which the target vehicle belongs; retrieving vehicle information corresponding to the target vehicle in the time periods before and after according to the vehicle identification code, performing data analysis on the vehicle information, and determining an analysis report, wherein the cloud stores the vehicle information including instrument data, central control screen display data, and appearance data.

[0008] In some possible embodiments of the present application, before obtaining hot events, it also includes: obtaining CAN message signals uploaded by vehicles; parsing the CAN message signals, and performing format conversion to generate the message signals in a preset format; clustering the message data in a preset format according to business flows using a preset data model, generating indicator data with labels and storing them in the cloud, wherein the type of the preset data model corresponds to the type of the indicator data, and the type of the indicator data is associated with the instrument data, the central control screen display data, and the appearance data.

[0009] In some possible embodiments of the present application, the target features, the time period, and the location area are input into the cloud for joint retrieval to determine the vehicle identification code to which the target vehicle belongs, including: inputting the target features, the time period, and the location area into a preset retrieval model in the cloud for joint retrieval to determine the vehicle identification code to which the target vehicle belongs; wherein the indicator data of the vehicle is mapped to the vehicle information to form a training set, and the preset retrieval model is obtained by training a convolutional neural network using the training set.

[0010] In some possible embodiments of the present application, the target features, the time period and the location area are input into a preset retrieval model in the cloud for joint retrieval to determine the vehicle identification code to which the target vehicle belongs, including: extracting the word vector of each text sample with a label, converting the word vector into a two-dimensional matrix, inputting the word vector into a convolutional neural network, and training to obtain a preset retrieval model; wherein each of the text samples stores the indicator data and the vehicle information of the same vehicle; combining the target features, the time period and the location area into a search text, segmenting the search text, and filtering modal particles to generate a search word vector; inputting the search word vector into the preset retrieval model to obtain the similarity between texts, and sorting them from high to low according to the similarity to determine the vehicle identification code to which the target vehicle belongs.

[0011] In some possible embodiments of the present application, it also includes: if it is monitored that any of the indicator data is missing within a certain period of time, analysis is performed based on the vehicle's current speed information, mileage, difference in location area before and after, and data missing period, and the indicator data is supplemented according to the analysis results to obtain complete the indicator data.

[0012] In some possible embodiments of the present application, data analysis is performed on the vehicle information to determine an analysis report, including: determining the change state of the vehicle's brake pedal, accelerator pedal, vehicle speed, steering wheel steering, seat belt status, and engine speed in the instrument data within the time period; based on the external environment and traffic conditions in the location area where the vehicle is located at that time, data analysis is performed on the change state of the vehicle's instrument data to determine whether the driver's current driving strategy has any misoperation; if the driver has no misoperation, a first analysis report on the vehicle status analysis is generated; if the driver has any misoperation, a second analysis report on the vehicle status and misoperation analysis is generated.

[0013] In some possible embodiments of the present application, after determining the analysis report, it also includes: sending the analysis report to the staff, conducting a return visit to the owner of the target vehicle based on the analysis report, and generating a return visit record; improving and optimizing the defects of the vehicle through the return visit record to improve the quality of the vehicle.

[0014] In some possible embodiments of the present application, it also includes: obtaining a standard vehicle image corresponding to the vehicle model, comparing the standard vehicle image with the image associated with the hot event, and judging the damaged parts on the target vehicle through differential comparison, and treating the damaged parts as abnormal parts.

[0015] The second aspect of the present application also provides a vehicle tracing and analysis system in hot events, including: a feature determination module, which is used to obtain hot events, identify target vehicles in images or videos associated with the hot events, and determine target features, wherein the target features are appearance data including vehicle model, paint color and abnormal parts, and the abnormal parts are parts of the target vehicle damaged due to an accident; a hot search module, which is used to search based on the release information of the hot event and determine the time period and location area where the hot event occurred; a retrieval and tracing module, which is used to input the target features, the time period and the location area into the cloud for joint retrieval to determine the vehicle identification code to which the target vehicle belongs; a report generation module, which is used to retrieve the vehicle information corresponding to the target vehicle in the time period before and after according to the vehicle identification code, perform data analysis on the vehicle information, and determine an analysis report, wherein the vehicle information includes instrument data, central control screen display data and appearance data.

[0016] The third aspect of the present application also provides an electronic device, including: a processor and a memory; the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the electronic device performs the above method.

[0017] The beneficial effects of the present application are as follows: by identifying the target vehicle in the image or video associated with the hot event, the target features are determined, which is conducive to identifying the target features of the vehicle related to the event and facilitating subsequent rapid positioning; searching based on the release information of the hot event, determining the time period and location area where the hot event occurred, which can narrow the search scope; inputting the target features, the time period and the location area into the cloud for joint retrieval to determine the vehicle identification code to which the target vehicle belongs; in this way, through the cloud joint retrieval and comprehensive vehicle information retrieval, accurate identification of the target vehicle is achieved, which is conducive to rapid and accurate public relations response, positive maintenance of the vehicle brand value, and at the same time, the efficiency of event handling is improved, and the vehicle information corresponding to the target vehicle in the time period before and after is retrieved according to the vehicle identification code, and the vehicle information is analyzed to determine the analysis report, which can not only analyze the cause of the accident caused by the vehicle, but also provide more comprehensive vehicle information support for the owner, which helps to reduce potential risks and improve vehicle safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flow chart of a vehicle tracing analysis method in a hot event in an embodiment of the present application;

[0019] Figure 2 It is a schematic diagram of processing target features in a vehicle tracing analysis method in a hot event in an embodiment of the present application;

[0020] Figure 3 It is a structural schematic diagram of a vehicle tracing and analysis system in a hot event in an embodiment of the present application;

[0021] Figure 4 It is a structural schematic diagram of an electronic device in an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The following describes the embodiments of the present invention through specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and sub-samples in the embodiments can be combined with each other without conflict.

[0023] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and thus the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0024] In the following description, numerous details are discussed to provide a more thorough explanation of the embodiments of the present invention. However, it is obvious to those skilled in the art that the embodiments of the present invention can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present invention difficult to understand.

[0025] The terms "first", "second", etc. in the specification and claims of the embodiments of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchanged where appropriate, so as to describe the embodiments of the embodiments of the present disclosure described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions.

[0026] Unless otherwise stated, the term "plurality" means two or more.

[0027] In the embodiment of the present disclosure, the character " / " indicates that the preceding and following objects are in an "or" relationship. For example, A / B indicates: A or B.

[0028] The term "and / or" is a description of the association relationship between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or, A and B.

[0029] Combination Figure 1 As shown, it is a flow chart of a vehicle tracing analysis method in a hot event in an embodiment of the present application, including:

[0030] Step S101, obtaining a hot event, identifying a target vehicle in an image or video associated with the hot event, and determining target features, wherein the target features include appearance data including vehicle model, paint color, and abnormal parts, wherein the abnormal parts are parts of the target vehicle damaged due to an accident;

[0031] Among them, the target vehicles are new energy vehicles and fuel vehicles. New energy vehicles include but are not limited to plug-in hybrid vehicles, extended-range electric vehicles, fuel cell vehicles, fuel cell vehicles, oil-electric hybrid vehicles, and pure electric vehicles. Fuel vehicles include but are not limited to sedans, SUVs, MPVs, pickups, sports cars, trucks, buses, and special-purpose vehicles.

[0032] Hot events refer to events or topics that have attracted widespread public attention and discussion and have become a common topic in society. For example, hot events on online platforms such as Douyin, Weibo, Kuaishou or other third-party platforms are one or more images or videos involving vehicle-related topics.

[0033] For example, hot events are obtained on the network platform through crawler software, and hot events related to vehicle topics are selected.

[0034] For another example, through image processing and analysis technology, relevant information of the vehicle is extracted from the image or video, including but not limited to the vehicle model (such as vehicle type), paint color, license plate number and other appearance features of the vehicle. Specifically, the input image or video is preprocessed, and the preprocessing includes denoising, contrast enhancement, graying, etc.; in the preprocessed image or video, the target detection algorithm (such as convolutional neural network, etc.) is used to detect the existence and position of the vehicle, and the image or video frame is traversed, and the vehicle is identified by using feature matching, deep learning and other technologies. The detected vehicle is feature extracted, including the vehicle model, paint color and abnormal parts (such as damaged parts), etc., and the various parts of the vehicle are identified by using image segmentation, edge detection, shape analysis and other technologies, and relevant features are extracted to determine the model and paint color of the vehicle. For abnormal parts, for example, machine learning or deep learning algorithms can be used to identify and analyze damaged parts and their degree.

[0035] It should be noted that the target features are features that can be obtained from the vehicle's appearance data, while the abnormal features are damaged vehicle parts. For example, the manifestations may be damaged window glass, damaged windshield, fallen tires, damaged paint color, deformation or damage of one or more parts of the vehicle due to an accident, etc., which will not be elaborated here.

[0036] In some embodiments, it also includes: obtaining a standard vehicle image corresponding to the vehicle model, comparing the standard vehicle image with the image associated with the hot event, and judging the damaged parts on the target vehicle through differential comparison, and treating the damaged parts as abnormal parts.

[0037] Specifically, the type of target vehicle is identified, the standard vehicle image corresponding to the target vehicle is determined, and an image difference algorithm (such as absolute difference, square difference, etc.) is used to compare the difference points of the target vehicle in the standard vehicle image and the hot event image, the pixel difference between the two images is calculated, and a difference image is generated, which will highlight the differences between the two images.

[0038] Analyze the difference image to identify which areas show significant differences. Combine the structural knowledge of the vehicle and the standard vehicle image to determine which parts of the vehicle these difference areas correspond to. Eliminate false positive differences due to changes in lighting, differences in viewing angles, or image quality issues. Mark the parts that are judged to be damaged as abnormal parts, for example, by highlighting them with a highlight color, border, or text annotation on the difference image. Generate a report containing information about the abnormal parts, including the name, location, degree of damage, etc. of the abnormal parts. If necessary, include the hot event image, standard vehicle image, and difference image in the report for further analysis or verification.

[0039] Step S102, searching based on the release information of the hot event to determine the time period and location area where the hot event occurs;

[0040] The source of the hot event is determined by obtaining the publisher of the hot event for the first time on the network platform and determining the real publisher of the hot event. For example, the time period and location area of ​​the hot event can be determined through the description information (publishing information) provided by the publisher, such as the specific time of a certain year and month, the specific location of a certain street or road section.

[0041] Specifically, the location area where the hot event occurred can be determined by analyzing a certain building sign or road section sign in the video or image, or the specific location area can be determined through the information released by the user (i.e., the publisher); similarly, the time period when the hot event occurred can be determined by analyzing the shooting time in the video or image, or the specific time period can be determined through the information released by the user (i.e., the publisher).

[0042] Step S103, inputting the target feature, the time period and the location area into the cloud for joint retrieval to determine the vehicle identification code to which the target vehicle belongs;

[0043] Among them, the acquired target features, time periods and location areas are input into the database in the cloud (server) for matching. For example, the database stores each vehicle in the corresponding time period, location information and corresponding sensor data, wherein the sensor data includes but is not limited to mileage, vehicle speed, fuel consumption, intelligent driving, steering wheel, seat functions, instrument display, central control screen display, doors and windows, and other complete vehicle data.

[0044] Exemplarily, the query conditions of target features, time period and location area are matched one by one in the database, and vehicles that simultaneously meet the query conditions are matched, so that the vehicle identification code to which the target vehicle belongs can be determined.

[0045] Step S104, retrieve the vehicle information corresponding to the target vehicle in the previous and next time periods according to the vehicle identification code, perform data analysis on the vehicle information, and determine an analysis report. The cloud stores the vehicle information including instrument data, central control screen display data, and appearance data.

[0046] Exemplarily, the VIN code is a unique identifier for the vehicle, which is used to accurately find relevant data of the target vehicle in the cloud database, and determine the time period when the vehicle information needs to be retrieved based on the time of occurrence of the hot event, including the time period before and after the event in a specific location area (where the accident occurred); using the VIN code as a query condition, the vehicle information of the target vehicle within the specified time period is retrieved from the cloud database, including but not limited to instrument data (such as speed, rotation speed, fuel level, etc.), central control screen display data (such as navigation records, entertainment system usage records, etc.) and appearance data (such as vehicle status, damage status, etc.).

[0047] For example, analyze the driving status of the vehicle before and after the accident in different time periods, such as average speed, acceleration, braking frequency, etc., to evaluate the vehicle's driving performance and driving behavior. For another example, analyze the central control screen display data, such as air conditioning, energy consumption, seat settings, ventilation and other habitual parameters, to determine the user's car usage habits in the current period. For another example, compare the vehicle appearance data before and after the incident to identify whether the vehicle has been damaged or changed, and evaluate the extent of the damage and possible causes. Use machine learning methods to detect abnormal values ​​or abnormal patterns in vehicle information, such as abnormal driving data, frequent fault alarms, etc., to determine potential problems or safety hazards in the vehicle, and write a detailed analysis report based on the data analysis results. The report should include an overview of the vehicle information, data analysis results, abnormal detection conditions, and possible conclusions and recommendations. Use visualization methods such as charts and images to intuitively display data analysis results and abnormal detection conditions, so that readers can better understand the content of the report and quickly identify key information. Review the analysis report to ensure the accuracy of the data and the rationality of the analysis. After the review is passed, the report will be released to relevant parties, such as car owners, insurance companies, and traffic management departments.

[0048] In some embodiments, performing data analysis on the vehicle information to determine an analysis report includes:

[0049] Determine the change status of the brake pedal, accelerator pedal, vehicle speed, steering wheel direction, seat belt status, and engine speed of the vehicle in the instrument data within the time period;

[0050] According to the external environment and traffic conditions of the vehicle in the location area at that time, performing data analysis on the change state of the instrument data of the vehicle to determine whether there is any misoperation in the driver's current driving strategy;

[0051] If the driver does not make any wrong operation, generating a first analysis report of the vehicle status analysis;

[0052] If the driver has made an erroneous operation, a second analysis report of the vehicle status and the erroneous operation analysis is generated.

[0053] For example, the instrument data of the target vehicle within a specified period of time is retrieved from the cloud database, including the brake pedal status, accelerator pedal position, vehicle speed, steering wheel steering angle, seat belt status, and engine speed, etc., and the possible impact of these factors on the change of vehicle instrument data is analyzed in combination with the external environment and traffic conditions. For example, rainy days make the road slippery, which in turn affects the speed and use of the brake pedal; traffic congestion will lead to frequent starts and stops and steering wheel turns. Similarly, the current road section has a clear speed limit of 120 kilometers per hour, but the instrument data collected before and after the accident has exceeded 150 kilometers per hour. Such an accident caused by obvious speeding is a kind of misoperation. In addition, when a pedestrian or obstacle is found in front of the vehicle, the accelerator pedal is stepped on when the brake pedal should be stepped on. This is also a kind of misoperation.

[0054] Through analysis, if the driver has not made any misoperation, a first analysis report of the vehicle status analysis is generated, and the first analysis report involves recording the vehicle's instrument data change status, external environment and traffic conditions, and the driver's driving strategy evaluation results. If the driver has made any misoperation, a second analysis report of the vehicle status and misoperation analysis is generated. In addition to recording the vehicle's instrument data change status, external environment and traffic conditions, and the driver's driving strategy evaluation results, it also includes the specific type of misoperation, the time and place of occurrence, and the consequences caused; at the same time, suggestions for improving driving strategies are given to help drivers improve driving safety and efficiency.

[0055] Through the above method, the vehicle's instrument data as well as the external environment and traffic conditions are comprehensively analyzed to determine whether the driver has made any misoperation, and a corresponding analysis report is generated to provide driving behavior feedback to the owner, help insurance companies assess risks, and provide decision-making support for traffic management departments.

[0056] Optionally, in some embodiments, after determining the analysis report, the method further includes:

[0057] Sending the analysis report to the staff, and conducting a return visit to the owner of the target vehicle based on the analysis report to generate a return visit record;

[0058] The defects of the vehicle are improved and optimized through the return visit records to improve the quality of the vehicle.

[0059] Specifically, the staff will professionally interpret the analysis report to determine the cause of the vehicle accident. In this way, during subsequent follow-up visits, on the one hand, the analysis report can be interpreted for the car owner to prevent the car owner from not understanding or not seeing the report. On the other hand, through the professional knowledge of the staff and subsequent service docking, it not only reflects the car dealer’s care and attention to users, but also can enhance the vehicle brand.

[0060] In some embodiments, before obtaining the hot events, the method further includes:

[0061] Get the CAN message signal uploaded by the vehicle;

[0062] Parsing the CAN message signal and performing format conversion to generate the message signal in a preset format;

[0063] The message data in a preset format is clustered according to the business flow using a preset data model to generate indicator data with labels and store them in the cloud. The type of the preset data model corresponds to the type of the indicator data, and the type of the indicator data is associated with the instrument data, the central control screen display data and the appearance data.

[0064] Specifically, the various sensors and controllers of the vehicle send message signals through the CAN bus, and through the OBD interface or other vehicle communication interface, a dedicated CAN bus adapter or ECU development tool is used to capture the CAN message signals sent by the vehicle in real time. For example, the hexadecimal CAN message signals sent by the vehicle-side controller are collected, including mileage, speed, fuel consumption, intelligent driving, steering wheel, seat function, instrument display, central control screen display, doors and windows, and other vehicle data. The captured CAN message signals are parsed to extract useful information and convert it into a message signal in a preset format. According to the CAN protocol (such as ISO 15765-2) and the vehicle-specific CAN database (DBC file), each field in the message signal is parsed, and the parsed data is converted into a specified data format, such as JSON, XML or binary format, for subsequent processing. Use the preset data model to cluster the converted message data and generate index data with labels. According to business needs, design a data model to define the type and structure of the index data. The model includes various performance indicators, safety indicators, comfort indicators, etc. of the vehicle. Use machine learning algorithms (such as K-means, hierarchical clustering, etc.) or rule engines to cluster the message data, classify similar data points into one category, and assign corresponding labels to them. Store the clustered index data in the cloud database and associate it with other vehicle data (such as instrument data, central control screen display data, appearance data).

[0065] Use database services (such as MySQL, MongoDB, DynamoDB, etc.) provided by cloud service providers (such as AWS, Azure, Alibaba Cloud, etc.) to store indicator data, and associate the indicator data with other vehicle data through the primary key-foreign key relationship of the database, the connection operation of the data table, or the ETL process of the data warehouse.

[0066] Through the above method, useful information is extracted from the vehicle CAN message signal and converted into indicator data valuable to the business, providing strong support for vehicle health management, fault diagnosis, performance optimization, etc.

[0067] In some embodiments, the target feature, the time period, and the location area are input into the cloud for joint retrieval to determine the vehicle identification code to which the target vehicle belongs, including:

[0068] Input the target feature, the time period and the location area into a preset search model in the cloud for joint search to determine the vehicle identification code to which the target vehicle belongs;

[0069] The index data of the vehicle is mapped to the vehicle information to form a training set, and the preset retrieval model is obtained by training a convolutional neural network using the training set.

[0070] Specifically, in one embodiment, a preset retrieval model can be used to select a target similar question with the highest confidence from each similar question as the similar question with the highest similarity to the question statement. The training model adopts a supervised learning training method, that is, when the preset supervision model is trained using a training sample set, the training samples have corresponding annotations, that is, in the training sample set, each pair of similar questions carries a corresponding label, which is used to indicate whether the two similar questions in each pair of similar questions belong to the same knowledge point, and the similarity value of the two similar questions in each pair of similar questions is also determined. The preset supervision model can be a logistic regression, a distributed gradient boosting library, or a gradient boosting algorithm library model.

[0071] Specifically, different target features, time periods and location areas are combined into question statements with different contents, the similarity values ​​between the question statements and each similar question are calculated, and all similarity values ​​are input into a preset retrieval model; based on the confidence results output by the preset retrieval model, the target similar question with the highest similarity to the question statement is determined.

[0072] Among them, the similarity value between the user's question statement and each similar question can be expressed by the features between the question statement and each similar question, and the features may include one or more of the following: minimum edit distance, longest common substring length, TFIDF weighted average word vector cosine distance, WMD score and sentence pair-based neural network features.

[0073] In another embodiment, taking an input image in a hot event as an example, the extracted target features, the time period, and the location area are combined as search tags, and the search tags are associated with the input image and bound into a sample to form a training sample set. The convolutional neural network is trained with the training sample set. In this way, after the training is completed, a preset retrieval model is formed, which can be deployed in the cloud for retrieving vehicles.

[0074] In some embodiments, the target feature, the time period, and the location area are input into a preset search model in the cloud for joint search to determine the vehicle identification code to which the target vehicle belongs, including:

[0075] Extract the word vector of each text sample with a label, convert the word vector into a two-dimensional matrix, input it into a convolutional neural network, and train to obtain a preset retrieval model; wherein each text sample stores the indicator data and the vehicle information of the same vehicle;

[0076] Combining the target feature, the time period and the location area into a search text, segmenting the search text, filtering modal particles, and generating a search word vector;

[0077] The search word vector is input into the preset retrieval model to obtain the similarity between the texts, and the texts are sorted from high to low according to the similarity to determine the vehicle identification code to which the target vehicle belongs.

[0078] Specifically, for each text sample (storing the index data and vehicle information of the same vehicle), extract the word vector, and convert each word in the text into a numerical vector of fixed length to capture the semantic relationship between words. The extracted word vector is converted into a two-dimensional matrix in a certain way (such as arranged in word order) as the input of the convolutional neural network. A training set containing multiple text samples (and their corresponding vehicle identification codes) is used to train the convolutional neural network. By learning the word vector representation of these text samples, a mapping relationship between vehicle features and their identification codes is established. After the training is completed, the convolutional neural network model obtained is the preset retrieval model, which can receive new text input and output the ranking of similar text (or vehicle identification codes).

[0079] Combine the target features (such as license plate number, vehicle model, etc.), time period and location area into a search text and search it in the preset retrieval model. Perform word segmentation on the search text to extract the keywords; at the same time, perform modal particle filtering to remove those words that have no substantial impact on the retrieval results. Convert the search text after word segmentation and filtering into a word vector representation, and use the search word vector as the input of the preset retrieval model. Input the search word vector into the preset retrieval model, and the model will calculate the similarity between the search text and each text sample in the training set, and sort the text samples from high to low according to the similarity. The vehicle identification code corresponding to the most similar text sample is the identification code of the target vehicle.

[0080] Through the above method, the vehicle identification code corresponding to the text sample with the highest similarity is selected as the retrieval result. The deep learning and natural language processing techniques are utilized to establish the mapping relationship between vehicle features and its identification code by training the convolutional neural network model, thereby achieving fast and accurate retrieval of the target vehicle.

[0081] In some embodiments, a vehicle tracing analysis method in a hot event in an embodiment of the present application is used, which is described in detail as follows:

[0082] 1. Data upload to the cloud: The data acquisition equipment on the vehicle side of new energy vehicles collects and stores real-time data of the whole vehicle and controller signals, and collects the hexadecimal CAN message signals sent by the vehicle-side controller, including mileage, vehicle speed, fuel consumption, intelligent driving, steering wheel, seat functions, instrument display, central control screen display, doors and windows, and other complete vehicle data, and then uploads them to the cloud.

[0083] 2. Data analysis: Cloud data is analyzed locally, and the hexadecimal signal data is parsed into decimal data according to rules.

[0084] 3. Data processing: The decimal data enters the data model (regression model, clustering model, etc.) of the local database and is then labeled.

[0085] 4. Data algorithm model: After the data is labeled, the algorithm model is built according to driving habits and driving conditions industry rules.

[0086] Each car is equipped with a mileage meter to display the total mileage, but each car manufacturer calculates the total mileage in different ways. Take the total mileage business indicator as an example:

[0087] a. Cumulative mileage is the total mileage of the vehicle from the time it rolls off the production line to the time it is scrapped; the cumulative mileage display range is: (0~999999)km, and the cumulative mileage cannot be reset by the customer; the total mileage display accuracy is 1km.

[0088] b. The instrument calculates the vehicle's mileage based on the vehicle speed signal, and the calculation error cannot be greater than ±2% (QC / T727-2007) "Automobile and Motorcycle Instruments".

[0089] c. When esc_vehicleSpeedValid = 0x0, the vehicle speed is invalid, where esc_vehicleSpeedValid indicates whether the vehicle speed is valid, and esc_vehicleSpeed ​​is the actual vehicle speed (in km / h) obtained from ESC (electronic stability control system); the mileage is calculated based on the vehicle speed equal to 0km / h; when esc_vehicleSpeedValid = 0x1, the vehicle speed is valid and responds to the vehicle speed signal. The mileage is calculated based on the actual vehicle speed.

[0090] Among them, 0x0 and 0x1 are respectively a value in the signal matrix table in the vehicle, 0x0 is used to represent that the vehicle speed is invalid, and 0x1 is used to represent that the vehicle speed is valid.

[0091] d. When the vehicle speed esc_vehicleSpeed>0x12C0, the meter calculates the mileage as an invalid value, that is, the mileage is calculated as 0km / h. Among them, 0x12C0 is used to represent the maximum speed value in the vehicle speedometer. If the vehicle speed exceeds the maximum speed value, the current speed cannot be read, and the speed is reset to zero.

[0092] e. By eliminating some invalid processing methods through the above method, the mileage can be accurately calculated from the vehicle speed signal, mileage value = s + Σ (V1 + V2 + V3 ... + Vn) T / 3400

[0093] Note: s = last mileage.

[0094] The vehicle speed used when calculating the mileage value V1…Vn, in km / h;

[0095] Calculation formula: Vn=V*1.02 (V: actual vehicle speed esc_vehicleSpeed).

[0096] T = period of vehicle speed signal, unit: s.

[0097] 5. Crawler software: Use Python tools to write crawler software to search for hot posts, pictures, videos and other information about the company's brand market vehicles.

[0098] 6. Image recognition: Image recognition algorithm is used to perform image clarification, information recognition, and keyword extraction on hot posts and pictures searched by crawler software.

[0099] 7. Video natural language processing: Use natural language processing algorithms to identify key information and extract keywords in videos.

[0100] 8. Intelligent retrieval: Use the keyword information extracted by image recognition and video natural language processing to compare with the cloud database of all vehicles of the company's brand to retrieve the vehicles that meet the requirements. The keyword information includes target features, time period and location area. By inputting the keyword data into the cloud for joint retrieval, the vehicle identification code of the target vehicle is determined to obtain vehicles that meet the retrieval requirements.

[0101] 9. Data analysis: Conduct data analysis on the vehicles retrieved intelligently to form a complete analysis report on the time point of the event.

[0102] See also Figure 2 , is a schematic diagram of processing target features in a vehicle tracing analysis method in a hot event in an embodiment of the present application; including but not limited to vehicle model, paint color and appearance data of abnormal parts, which will not be repeated here.

[0103] Compared with the prior art, the beneficial effects of this technical solution are:

[0104] First, use image recognition to locate the location of the incident, and use data analysis to determine the condition of the vehicle at the time of the incident and identify the vehicle.

[0105] Secondly, it can not only locate the location of the incident, but also more accurately locate the VIN of the target vehicle. Through background data analysis, it can determine the condition of the vehicle involved in the incident and the status of the vehicle at the time, and fully grasp all key data of the vehicle at the time of the incident.

[0106] Third, this application can quickly and easily accurately locate hot vehicles, effectively enhancing brand reputation.

[0107] See also Figure 3 , is a schematic diagram of the structure of a vehicle tracing analysis system in a hot event in an embodiment of the present application, including:

[0108] The feature determination module 301 is used to obtain hot events, identify target vehicles in images or videos associated with the hot events, and determine target features, where the target features include appearance data of vehicle model, paint color, and abnormal parts, where the abnormal parts are parts of the target vehicle damaged due to an accident;

[0109] A hotspot search module 302 is used to search based on the release information of the hotspot event to determine the time period and location area where the hotspot event occurs;

[0110] A search and tracing module 303 is used to input the target feature, the time period and the location area into the cloud for joint search to determine the vehicle identification code to which the target vehicle belongs;

[0111] The report generation module 304 is used to retrieve the vehicle information corresponding to the target vehicle in the previous and subsequent time periods according to the vehicle identification code, perform data analysis on the vehicle information, and determine an analysis report. The vehicle information includes instrument data, central control screen display data, and appearance data.

[0112] Among them, the feature determination module 301 is used to execute the aforementioned step 101, the hotspot search module 302 is used to execute the aforementioned step 102, the search and tracing module 303 is used to execute the aforementioned step 103, and the report generation module 304 is used to execute the aforementioned step 104.

[0113] The above modules correspond to the steps of the vehicle tracing analysis method in hot events, and their execution process corresponds to the corresponding steps, which will not be described in detail here.

[0114] The vehicle tracing and analysis system in hot events provided by the embodiment of the present disclosure is used to identify the target vehicle in the image or video associated with the hot event and determine the target features, which is conducive to identifying the target features of the vehicle related to the event and facilitating subsequent rapid positioning; based on the release information of the hot event, a search is performed to determine the time period and location area where the hot event occurred, which can narrow the search scope; the target features, the time period and the location area are input into the cloud for joint retrieval to determine the vehicle identification code to which the target vehicle belongs; in this way, through the cloud joint retrieval and comprehensive vehicle information retrieval, accurate identification of the target vehicle is achieved, which is conducive to rapid and accurate public relations response, positive maintenance of vehicle brand value, and at the same time, the efficiency of event processing is improved, and the vehicle information corresponding to the target vehicle in the time period before and after is retrieved according to the vehicle identification code, and the vehicle information is analyzed to determine the analysis report, which can not only analyze the cause of the accident caused by the vehicle, but also provide more comprehensive vehicle information support for the owner, which helps to reduce potential risks and improve vehicle safety.

[0115] An embodiment of the present disclosure further provides an electronic device, including: a processor and a memory; the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the electronic device executes the above method.

[0116] Figure 4 The structure diagram of the computer system suitable for implementing the electronic device of the embodiment of the present application is shown. It should be noted that: Figure 4 The computer system 400 of the electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0117] like Figure 4 As shown, the computer system 400 includes a central processing unit (CPU) 401, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 402 or the program loaded from the storage part 408 to the random access memory (RAM) 403, such as executing the method in the above embodiment. In the RAM 403, various programs and data required for system operation are also stored. The CPU 401, the ROM 402 and the RAM 403 are connected to each other through the bus 404. The input / output (I / O) interface 405 is also connected to the bus 404.

[0118] The following components are connected to the I / O interface 405: an input section 406 including a keyboard, a mouse, etc.; an output section 407 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to the I / O interface 405 as needed. A removable medium 411, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 410 as needed so that a computer program read therefrom is installed into the storage section 408 as needed.

[0119] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software may depend on the specific application and design constraints of the technical solution. Technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the embodiments of the present disclosure. Technicians can clearly understand that for the convenience and simplicity of description, the specific working process of the systems, devices and units described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0120] The above embodiments are merely illustrative of the principles and effects of the present application and are not intended to limit the present application. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present application. Therefore, all equivalent modifications or changes made by a person of ordinary skill in the art without departing from the spirit and technical ideas disclosed in the present application shall still be covered by the claims of the present application.

Claims

1. A vehicle tracing analysis method in a hot event, characterized in that: include: Acquire hot events, identify target vehicles in images or videos associated with the hot events, and determine target features, where the target features include appearance data including vehicle model, paint color, and abnormal parts, where the abnormal parts are parts of the target vehicle that are damaged due to an accident; Search based on the release information of the hot event to determine the time period and location area where the hot event occurs; Input the target feature, the time period and the location area into the cloud for joint retrieval to determine the vehicle identification code to which the target vehicle belongs; The vehicle information corresponding to the target vehicle in the time periods before and after is retrieved according to the vehicle identification code, data analysis is performed on the vehicle information, and an analysis report is determined. The cloud stores the vehicle information including instrument data, central control screen display data and appearance data.

2. The method according to claim 1, characterized in that Before getting hot events, it also includes: Get the CAN message signal uploaded by the vehicle; Parsing the CAN message signal and performing format conversion to generate the message signal in a preset format; The message data in a preset format is clustered according to the business flow using a preset data model to generate indicator data with labels and store them in the cloud. The type of the preset data model corresponds to the type of the indicator data, and the type of the indicator data is associated with the instrument data, the central control screen display data and the appearance data.

3. The method according to claim 2, characterized in that Inputting the target feature, the time period and the location area into the cloud for joint retrieval to determine the vehicle identification code to which the target vehicle belongs, including: Input the target feature, the time period and the location area into a preset search model in the cloud for joint search to determine the vehicle identification code to which the target vehicle belongs; The index data of the vehicle is mapped to the vehicle information to form a training set, and the preset retrieval model is obtained by training a convolutional neural network using the training set.

4. The method according to claim 3, characterized in that: Inputting the target feature, the time period and the location area into a preset search model in the cloud for joint search to determine the vehicle identification code to which the target vehicle belongs, including: Extract the word vector of each text sample with a label, convert the word vector into a two-dimensional matrix, input it into a convolutional neural network, and train to obtain a preset retrieval model; wherein each text sample stores the indicator data and the vehicle information of the same vehicle; Combining the target feature, the time period and the location area into a search text, segmenting the search text, filtering modal particles, and generating a search word vector; The search word vector is input into the preset retrieval model to obtain the similarity between the texts, and the texts are sorted from high to low according to the similarity to determine the vehicle identification code to which the target vehicle belongs.

5. The method according to claim 2, characterized in that: Also includes: If it is monitored that any of the indicator data is missing within a certain period of time, an analysis is performed based on the vehicle's current speed information, mileage, location area changes before and after, and the data missing period, and the indicator data is supplemented according to the analysis results to obtain complete indicator data.

6. The method according to claim 1, characterized in that Perform data analysis on the vehicle information to determine an analysis report, including: Determine the change status of the brake pedal, accelerator pedal, vehicle speed, steering wheel direction, seat belt status, and engine speed of the vehicle in the instrument data within the time period; According to the external environment and traffic conditions of the vehicle in the location area at that time, performing data analysis on the change state of the instrument data of the vehicle to determine whether there is any misoperation in the driver's current driving strategy; If the driver does not make any wrong operation, generating a first analysis report of the vehicle status analysis; If the driver has made an erroneous operation, a second analysis report of the vehicle status and the erroneous operation analysis is generated.

7. The method according to claim 6, characterized in that After the analysis report is determined, it also includes: Sending the analysis report to the staff, and conducting a return visit to the owner of the target vehicle based on the analysis report to generate a return visit record; The defects of the vehicle are improved and optimized through the return visit records to improve the quality of the vehicle.

8. The method according to any one of claims 1 to 7, characterized in that: Also includes: Obtain a standard vehicle image corresponding to the vehicle model, compare the standard vehicle image with the image associated with the hot event, and determine the damaged parts on the target vehicle through differential comparison, and treat the damaged parts as abnormal parts.

9. A vehicle tracing and analysis system for hot events, characterized in that: include: A feature determination module is used to obtain hot events, identify target vehicles in images or videos associated with the hot events, and determine target features, where the target features include vehicle model, paint color, and appearance data of abnormal parts, where the abnormal parts are parts of the target vehicle that are damaged due to an accident; A hotspot search module, used to search based on the release information of the hotspot event to determine the time period and location area where the hotspot event occurs; A search and tracing module, used to input the target feature, the time period and the location area into the cloud for joint search to determine the vehicle identification code to which the target vehicle belongs; A report generation module is used to retrieve the vehicle information corresponding to the target vehicle in the time period before and after according to the vehicle identification code, perform data analysis on the vehicle information, and determine an analysis report. The vehicle information includes instrument data, central control screen display data and appearance data.

10. An electronic device, characterized in that: include: Processor and memory; The memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the electronic device executes the method according to any one of claims 1 to 8.

Images