Method, device, medium and program product for detecting traffic accidents of vehicles

Abstract

The present invention provides a method, device, medium and program product for discovering a traffic accident of a vehicle, the method comprising: receiving positioning feature data and alarm feature data of a vehicle within a first predetermined time range; determining the braking deceleration feature data and alarm feature score data of the vehicle within the first predetermined time range according to the received positioning feature data and alarm feature data; the braking deceleration feature data comprises: braking deceleration feature data and braking distance feature data; the alarm feature score data comprises: feature score data of each alarm type in the predetermined alarm type; inputting the braking deceleration feature data and alarm feature score data of the vehicle within the first predetermined time range into a pre-trained accident discovery machine learning model, and obtaining the result output by the accident discovery machine learning model; wherein the result comprises: normal driving and accident. By using the above technical solution, it is possible to timely and accurately discover whether a vehicle is in an accident.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and in particular to a method, device, medium and program product for detecting a traffic accident involving a vehicle. Background Art

[0002] Currently, existing technologies mainly use monitoring equipment or vehicle status information, driver's vital signs and other technical means to detect whether a vehicle has been involved in a traffic accident. It does not conduct in-depth analysis of the data from the on-board safety terminal, and has the problems of high technical application costs, difficulty in popularization and low accuracy in identifying vehicle accidents. Summary of the invention

[0003] The embodiments of the present invention provide a method, device, medium and program product for detecting a traffic accident of a vehicle, so as to timely detect whether a traffic accident of the vehicle occurs and improve the recognition accuracy of the traffic accident of the vehicle.

[0004] In order to achieve the above object, a method for detecting a traffic accident of a vehicle is provided, comprising:

[0005] Receive positioning feature data and alarm feature data of a vehicle within a first predetermined time range; the positioning feature data includes: vehicle frame number, positioning latitude and longitude, positioning time, driving speed and driving mileage data; the alarm feature data includes: alarm time and alarm number of predetermined alarm types; the predetermined alarm types include: forward collision alarm, vehicle distance alarm, vehicle collision alarm and / or sudden deceleration alarm;

[0006] Determine the braking deceleration characteristic data and the alarm characteristic score data of the vehicle within the first predetermined time range according to the received positioning characteristic data and the alarm characteristic data; the braking deceleration characteristic data includes: braking deceleration characteristic data and braking distance characteristic data; the alarm characteristic score data includes: characteristic score data of each alarm type in the predetermined alarm type;

[0007] Input the braking and deceleration feature data and alarm feature score data of the vehicle within the first predetermined time range into a pre-trained accident detection machine learning model, and obtain the results output by the accident detection machine learning model; the results include: normal driving and accident occurrence; wherein the accident detection machine learning model is trained in advance using historical data of the selected vehicle; the historical data includes: the positioning feature data, alarm feature data, braking and deceleration feature data and alarm feature score data of the vehicle within the first predetermined time range before the time point of the accident occurrence, and the positioning feature data, alarm feature data, braking and deceleration feature data and alarm feature score data of the vehicle within the first predetermined time range when the vehicle is driving normally, i.e., when no accident occurs, that is, the historical data includes the accident occurrence data of the vehicle when the accident occurs and the normal driving data of the vehicle when driving normally;

[0008] Wherein, determining the braking deceleration characteristic data of the vehicle within the first predetermined time range includes:

[0009] According to the driving speeds v1 and v2 corresponding to the two selected driving time points t1 and t2, the braking deceleration dt between t1 and t2 is determined, where: Among them, t2>t1; when dt is a positive number, it indicates that the vehicle is decelerating; when dt is a negative number, it indicates that the vehicle is accelerating; when dt is zero, it indicates that the speed of the vehicle has not changed or the vehicle is stopped;

[0010] According to the driving mileages m1 and m2 corresponding to the two selected driving time points t1 and t2, the braking distance Δm between t1 and t2 is determined, wherein Δm=m2-m1; when Δm=0, it indicates that the vehicle has stopped driving;

[0011] Wherein, determining the alarm feature score data of the vehicle within the first predetermined time range includes:

[0012] Acquire the alarm time and the number of alarms of the predetermined alarm type within a second predetermined time range from the driving start time selected within the first predetermined time range, and determine the average value of the alarms within the second predetermined time range. in, The selected driving start time is the earliest time when the alarm occurs within the second predetermined time range;

[0013] The normalized index of each alarm type in the predetermined alarm type As the alarm correction coefficient of the alarm type within the second predetermined time range, wherein, Among them, j corresponds to alarm type j, b jis the number of alarms of alarm type j occurring within the second predetermined time range, and k is the number of alarm types;

[0014] For each alarm type j, the alarm correction coefficient corresponding to the alarm type is used. Determine the alarm feature score c of the alarm type within the second predetermined time range j ,in:

[0015]

[0016] Preferably, in the method, the first predetermined time range is 1 to 5 minutes; and / or the second predetermined time range is a predetermined N days; and N is greater than 2.

[0017] Preferably, in the method, the positioning feature data and the alarm feature data are received from a vehicle-mounted safety terminal; the vehicle-mounted safety terminal collects the positioning feature data and the alarm feature data of the vehicle in real time.

[0018] Preferably, in the method described above, the accident discovery machine learning model samples a random forest algorithm to perform machine learning; and the historical data for training the accident discovery machine learning model is pre-labeled.

[0019] Preferably, the method described herein further includes the steps of cleaning and normalizing the received positioning feature data and alarm feature data before determining the braking and deceleration feature data and alarm feature score data of the vehicle within the first predetermined time range based on the received positioning feature data and alarm feature data.

[0020] Preferably, the method further comprises:

[0021] When the result is an accident, an accident occurrence report for reporting is generated based on the positioning feature data, alarm feature data, braking and deceleration feature data and alarm feature score data of the vehicle involved in the accident within the first predetermined time range.

[0022] Preferably, the method further comprises:

[0023] The vehicle accident emergency rescue system takes rescue measures for the vehicle involved in the accident according to the accident occurrence report.

[0024] On the other hand, a device for detecting a traffic accident of a vehicle is also provided, comprising a memory and a processor, wherein the memory stores at least one program, and the at least one program is executed by the processor to implement the steps of any of the above methods. Preferably, the device is an electronic device.

[0025] On the other hand, a computer-readable storage medium is provided, wherein at least one program is stored in the storage medium, and the at least one program is executed by a processor to implement the steps of any of the methods described above.

[0026] On the other hand, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the steps of any of the methods described above are implemented.

[0027] The above technical solution has the following technical effects:

[0028] The embodiment of the present invention obtains high-level features that can characterize whether a vehicle has an accident based on vehicle positioning feature data and alarm feature data within a predetermined time range: the vehicle's braking and deceleration feature data and alarm feature score data, and inputs the above data into an accident detection machine learning model trained with labeled historical data to obtain a classification result of whether the vehicle has an accident. The method of the embodiment of the present invention can be used to timely discover whether a vehicle has an accident, and improve the accuracy of identifying vehicle accidents; thereby, emergency rescue can be carried out in a timely manner and the driver's personal and property safety can be guaranteed. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 The figure is a flow chart of a method for detecting a traffic accident involving a vehicle according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] To further illustrate the various embodiments, the present invention provides drawings. These drawings are part of the disclosure of the present invention, which are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, a person of ordinary skill in the art should be able to understand other possible implementations and advantages of the present invention. The components in the figures are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0031] The present invention will now be further described with reference to the accompanying drawings and specific implementation methods.

[0032] Embodiment 1:

[0033] Figure 1 FIG. 1 is a flow chart of a method for detecting a vehicle traffic accident according to an embodiment of the present invention. Figure 1 The method for discovering a traffic accident involving a vehicle in this embodiment includes:

[0034] Receive the positioning feature data and alarm feature data of the vehicle within a first predetermined time range; the positioning feature data includes: frame number, positioning longitude and latitude, positioning time, driving speed and driving mileage data; the alarm feature data includes: alarm time and alarm number of predetermined alarm types; the predetermined alarm types include: forward collision alarm, too close distance alarm, vehicle collision alarm and / or sudden deceleration alarm; in a specific implementation, the above positioning feature data and alarm feature data are received from the vehicle-mounted safety terminal; the vehicle-mounted safety terminal is installed on the vehicle, and the positioning feature data and alarm feature data of the vehicle are collected in real time and stored; in a specific implementation, the positioning feature data and alarm feature data of the vehicle within the first predetermined time range are a sequence of positioning feature data and alarm feature data arranged according to time; in a specific implementation, the first predetermined time range is determined according to the required number of data points and the processing power of the hardware; preferably, the range of the first predetermined time is 1 to 5 minutes;

[0035] Determine the braking deceleration characteristic data and the alarm characteristic score data of the vehicle within the first predetermined time range according to the received positioning characteristic data and the alarm characteristic data; the braking deceleration characteristic data includes: braking deceleration characteristic data and braking distance characteristic data; the alarm characteristic score data includes: characteristic score data of each alarm type in the predetermined alarm type;

[0036] Input the braking and deceleration feature data and alarm feature score data of the vehicle within a first predetermined time range into a pre-trained accident detection machine learning model, and obtain the results output by the accident detection machine learning model; wherein the results include: normal driving and accidents; wherein the accident detection machine learning model is trained in advance using historical data of the selected vehicle; the historical data includes: the positioning feature data, alarm feature data, braking and deceleration feature data and alarm feature score data of the vehicle within the first predetermined time range before the time point of the accident, and the positioning feature data, alarm feature data, braking and deceleration feature data and alarm feature score data of the vehicle within the first predetermined time range when the vehicle is driving normally without an accident;

[0037] Wherein, determining the braking deceleration characteristic data of the vehicle within the first predetermined time range includes:

[0038] According to the driving speeds v1 and v2 corresponding to the two selected driving time points t1 and t2, the braking deceleration dt between t1 and t2 is determined, where: Among them, t2>t1; when dt is a positive number, it indicates that the vehicle is decelerating; when dt is a negative number, it indicates that the vehicle is accelerating; when dt is zero, it indicates that the speed of the vehicle has not changed or the vehicle is stopped;

[0039] According to the driving mileages m1 and m2 corresponding to the two selected driving time points t1 and t2, the braking distance Δm between t1 and t2 is determined, wherein Δm=m2-m1; when Δm=0, it indicates that the vehicle has stopped driving;

[0040] Wherein, determining the alarm feature score data of the vehicle within the first predetermined time range includes:

[0041] Acquire the alarm time and the number of alarms of the predetermined alarm type within a second predetermined time range from the driving start time of the first predetermined time range, and determine the alarm average value within the second predetermined time range in,

[0042] The normalized index of each alarm type in the predetermined alarm type As the alarm correction coefficient of the alarm type within the second predetermined time range, wherein, Among them, j corresponds to alarm type j, b k is the number of alarms of alarm type j occurring within the second predetermined time range, and k is the number of alarm types;

[0043] For each alarm type j, use the alarm correction coefficient corresponding to the alarm type. Determine the alarm feature score c of the alarm type within the second predetermined time range j ,in:

[0044]

[0045] Embodiment 2:

[0046] The method for discovering a traffic accident involving a vehicle in this embodiment of the present invention comprises the following steps:

[0047] S1. The on-board safety terminal, such as the on-board safety terminal of the new energy online car-hailing vehicle, receives vehicle data in real time and cleans the received vehicle data, including: processing duplicate data, processing outliers and processing missing values; wherein the above-mentioned vehicle data includes: vehicle positioning feature data and vehicle alarm feature data of predetermined alarm types; wherein the vehicle positioning feature data includes: frame number, positioning longitude and latitude, positioning time, driving speed and driving mileage; the predetermined alarm types include: forward collision alarm, vehicle-too-close alarm, vehicle collision alarm and sudden deceleration alarm; the vehicle alarm feature data includes: alarm time, number of forward collision alarms, number of vehicle-too-close alarms, number of vehicle collision alarms, number of sudden deceleration alarms; wherein the alarm time is the time when the alarm occurs.

[0048] S2, performing data analysis on the above-mentioned vehicle data received in real time according to a sequence within a predetermined time range to calculate the corresponding braking deceleration feature data and alarm feature score data; in a specific implementation, the real-time received vehicle positioning data and alarm feature data are aggregated into a vehicle original positioning data set and a vehicle alarm original data set; data analysis and calculation are performed on the vehicle original positioning data set and the vehicle alarm original data set; wherein, the relevant high-level features of the vehicle original positioning data set are calculated according to the vehicle original positioning data set: braking deceleration feature data and braking distance feature data; and they are aggregated into a vehicle braking deceleration feature data set; data analysis is performed on the vehicle alarm original data set, and the relevant high-level features, namely, alarm feature score data, are calculated according to the alarm calibration coefficient and weighting algorithm within a certain time range of each vehicle, including: forward collision alarm feature score data, too close distance alarm feature score data, vehicle collision alarm feature score data and sudden deceleration alarm feature score data, and they are aggregated into a vehicle alarm feature score data set. In the specific implementation, the vehicle braking deceleration feature data set and the vehicle alarm feature score data set presented in the form of data series within a certain time range are normalized to eliminate the influence of data of different dimensions, and obtain the standard normalized vehicle braking deceleration feature data set and the standard normalized vehicle alarm feature score data set. At the same time, the original vehicle positioning data set, the original vehicle alarm data set, the vehicle braking deceleration feature data set, the vehicle alarm feature score data set, the standard normalized vehicle braking deceleration feature data set, and the standard normalized vehicle alarm feature score data set within a certain time range are stored using a time series database.

[0049] S3, for the braking and deceleration feature data and the alarm feature score data calculated above, use a pre-trained accident detection machine learning model to classify the new energy online car-hailing vehicle accident machine learning model to obtain a corresponding classification result, that is, whether the vehicle has an accident;

[0050] In the specific implementation, historical data collected in advance and presented in a serial manner within a certain time range predetermined before an accident occurs in a vehicle such as a new energy online-hailing vehicle, and historical data collected in advance and presented in a serial manner within a certain time range predetermined when the vehicle is driving normally are used, and the above-mentioned analysis method of the embodiment of the present invention is used to obtain a historical vehicle braking and deceleration feature data set, a historical vehicle alarm feature score data set, a historical standard-normalized vehicle braking and deceleration feature data set, and a historical standard-normalized vehicle alarm feature score data set; the historical standard-normalized vehicle braking and deceleration feature data set and the historical standard-normalized vehicle alarm feature score data set are used and the random forest algorithm is applied for machine learning, thereby training a binary classification accident detection machine learning model for identifying whether an accident has occurred and whether an accident has not occurred; the obtained standard-normalized vehicle braking and deceleration feature data set and the standard-normalized vehicle alarm feature score data set are input into the above-mentioned accident detection machine learning model, and the result of whether a vehicle such as a new energy online-hailing vehicle has an accident is output.

[0051] S4. When the accident detection machine learning model outputs that a vehicle has an accident, the original positioning data set, vehicle alarm original data set, vehicle braking and deceleration feature data set, and vehicle alarm feature score data set of the vehicle are obtained, and an accident occurrence report is generated.

[0052] S5, the vehicle accident emergency rescue system takes rescue measures for the vehicle involved in the accident according to the reported accident report; for example, it responds promptly to carry out emergency rescue for the driver involved in the accident to ensure the safety of the driver's life and property.

[0053] Based on the above technical solution, the specific implementation of each step is described below.

[0054] A specific implementation of step S1 includes:

[0055] First, use the vehicle safety terminal protocol that complies with standards, regulations and specifications to connect to the vehicle safety terminal equipment of the new energy online car-hailing service to receive vehicle data;

[0056] Then, for the received duplicate vehicle data, the duplicate data is deleted and only one piece of data is retained to ensure the uniqueness of the vehicle data at each positioning time; however, the processing of duplicate vehicle data is not limited to retaining multiple pieces of data;

[0057] Next, the received vehicle abnormal value data is filled with the average data of the vehicle in the last hour; but it is not limited to deleting the abnormal data row or using manual calibration, filling with other statistical methods, filling with previous and next data, etc.;

[0058] Finally, the missing value data of the received vehicles is filled with default values, such as using zero filling, but not limited to manual calibration, deleting data rows, using other statistical methods to fill, etc.

[0059] A specific implementation method of step S2 includes:

[0060] According to the vehicle data received in real time, the vehicle positioning feature data and vehicle alarm feature data with a sequence length of N within a certain time range are analyzed and calculated to obtain high-level features, and the vehicle braking and deceleration feature data set and the vehicle alarm feature score data set are obtained, and the standard normalized vehicle braking and deceleration feature data set and the standard normalized vehicle alarm feature score data set are obtained through standard normalization. Preferably, N=5, and data is collected every 30 seconds; then the time range is 2 and a half minutes. In other implementations, different N and collection time can be selected according to actual needs.

[0061] The calculation of braking deceleration characteristic data includes:

[0062] First, the driving speed unit transmitted by the device is uniformly converted into driving speed data of standard unit per meter per second (m / s), and the speed difference is calculated by subtracting the previous driving speed v1 from the next driving speed v2. If it is a negative number, it means deceleration occurs, otherwise it is acceleration. The driving speed difference calculation method is: Δv = v1-v2, where Δv is the driving speed difference (per meter per second) and retains one decimal place;

[0063] Next, the driving time difference (seconds) is calculated by subtracting the next positioning time from the previous positioning time, which represents the number of seconds the vehicle has spent driving between the two positioning points. The vehicle driving time difference (seconds) is calculated as: Δt = t2-t1, where Δt is the driving time difference (seconds), t2 is the next driving time, and t1 is the previous driving time;

[0064] Finally, the vehicle braking deceleration is calculated based on the calculated driving speed difference and driving time difference, and the calculation method is: Where dt is the vehicle braking deceleration. If it is a positive number, it means the vehicle is decelerating. Otherwise, it is accelerating. If it is zero, it means the vehicle speed has not changed or has stopped. Δu is the driving speed difference (meters per second) and is rounded to two decimal places. Δt is the driving time difference (seconds).

[0065] An example of calculating vehicle braking deceleration characteristic data is shown in Table 1:

[0066]

[0067] Table 1

[0068] The calculation of braking distance characteristic data includes:

[0069] The driving mileage positioning transmitted by the equipment is uniformly converted into standard driving mileage data in meters (m), and the driving mileage (m) is calculated by subtracting the latter driving mileage data from the previous driving mileage, which is called the braking distance. The calculation method of the braking distance (m) is: Δm = m2-m1, where Δm is the braking distance (m), m2 is the latter driving mileage (m), and m1 is the former driving mileage (m). If the braking distance (m) is zero, it means that the vehicle has stopped.

[0070] An example of calculation of braking distance characteristic data is shown in Table 2:

[0071] Serial number Positioning time Driving mileage (km) Braking distance(m) 1 2024-06-21 05:48:24 27440.1 2 2024-06-21 05:48:27 27440.2 100 3 2024-06-21 05:48:32 27440.4 200 4 2024-06-21 05:49:03 27440.5 100

[0072] Table 2

[0073] The calculation method and steps of the corresponding alarm feature score of the vehicle alarm feature score data set are as follows:

[0074] In this example, the vehicle alarm feature score data set includes: forward collision alarm feature score data, close vehicle distance alarm feature score data, vehicle collision alarm feature score data and sudden deceleration alarm feature score data, which use the alarm calibration coefficient within a certain time range of each vehicle and apply a weighted algorithm to calculate the score.

[0075] First, from the vehicle alarm feature data within a certain time range, the alarm time of the first alarm is obtained as the driving start time within the time range.

[0076] Next, the alarm time and the corresponding alarm type sequence historical data of the number of vehicle forward collision alarms, the number of vehicle-to-vehicle close distance alarms, the number of vehicle collision alarms, and the number of sudden deceleration alarms are obtained from the driving start time of the sequence within a certain time range, such as seven days, and the alarm average value of all the listed alarm type sequence data within the seven-day historical range is calculated: in The average number of alarms of all alarm types listed in the seven-day historical range, a j is the number of historical alarms of the corresponding alarm type j, M is the number of time points corresponding to the listed historical alarm times of the alarms, and k is the number of all listed alarm types;

[0077] Next, the alarm time and the corresponding alarm type sequence historical data of the number of vehicle forward collision alarms, the number of vehicle-to-vehicle close distance alarms, the number of vehicle collision alarms, and the number of sudden deceleration alarms are obtained from the above driving start time sequence within a certain time range for several days, such as seven days, and the normalization function, i.e., the softmax operation, is used to calculate the alarm calibration coefficient corresponding to each alarm type within a predetermined time range:

[0078] The softmax operation formula is:

[0079]

[0080] here, The numerator is the total number of a certain alarm type within a certain time range and The denominator is the sum of the total number of alarm types listed within a certain time range and The sum of the subtracted exponential functions is thus can be considered as the correct probability distribution;

[0081] in Here j is the number of alarms of a certain alarm type occurring within a certain time range, and k is the number of listed alarm types;

[0082] The calculated normalized function value is used as the corresponding alarm correction coefficient, that is, the alarm calibration coefficient corresponding to the corresponding alarm type of the data sequence within a certain time range is obtained according to the softmax operation;

[0083] Finally, the corresponding alarm feature score is obtained by weighted calculation based on the calculated alarm calibration coefficient and the total number of a certain alarm type within a certain time range. In the specific implementation, the obtained data is aggregated into a vehicle alarm feature score data set; its calculation formula is: c j is the feature score of the corresponding alarm type j, The alarm calibration coefficient corresponding to the alarm type. In a specific implementation, it also includes updating In one specific implementation, the selected historical data is used to update the The value of .

[0084] Take the data of the driving start time of a certain vehicle within a certain time range and the previous seven days as an example to calculate the vehicle alarm feature score:

[0085] Assume that the time range of a vehicle's current driving time series is from 2024-06-21 23:59:59 to 2024-06-22 00:09:59, its driving start time is 2024-06-21 23:59:59, and the previous seven days range is from 2024-06-14 23:59:59 to 2024-06-21 23:59:59;

[0086] The number of historical alarms of corresponding alarm types generated within the seven-day period from 2024-06-14 23:59:59 to 2024-06-21 23:59:59 is shown in Table 3:

[0087]

[0088]

[0089] Table 3

[0090] The average alarm value in the seven-day range from 2024-06-14 23:59:59 to 2024-06-21 23:59:59 is: 2.3 = 23 / 10;

[0091] An example of the calculation process of the corresponding alarm type alarm number and alarm calibration coefficient and the corresponding alarm type feature score for the current driving time series of a vehicle ranging from 2024-06-21 23:59:59 to 2024-06-22 00:09:59 is shown in Table 4:

[0092]

[0093]

[0094] Table 4

[0095] An example is given to illustrate the calculation of the forward collision warning correction factor:

[0096] 1.63 / (1.63+1.09+1.63+1.63)=0.27;

[0097] A specific implementation of step S3 includes:

[0098] After data preprocessing and data analysis and calculation, the vehicle braking and deceleration feature data set normalized by historical standards and the vehicle alarm feature score data set normalized by historical standards are obtained. The data have been distinguished between positive and negative labels, that is, the positive and negative labels are pre-labeled; among them, the positive label is normal driving, and the negative label is an accident. The data of the positive and negative labels are used to calculate the proportion of the two categories. The data of the negative label is less than that of the positive label; in this example, the training uses the SMOTE algorithm to perform data oversampling processing to generate negative label data to ensure that the distribution ratio of positive and negative data is consistent.

[0099] In this example, the random forest algorithm is used for machine learning. The random forest algorithm builds multiple trees based on decision trees and uses the bagging mechanism to improve the generalization ability of the model. Through the self-service sampling method, each decision tree training uses sample set randomness and randomly selected feature subsets to increase the diversity of the model. In this example, the number of trees for the random forest algorithm training is controlled between 200 and 300. The maximum depth value of the decision tree is selected using network search to select the best value. The maximum number of features is set to the square root of the total number of features, and random states are added. The confusion matrix is ​​used to evaluate this model, so as to train and obtain a binary classification accident detection machine learning model for identifying accidents and non-accidents, such as the new energy online car-hailing accident detection model. The standard normalized vehicle braking and deceleration feature data set and the standard normalized vehicle alarm feature score data set are obtained in real time and input into the new energy online car-hailing accident detection model to output the result of whether the new energy online car-hailing accident occurred.

[0100] A specific implementation of step S4 includes:

[0101] The accident discovery machine learning model output identifies that a vehicle, such as a new energy online-hailing vehicle, has an accident, and obtains the corresponding vehicle original positioning data set, vehicle alarm original data set, vehicle braking and deceleration feature data set, and vehicle alarm feature score data set. Based on the acquired data sets, an accident report, such as a new energy online-hailing vehicle accident report, is generated.

[0102] In a specific implementation, the report displays the frame numbers of new energy online-hailing vehicles that have accidents within a certain time range, the range of longitude and latitude of the accident and convert it into an address, the range of the time of the accident, the alarm time range, the total number of forward collision alarms, the total number of vehicle-to-vehicle distance alarms, the total number of vehicle collision alarms, and the total number of sudden deceleration alarms. The basic vehicle information and driver-related information are matched according to the frame number of the new energy online-hailing vehicle, and an accident report of the new energy online-hailing vehicle is presented, and the accident report data is reported to the vehicle accident emergency rescue system at the same time.

[0103] A specific implementation of step S5 includes:

[0104] Upon receiving the accident report data of the new energy online ride-hailing vehicle, the vehicle accident emergency rescue system will immediately display and sound a reminder that an accident has occurred with the new energy online ride-hailing vehicle. The relevant on-duty personnel will obtain the driver's basic information and the circumstances of the accident through the accident report data of the new energy online ride-hailing vehicle, and contact the driver or the national alarm and emergency agency in a timely manner to provide timely assistance to the driver and protect the driver's life and property.

[0105] Embodiment three:

[0106] The present invention also provides a device for detecting a traffic accident of a vehicle, the device comprising a memory and a processor, the memory storing at least one program, the at least one program being executed by the processor to implement the steps of any of the methods described above. In a specific implementation, the device is an electronic device; in a specific implementation, the device is a vehicle-mounted terminal.

[0107] Further, as an executable solution, the device for detecting a traffic accident of a vehicle may be a computer unit, and the computer unit may be a computing device such as a desktop computer, a notebook, a PDA, and a cloud server. The computer unit may include, but is not limited to, a processor and a memory. Those skilled in the art may understand that the composition structure of the above-mentioned computer unit is only an example of a computer unit and does not constitute a limitation on the computer unit, and may include more or less components than the above-mentioned, or a combination of certain components, or different components. For example, the computer unit may also include input and output devices, network access devices, buses, etc., which are not limited in the embodiments of the present invention.

[0108] Further, as an executable solution, the processor may be a central processing unit (CPU), 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. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the computer unit, and various interfaces and lines are used to connect the various parts of the entire computer unit.

[0109] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the computer unit by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system and at least one application required for a function; the data storage area can store data created according to the use of the mobile phone, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0110] Embodiment 4:

[0111] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above method in the embodiment of the present invention are implemented.

[0112] If the module / unit integrated in the computer unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory) and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction.

[0113] Embodiment five:

[0114] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the steps of the method described above are implemented.

[0115] Although the present invention has been specifically shown and described in conjunction with the preferred embodiments, it should be understood by those skilled in the art that various changes may be made to the present invention in form and details without departing from the spirit and scope of the present invention as defined by the appended claims, all of which are within the scope of protection of the present invention.

Claims

1. A method for discovering a traffic accident involving a vehicle, characterized in that: include: Receiving positioning feature data and alarm feature data of a vehicle within a first predetermined time range; The positioning feature data includes: vehicle frame number, positioning latitude and longitude, positioning time, driving speed and driving mileage data; the alarm feature data includes: alarm time and alarm number of predetermined alarm types; the predetermined alarm types include: forward collision alarm, vehicle distance alarm, vehicle collision alarm and / or sudden deceleration alarm; Determine the braking deceleration characteristic data and the alarm characteristic score data of the vehicle within the first predetermined time range according to the received positioning characteristic data and the alarm characteristic data; the braking deceleration characteristic data includes: braking deceleration characteristic data and braking distance characteristic data; the alarm characteristic score data includes: characteristic score data of each alarm type in the predetermined alarm type; Input the braking and deceleration feature data and alarm feature score data of the vehicle within the first predetermined time range into a pre-trained accident detection machine learning model, and obtain the results output by the accident detection machine learning model; the results include: normal driving and accidents; wherein the accident detection machine learning model is trained in advance using historical data of the selected vehicle; the historical data includes: the positioning feature data, alarm feature data, braking and deceleration feature data and alarm feature score data of the vehicle within the first predetermined time range before the time point of the accident, and the positioning feature data, alarm feature data, braking and deceleration feature data and alarm feature score data of the vehicle within the first predetermined time range when the vehicle is driving normally without an accident; Wherein, determining the braking deceleration characteristic data of the vehicle within the first predetermined time range includes: According to the driving speeds v1 and v2 corresponding to the two selected driving time points t1 and t2, the braking deceleration dt between t1 and t2 is determined, where: Among them, t2>t1; when dt is a positive number, it indicates that the vehicle is decelerating; when dt is a negative number, it indicates that the vehicle is accelerating; when dt is zero, it indicates that the speed of the vehicle has not changed or the vehicle is stopped; According to the driving mileages m1 and m2 corresponding to the two selected driving time points t1 and t2, the braking distance Δm between t1 and t2 is determined, wherein Δm=m2-m1; when Δm=0, it indicates that the vehicle has stopped driving; Wherein, determining the alarm feature score data of the vehicle within the first predetermined time range includes: Acquire the alarm time and the number of alarms of the predetermined alarm type within a second predetermined time range from the driving start time selected within the first predetermined time range, and determine the average value of the alarms within the second predetermined time range. in, The selected driving start time is the earliest time when the alarm occurs within the second predetermined time range; The normalized index of each alarm type in the predetermined alarm type As the alarm correction coefficient of the alarm type within the second predetermined time range, wherein, Among them, j corresponds to alarm type j, b j is the number of alarms of alarm type j occurring within the second predetermined time range, and k is the number of alarm types; For each alarm type j, the alarm correction coefficient corresponding to the alarm type is used. Determine the alarm feature score c of the alarm type within the second predetermined time range j ,in:

2. The method according to claim 1, characterized in that: The first predetermined time range is 1 to 5 minutes; and / or the second predetermined time range is a predetermined N days; and N is greater than 2.

3. The method according to claim 1, characterized in that The positioning feature data and the alarm feature data are received from the vehicle-mounted safety terminal; the vehicle-mounted safety terminal collects the vehicle's positioning feature data and the alarm feature data in real time.

4. The method according to claim 1, characterized in that: The accident discovery machine learning model samples the random forest algorithm to perform machine learning; and the historical data for training the accident discovery machine learning model is pre-labeled.

5. The method according to claim 1, characterized in that Before determining the braking and deceleration characteristic data and the alarm characteristic score data of the vehicle within the first predetermined time range according to the received positioning characteristic data and the alarm characteristic data, the method further includes: cleaning and normalizing the received positioning characteristic data and the alarm characteristic data.

6. The method according to claim 1, characterized in that Also includes: When the result is an accident, an accident occurrence report for reporting is generated based on the positioning feature data, alarm feature data, braking and deceleration feature data and alarm feature score data of the vehicle involved in the accident within the first predetermined time range.

7. The method according to claim 6, characterized in that Also includes: The vehicle accident emergency rescue system takes rescue measures for the vehicle involved in the accident according to the accident occurrence report.

8. A device for detecting a vehicle traffic accident, characterized in that: The method comprises a memory and a processor, wherein the memory stores at least one program, and the at least one program is executed by the processor to implement the steps of any one of the methods according to claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The storage medium stores at least one program, and the at least one program is executed by a processor to implement the steps of any one of the methods described in claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images