Vehicle collision accident detection method, device, system and computer readable medium

Abstract

The present invention discloses a method and device for detecting vehicle collision accidents. The method includes: obtaining vehicle driving data through a processor, and dividing the vehicle driving journey into multiple sub-trips according to the time axis; calculating the suspected collision time point in the sub-trip based on the change of acceleration in the lateral, longitudinal and vertical directions, and setting the sub-trip where the suspected collision time point is located as the suspected collision journey; establishing a map of the characteristic data of the driver's driving behavior and the vehicle's driving state in the suspected collision journey; based on the recognition algorithm model established in advance by performing map analysis and machine learning on the characteristic data of historical collision accidents, calculating the map of the characteristic data of the driver's driving behavior and the vehicle's driving state in the suspected collision journey, so as to identify the target collision time point from the suspicious collision time point, and output information about the target collision time point. This implementation method can improve the accuracy of detecting the time point of a vehicle collision accident.

Description

Technical Field

[0001] The present invention relates to the field of vehicle information technology and traffic accident identification, and in particular to a control method, device, system and computer-readable medium for vehicle collision accidents. Background Art

[0002] With the development of the national new energy vehicle industry, especially the advancement of intelligent driving assistance technology to L4 and L5 advanced development, the application of new energy vehicles and intelligent driving technology has given rise to a new liability factor in the field of traffic accident identification: new energy vehicle intelligent assisted driving safety liability. That is, the collision accident was caused by the intelligent assisted driving program processing control errors during the intelligent assisted driving period. The important basis for this type of traffic accident identification is the accurate acquisition or calculation of the time point of the collision accident, for example, it needs to be refined to the second level or even the millisecond level.

[0003] In addition, with the substantial growth in the number of new energy vehicles, their proportion of auto insurance premiums has become increasingly larger, and the accident rate of new energy vehicles is higher than that of traditional fuel vehicles, and insurance companies generally suffer losses in the operation of new energy vehicle insurance. How to accurately determine the time of vehicle collision and the driving status data before and after the collision, which provides important scientific basis for insurance companies to identify the authenticity of the case and the liability of insurance accidents, so that they can obtain investigation reports without manual on-site verification, has become a concern for insurance companies to reduce costs.

[0004] It can be seen that in the process of detecting vehicle collision accidents, how to accurately detect the time point of the vehicle collision accident and the driving status data before and after the accident, so as to accurately determine the responsibility for the traffic accident is an issue that needs to be considered urgently. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a method and device for detecting a vehicle collision accident, which can improve the accuracy of detecting the time point of the vehicle collision accident.

[0006] To achieve the above object, according to a first aspect of the present invention, a method for detecting a vehicle collision accident is provided, comprising:

[0007] The vehicle's driving data is obtained through the processor, and the vehicle's driving journey is divided into multiple sub-trips according to the time axis.

[0008] Calculating a suspected collision time point in the sub-trip based on changes in the acceleration of the vehicle in the lateral, longitudinal and vertical directions, and setting the sub-trip where the suspected collision time point is located as a suspected collision trip;

[0009] Establishing a graph of the characteristic data of the driver's driving behavior and the characteristic data of the vehicle's driving state during the suspected collision trip;

[0010] Based on a recognition algorithm model established in advance by performing graph analysis and machine learning on characteristic data of historical collision accidents, a graph of characteristic data of the driver's driving behavior and characteristic data of the vehicle's driving state of the suspected collision stroke is calculated to identify a target collision time point that meets the characteristics of the driving behavior and driving state in the vehicle collision accident from the suspected collision time point, and output information about the target collision time point.

[0011] Preferably, the recognition algorithm model is established through the following steps:

[0012] Based on the data of historical collision accidents, the characteristic data of the driver's driving behavior and vehicle driving status near the collision time point of the historical collision accidents are analyzed to obtain the causal factors of the collision accidents;

[0013] Based on the obtained collision accident causal factors, the characteristics of the collision time points of historical collision accidents are marked with a graph; and

[0014] The marked historical collision accident data is used for model training using machine learning to establish the recognition algorithm model.

[0015] Preferably, before setting the sub-trip where the suspected collision time point is located as the suspected collision trip, the vehicle collision accident detection method further comprises: filtering the sub-trip, wherein:

[0016] In the sub-trip, determining whether the suspected collision time point exists within a predetermined time before the time point when the gear position is changed to the N gear or the P gear;

[0017] If the suspected collision time point does not exist within the predetermined time, the subtrip is deleted.

[0018] Preferably, before setting the sub-trip where the suspected collision time point is located as the suspected collision trip, the vehicle collision accident detection method further comprises: filtering the calculated suspected collision time point, wherein:

[0019] Based on the data of driving behavior and vehicle status near the suspected collision time point, determining whether the change in acceleration of the vehicle in the lateral, longitudinal and vertical directions at the suspected collision time point is an abnormal mutation of acceleration;

[0020] In the case where the change in acceleration is an abnormal mutation, the suspected collision time point is marked on the map as a false suspected collision time point for identification based on the identification algorithm model.

[0021] Preferably, the characteristic data of the driver's driving behavior that is mapped is the driver's driving behavior data that is different from the driving behavior data when the vehicle is driving normally.

[0022] Preferably, the mapped characteristic data of the driver's driving behavior is based on at least one of the following data: a change in the effective stroke of the brake pedal, a change in the steering wheel angle, a rotation speed of the steering wheel angle, and a change in the accelerator pedal opening.

[0023] Preferably, the characteristic data of the vehicle driving state that is mapped is vehicle driving state data that is different from the driving state data of the vehicle when it is driving normally.

[0024] Preferably, the mapped characteristic data of the vehicle driving state is based on the change of acceleration of the vehicle in at least one of the lateral, longitudinal and vertical directions.

[0025] Preferably, dividing the vehicle's travel journey into a plurality of sub-trips according to the time axis comprises:

[0026] The driving trip of the vehicle is divided into the plurality of sub-trips according to at least one of the gear information and the door opening condition in the driving data.

[0027] Preferably, dividing the vehicle's travel journey into a plurality of sub-trips according to the time axis comprises:

[0028] determining whether there is a sub-trip in which the vehicle speed is always 0 among the plurality of sub-trips;

[0029] If there is a sub-trip in which the vehicle speed is always 0, the sub-trip is deleted.

[0030] According to the detection method of vehicle collision accident of the first aspect of the present invention, the collision time point and related information of the vehicle collision accident can be accurately and effectively detected, which greatly improves the accuracy of the collision time point detection, so that the responsibility of the road traffic accident can be accurately identified, and the cost expenditure of insurance accident claims can be reduced through the electronic investigation of the accident site. In addition, the present invention can facilitate machine learning and enable the computer to recognize the same kind of maps by mapping the collision data (historical collision data and suspected collision travel data). Compared with the recognition of simple digital data, the recognition of the map has better tolerance and redundancy, so that the collision time point can be more effectively identified. Moreover, the present invention can realize the automatic detection of collision accidents. When it is sensed that an extremely serious collision accident occurs (the high-voltage battery and electric drive of new energy vehicles have the characteristics of fast acceleration, high voltage, easy to catch fire, etc., so the degree of injury is large when a collision occurs), the back-end service personnel can make remote calls and tracking, and assist in on-site processing when the driver is incapable of acting, such as calling medical ambulances and traffic police at the first time, so as to provide timely and effective accident processing.

[0031] A second aspect of the present invention provides a vehicle collision accident detection device, comprising:

[0032] A trip division unit, which obtains the driving data of the vehicle through a processor, and divides the driving trip of the vehicle into a plurality of sub-trips according to a time axis;

[0033] a suspicious collision time point calculation unit, which calculates the suspicious collision time point in the sub-trip based on the changes in the acceleration of the vehicle in the lateral, longitudinal and vertical directions, and sets the sub-trip where the suspicious collision time point is located as the suspicious collision trip;

[0034] A map establishing unit is configured to establish a map based on the characteristic data of the driver's driving behavior and the characteristic data of the vehicle's driving state during the suspected collision trip; and

[0035] The target collision time point recognition unit calculates the graph of the characteristic data of the driver's driving behavior and the characteristic data of the vehicle's driving state of the suspected collision stroke based on a recognition algorithm model that is previously established by performing graph analysis on characteristic data of historical collision accidents and performing machine learning, so as to recognize a target collision time point that meets the characteristics of the driving behavior and driving state in the vehicle collision accident from the suspected collision time points, and outputs information about the target collision time point.

[0036] The vehicle collision accident detection device according to the second aspect of the present invention can achieve various technical effects similar to those of the first aspect.

[0037] The third aspect of the present invention provides an electronic device, comprising: one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in the first aspect.

[0038] A fourth aspect of the present invention provides a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in the first aspect.

[0039] The second to fourth aspects of the present invention can achieve the same technical effects as the first aspect.

[0040] The further effects of the above-mentioned non-conventional optional manner will be described below in conjunction with the specific implementation manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention.

[0042] Figure 1 is a schematic diagram of the main process of a method for detecting a vehicle collision accident according to an embodiment of the present invention;

[0043] Figure 2 is a schematic diagram showing an example of trip division;

[0044] Figure 3A and Figure 3B is a schematic diagram showing the marking of a spectrum of false suspicious collision time points;

[0045] Figures 4A to 4G is a schematic diagram showing the mapping of driver behavior and vehicle travel data characteristics;

[0046] Figure 5 is a schematic diagram of the steps of establishing a recognition algorithm model according to an embodiment of the present invention;

[0047] Figure 6 is a schematic diagram showing the form of a vehicle collision accident;

[0048] Figures 7A to 7F is a schematic diagram showing an example of marking collision time point features for a known collision accident case;

[0049] Figure 8 is a schematic diagram showing an example of the operation flow of the collision accident detection method of the present invention;

[0050] Fig. 9 is a schematic diagram of the main configuration of a vehicle collision accident detection device according to an embodiment of the present invention;

[0051] Fig.10is an exemplary system architecture diagram to which embodiments of the present invention may be applied;

[0052] Fig.11 It is a schematic diagram of the structure of a computer system of a terminal device or a server suitable for implementing an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0054] The vehicle involved in the vehicle collision accident detection method in the present application is particularly suitable for four-wheel passenger cars, but is not limited thereto.

[0055] Figure 1 FIG. 1 is a schematic diagram of the main process of the method for detecting a vehicle collision accident according to an embodiment of the present invention. Figure 1 As shown, the vehicle collision accident detection method according to the embodiment of the present invention mainly includes steps S101 to S104.

[0056] In step S101, the vehicle's driving data is acquired by a processor, and the vehicle's driving journey is divided into a plurality of sub-journals according to a time axis. For example, the vehicle's driving journey may be divided into a plurality of sub-journals according to at least one of gear information and door opening conditions in the driving data.

[0057] Specifically, for example, according to the gear changes during vehicle driving, the opening of the driver's door, and the intelligent assisted driving status, the data stream of the vehicle traveling along the timeline is cut into sub-trips.

[0058] For example, the vehicle starts driving in D gear or R gear, then stops, starts again, drives again, stops again, etc. Starting-driving-stopping-starting again-driving again-stopping again constitutes multiple small trips in the vehicle's driving journey. Combined with the opening state of the driver's door, the small trips are summarized into 26 types, and the 26 trip gauges are used to convert them during the trip division process, as shown in Table 1, where D, N, P, and R represent the vehicle's D gear, N gear, P gear, and R gear, respectively.

[0059] Table 1: Travel gauge examples

[0060]

[0061] Figure 2An example of trip cutting (division) is shown. Figure 2 As shown, in subtrip ①[RR], the next second after the last R is D, RD is not within the gauge, and the last R is taken as the end point. In subtrip ②[DD], the next second after the last D is R, DR is not within the gauge, and the last D is taken as the end point. In subtrip ③[RP], the next second after the last R is P, RP is within the gauge, and P is combined; the next second after the last P is R, RPR is not within the gauge, and the last P is taken as the end point; R is the starting point of the new trip. In subtrip ④[RNPKP], the next second after the last R is N, RN is within the gauge, and N is combined; the next second after the last N is P, RNP is within the gauge, and P is combined. There is a door open between the first P and the last P, so the trip is [RNPKP].

[0062] In addition, preferably, the subtrips can also be filtered. That is, it is determined whether there is a subtrip in which the vehicle speed is always 0 among the multiple subtrips, and if there is a subtrip in which the vehicle speed is always 0, the subtrip is deleted. Thus, the subtrip with no vehicle speed, that is, the subtrip with no collision time point can be deleted, and the recognition accuracy and efficiency of the machine learning recognition model described later can be improved.

[0063] return Figure 1 ,like Figure 1 As shown, in step S102, based on the changes in the acceleration of the vehicle in the lateral, longitudinal and vertical directions, the suspicious collision time point in the sub-trip is calculated and the sub-trip where the suspicious collision time point is located is set as the suspicious collision trip. When a vehicle collides, it can reflect the occurrence of the collision accident based on the changes in the acceleration of the vehicle in the three directions of the X-axis (lateral), Y-axis (longitudinal) and Z-axis (vertical). For example, a three-axis sensor set on the vehicle can be used to sense the changes in the acceleration of the vehicle in the longitudinal (Y-axis), lateral (X-axis) and vertical (Z-axis) directions respectively to perceive the stability of the vehicle's driving. When a collision occurs, there will be a significant change in the acceleration in at least one direction of the X, Y and Z axes. For example, a predetermined threshold value can be set for the change in the comprehensive acceleration of the XYZ axes through a large amount of known collision accident data, so as to screen out suspicious collision time points based on the threshold value.

[0064] For example, according to the international standard ISO / DIS6597, the minimum braking deceleration of a cold test of a car is not less than 5.8m / s2, which is equivalent to 0.59G. According to a large number of known collision cases, at a certain point in time, when the absolute value of the acceleration variable in a certain direction of the X, Y, or Z axis is greater than or equal to 0.1G, it is considered a sudden change unstable state. The three-axis comprehensive acceleration is When reversing in R gear, the vehicle speed is usually low, so △a(xyz) ≥ 0.13G, it can be determined as a suspicious collision time point; when driving forward in D gear, △a(xyz) ≥ 0.15G, it can be determined as a suspicious collision time point. The following (1)-(9) show examples of the acceleration at the suspicious collision time point and its corresponding vehicle driving state:

[0065] (1) At the time of the suspected collision, △ax ≥ 0.1G (acceleration change on the X axis), and the vehicle is in a "left-swinging and right-swinging" driving state;

[0066] (2) At the time of the suspected collision, △ax ≤ -0.1G, and the vehicle was in a "right-swinging and left-swinging" driving state;

[0067] (3) At the time of the suspected collision, -0.1G < △ax < 0.1G, and the vehicle does not swing significantly left or right;

[0068] (4) At the time of the suspected collision, △ay ≥ 0.1G (acceleration change on the Y axis), and the vehicle is in a state of "rapid forward acceleration";

[0069] (5) At the time of the suspected collision, △ay≤-0.1G, and the vehicle was in a state of "rapid deceleration";

[0070] (6) At the time of the suspected collision, -0.1G < △ay < 0.1G, and the vehicle was not in an obvious rapid acceleration or deceleration state;

[0071] (7) At the time of the suspected collision, △az ≥ 0.1G (acceleration change on the Z axis), and the vehicle is in a "tilted forward and backward" driving state; (i)

[0072] (8) At the time of the suspected collision, △az ≤ -0.1G, and the vehicle was in a "nodding head and lifting tail" driving state;

[0073] (9) At the time of the suspected collision, -0.1G < △az < 0.1G, and there was no obvious up and down movement of the front and rear ends of the vehicle.

[0074] In addition, preferably, the sub-trips can be screened and filtered. After a vehicle collision occurs, the vehicle speed will drop to 0, and the driver will often put the gear back to N or P, and may open the driver's door to get out of the car to check (in the case of an extremely serious collision and the driver loses the ability to act, it can be directly judged by the XYZ comprehensive acceleration change). Therefore, in the sub-trip, it can be judged whether there is a suspicious collision time point within the predetermined time before the gear is changed to N or P; if there is no suspicious collision time point within the predetermined time, the sub-trip is deleted. Therefore, by filtering the sub-trip, the recognition accuracy and efficiency can be further improved.

[0075] For example, in a sub-trip, the driver shifts the gear back to N or P, and the first time is taken as the starting point, and 60 seconds are counted back as the collision trip; if there is no suspicious collision time point in the 60-second collision trip, the sub-trip is filtered out.

[0076] In addition, preferably, the calculated suspicious collision time points can also be filtered. There may be multiple suspicious collision time points in the collision journey, but in actual driving scenarios such as the vehicle passing over speed bumps, rapid acceleration, and rapid deceleration, abnormal mutations in the comprehensive acceleration changes of the XYZ axes may also occur. Therefore, the suspicious collision time points that have mutations in the above scenarios can be filtered out to avoid interference, thereby further improving recognition accuracy and efficiency.

[0077] For example, based on the calculated data of driving behavior and vehicle status near the suspected collision time point, it can be determined whether the change in the acceleration of the vehicle in the lateral, longitudinal and vertical directions at the suspected collision time point is an abnormal mutation in acceleration; if the change in acceleration is an abnormal mutation, the suspected collision time point is marked on the map as a false suspected collision time point for identification based on the recognition algorithm model.

[0078] For example, through manual methods, the time points of false suspicious collisions are marked on a map, and then identified and judged using a machine learning algorithm model. Figure 3A and Figure 3B An example of marking a false suspected collision time point in the spectrum is shown. Figure 3A As shown in , at the suspected collision time point in the graph, the vehicle speed is 0 at the time points 1 second before and 1 second after it, so the suspected collision time point is marked as a "false suspected collision time point". Figure 3B As shown, at the suspicious collision time point in the graph, there is no feature of sudden braking driving behavior within 5 seconds before and after it, so the suspicious collision time point is marked as a "false suspicious collision time point".

[0079] Return Figure 1 ,like Figure 1 As shown, in step S103, a map is created for the characteristic data of the driver's driving behavior and the characteristic data of the vehicle's driving state in the suspected collision trip. Before and after the vehicle collision accident, the driver's behavior and the vehicle's driving state must be significantly different from those during normal driving. Therefore, the data characteristics that may be different can be mapped to facilitate computer learning and recognition. That is, the characteristic data of the driver's driving behavior and the characteristic data of the vehicle's driving state that are mapped can be data that are different from the driving behavior data and the driving state data when the vehicle is driving normally.

[0080] Preferably, the mapped characteristic data of the driver's driving behavior is based on at least one of the following data: a change in the effective travel of the brake pedal, a change in the steering wheel angle, a rotation speed of the steering wheel angle, and a change in the accelerator pedal opening. The mapped characteristic data of the vehicle's driving state is based on a change in the acceleration of the vehicle in at least one of the lateral, longitudinal and vertical directions, for example.

[0081] refer to Figures 4A-4G , examples 1-6 of using graphs to portray driver behavior and vehicle driving data characteristics are described.

[0082] In Example 1, a graph showing the behavior characteristics of a driver stepping on the brakes is shown. Figure 4A As shown in the figure, ① represents "sudden braking", that is, the effective travel of the brake pedal, the variable changes suddenly between two time points, and according to the actual variable size, it can be divided into three levels: super-acute braking, emergency braking, and preventive emergency braking; ② represents "slow braking", that is, the effective travel of the brake pedal gradually increases between two consecutive time points, but the variables at the two time points do not mutate, and this graph is slow braking; ③ represents "brake hold", that is, the effective travel of the brake pedal is in a deep braking state value at at least two consecutive time points. But the variables at the two time points do not mutate.

[0083] In Example 2, a graph showing the behavioral characteristics of a driver turning the steering wheel sharply is shown. Figure 4B As shown in FIG. 1 , ① ② ③ ④ are all characteristic graphs of the driver's sudden steering wheel turning behavior. The characteristic of the sudden steering wheel turning behavior is that between two time points, the change in the steering wheel angle is greater than a predetermined angle (e.g., 50 degrees), and the change in the steering wheel speed is greater than a predetermined value (e.g., 100 degrees / second).

[0084] In Example 3, a graph showing the behavior characteristics of a driver turning left and turning right is shown. Figure 4C As shown, ① is a graph representing the driving characteristics when the driver turns the vehicle to the left, wherein the steering wheel angle gradually increases from 0 degree, reaches a peak value and then gradually returns to 0 degree; ② is a graph representing the driving characteristics when the driver turns the vehicle to the right, wherein the steering wheel angle gradually becomes negative from 0 degree, reaches a negative peak and then gradually returns to 0 degree.

[0085] In Example 4, a graph showing the behavior characteristics of a driver changing lanes to the right is shown. Figure 4D As shown, ① represents the changing characteristics of the steering wheel angle when the driver changes lanes to the right, wherein the steering wheel changes from 0 degrees to a negative value, reaches a negative peak, then changes toward 0 degrees, crosses 0 degrees to a positive value, reaches a positive peak, and then returns to 0 degrees.

[0086] In Example 5, a graph showing the behavior characteristics of slow acceleration and rapid acceleration of the driver is shown. Figure 4E As shown, ① represents the behavioral data map characteristics of the driver's slow acceleration, wherein, between two consecutive time points, the accelerator pedal opening does not change suddenly, but increases slowly, and the vehicle speed does not change suddenly, but increases slowly; ② represents the behavioral data map characteristics of the driver's sudden stepping on the accelerator pedal to accelerate, wherein between two time points, the accelerator pedal opening changes suddenly, and the vehicle speed also increases suddenly.

[0087] In Example 6, a characteristic map of the vehicle driving state is shown. Figure 4F As shown, ① indicates that the acceleration of the vehicle's X-axis and Z-axis suddenly changes, the three-axis comprehensive acceleration of △a(xyz) suddenly changes, the Z-axis acceleration suddenly changes from 0 to a positive value, indicating that the vehicle's driving posture has a tendency to "pitch forward and flip backward", and the X-axis acceleration suddenly changes from 0 to a negative value, indicating that the vehicle's driving posture has a tendency to "turn right and swing left"; combined with Figure 4G The characteristics of vehicle speed and accelerator pedal opening are shown, and the vehicle driving state characteristics are related to the driver's sudden acceleration by pressing the accelerator pedal. It indicates that ② the vehicle's X-axis and Z-axis accelerations suddenly change, and the △a (xyz) three-axis comprehensive acceleration suddenly changes. The Z-axis acceleration suddenly changes from 0 to a negative value, indicating that the vehicle's driving posture has a "nodding head and tilting tail" trend, and the X-axis acceleration suddenly changes from 0 to a positive value, indicating that the vehicle's driving posture has a "left turn head and right swing tail" trend; combined with Figure 4G The vehicle speed and brake pedal effective travel characteristics shown are related to the vehicle driving state characteristics and the driver's sudden braking by pressing the brake pedal.

[0088] Refer to Figure 1 In step S104, based on the recognition algorithm model established in advance by performing graph analysis and machine learning on the characteristic data of historical collision accidents, the graph of the characteristic data of the driver's driving behavior and the characteristic data of the vehicle's driving state of the suspected collision stroke is calculated to identify the target collision time point that meets the characteristics of the driving behavior and driving state in the vehicle collision accident from the suspicious collision time point and output information about the target collision time point.

[0089] Preferably, if Figure 5 As shown, the recognition algorithm model can be established through the following steps: step S201, based on the data of historical collision accidents, the characteristic data of the driver's driving behavior and the vehicle's driving status near the collision time point of the historical collision accident are analyzed to obtain the collision accident causal factors; step S202, according to the obtained collision accident causal factors, the characteristics of the collision time point of the historical collision accident are marked with a map; and step S203, the marked historical collision accident data are used for model training using machine learning, so as to establish the recognition algorithm model.

[0090] Specifically, in step S201 of analyzing the causal factors of the collision accident, the data of a large number of known collision accidents can be analyzed, and the characteristics of the driver's behavior and the actual driving state data of the vehicle at each time point within a predetermined time before and after the collision (for example, 5 seconds before and 5 seconds after the collision) can be analyzed to find the causal factors of the collision accident and the weight of the factors affecting the collision accident. For example, based on at least one of the form of the collision accident, the different gears during driving (for example, the gears R and D are different), and the different collision parties (that is, the active collision and the collision by the third party vehicle are different), the data of the known collision accidents can be analyzed (for example, analyzing the characteristics of the driver's behavior and the vehicle driving state data 5 seconds before and after the collision time point and the probability of the occurrence of the characteristics), so as to obtain the causal factors of the collision accident, and provide a methodology for the subsequent collision time point feature labeling and machine learning. Figure 6 An example of the form of a vehicle collision accident is shown, where Ⅰ to Ⅳ belong to the frontal collision form; Ⅴ to Ⅷ belong to the side collision form; Ⅸ to Ⅻ belong to the rear-end collision form. In addition, the corresponding weights of the impact on the collision accident can be set for the obtained collision accident causal factors. For example, extra-acute braking has a great impact on the accident, so the corresponding collision accident causal factors have a high weight value. General emergency braking may be a preventive point brake, so the corresponding collision accident causal factors have a lower weight value. By performing machine learning based on the collision accident causal factors and the weights of the factors, it is possible to more accurately identify collision accidents and their collision time points.

[0091] The factors contributing to the collision accident are as follows:

[0092] [When you step on the accelerator to accelerate, the vehicle collides and slows down];

[0093] [5 seconds before the vehicle collision, if the speed is greater than 80km / h, brake very quickly and hard];

[0094] [The vehicle accelerates forward while the brakes are applied suddenly, causing the vehicle to be rear-ended or hit from the front];

[0095] [Extremely urgent braking 3 seconds before vehicle collision];

[0096] [Extremely urgent braking 2 seconds before vehicle collision];

[0097] [Extremely urgent braking 1 second before vehicle collision];

[0098] [Extremely urgent braking when vehicles collide];

[0099] [Extremely urgent braking 1 second after vehicle collision];

[0100] [Extremely urgent braking 2 seconds after vehicle collision];

[0101] [Extremely urgent braking 3 seconds after vehicle collision];

[0102] [When the vehicle collides in R gear, brake suddenly and turn the steering wheel sharply at a large angle at the same time].

[0103] In step S202 of marking the features of the collision time point, the dominant features of the collision time point in the known collision accident within a predetermined time before and after the collision time point are marked with a map using the collision accident causation factor. For example, when a collision occurs when the vehicle is moving forward or backward, there are dominant features at the collision time point and within 5 seconds before and after the collision time point according to the analysis of the collision accident causation factor method, and the dominant features of the collision time point of the known collision accident case are marked by manual methods, and then the target collision time point is automatically identified through the machine learning training model.

[0104] 7A to 7F An example of collision time point feature marking for a known collision accident case is shown. Fig. 7A and 7B The figure shows the driving state characteristics of a vehicle being rear-ended by a third-party vehicle and an example of its characteristic marking. Fig. 7A As shown in the figure, car A is a new energy vehicle and is rear-ended by a third-party car B. Figure 7B As shown, the features of this rear-end collision case are marked as follows:

[0105] [One second before the vehicle collision, sudden braking and deceleration (deceleration greater than 0.5G)];

[0106] [At 1 second after the collision, the vehicle accelerates forward while braking at the depth of the effective brake travel, resulting in a rear-end collision or a forward collision];

[0107] [Level 1 warning for the left blind spot of the vehicle within 5 seconds before a vehicle collision];

[0108] [The vehicle stops for more than 2 minutes after a collision];

[0109] [The driver opens the door and gets out to check within 2 minutes after the vehicle collision].

[0110] Figure 7C and 7D The following is an example of the driving state characteristics and feature marking of a vehicle actively rear-ending a third-party vehicle. Figure 7C As shown in the figure, vehicle A represents a new energy vehicle, which actively rear-ends and collides with a third-party vehicle B. Fig.7D As shown, the features of this collision case are marked as follows:

[0111] [The collision occurred when the vehicle was changing lanes to the right];

[0112] [Vehicle collision time, acute braking];

[0113] [At the time of vehicle collision, sudden braking and deceleration (0.2-0.5G deceleration)];

[0114] [Level 1 warning for the left blind spot of the vehicle within 5 seconds before a vehicle collision];

[0115] [Level 1 warning for the right blind spot of the vehicle within 5 seconds before a vehicle collision];

[0116] [The vehicle stops for more than 2 minutes after a collision];

[0117] [The driver opens the door and gets out to check within 2 minutes after the vehicle collision].

[0118] Fig. 7E and 7F An example of an extremely severe collision is shown. When the test vehicle collides, the driver's airbag ignites and explodes to protect the occupants. The general conditions are: Fig. 7E As shown, when the vehicle is moving forward, the speed is ≥30km / h, and the direction of the collision force comes from the 150-degree fan-shaped area between the left and right headlight positions. Assuming that a collision occurs at a speed of 30km / h, and the speed drops to 0 within 1 second, the collision deceleration is 0.85G. According to the above, according to the international standard ISO / DIS6597, the minimum braking deceleration of the automobile cold test is not less than 5.8m / s2, which is equivalent to 0.59G. Therefore, the present invention sets the collision under the condition of three-axis comprehensive acceleration △a(xyz)≥2G as an extremely serious collision.

[0119] like Figure 7F As shown, ① 1 second before the collision, △a(xyz) = 3.25G; ② at the collision, △a(xyz) = 4.0G; ③ 1 second after the collision, △a(xyz) = 5.93G; ④ 2 seconds after the collision, △a(xyz) = 3.51G; ⑤ 3 seconds after the collision, △a(xyz) = 2.29G. Based on this, it can be judged that the vehicle has experienced multiple extremely serious collisions. At 2:52:13, the driver suddenly braked at a speed of 117.06 km / h; at 21:52:14, the driver's effective braking range reached its peak, and when the speed dropped to 89.94 km / h, the braking deceleration = 0.769G; at 21:52:16, the driver kept braking and a collision occurred. At the time of the collision, △a(xyz) = 4.0G, and the speed after the collision = 13.56 km / h. The characteristics of this collision case are marked as follows:

[0120] [3 seconds before the vehicle collision, when the vehicle speed is greater than 80km / h, turn the steering wheel sharply at a large angle];

[0121] [Two seconds before the vehicle collision, the vehicle brakes suddenly and decelerates (deceleration greater than 0.5G) when the speed is greater than 80km / h];

[0122] [Lose control of the steering wheel at the time of collision];

[0123] [At 1 second after the collision, the vehicle accelerates forward while braking at the depth of the effective brake travel, resulting in a rear-end collision or a forward collision];

[0124] [The vehicle is in an extremely severe collision, with a collision force greater than 2G];

[0125] [The vehicle stops for more than 2 minutes after a collision];

[0126] [The driver opens the door and gets out to check within 2 minutes after the vehicle collision].

[0127] After the known collision accidents are marked, the marked known collision accidents are trained with machine learning to establish a recognition model. Based on the recognition model, the real collision time point is identified from the suspicious collision time.

[0128] Figure 8 FIG. 2 shows an example of the operation flow of the collision accident detection method of the present invention. Figure 8 As shown, first, the data of the vehicle traveling along the time axis is cut, and the sub-trips are generated based on the gear position signal at the time point. It is judged whether the trip meets the discarding conditions in the trip cutting gauge. Generate the valid trip program number, name, starting time, end time, trip duration, etc., and perform the first screening of the valid trip according to the collision trip screening rules. It is judged whether there is a time point when the vehicle speed is greater than 0 in the sub-trip. If not, the sub-trip without vehicle speed is filtered out. Then, the △a(xyz) of each time point within the 60-second collision period of each valid sub-trip is calculated to determine whether there is a time point in the trip when △a(xyz) ≥ 0.15G in D gear or △a(xyz) ≥ 0.13G in R gear. If there is no such time point, the trip is filtered out. Thus, the results of the second trip screening and the results of the suspicious collision time points are calculated. Based on the algorithm model that marks known collision cases and performs machine learning and training on the marked cases, the suspicious collision time points are calculated to identify the real collision trip and collision time points.

[0129] The present invention is based on the application of assisted driving technology for new energy vehicles, and collects vehicle driving data (such as vehicle speed, effective travel of brake pedal, accelerator pedal opening, steering wheel angle and speed, three-axis acceleration, etc.) (for example, in real time), analyzes driver behavior characteristics and causes of collision accidents through, for example, big data mining, and then realizes automatic detection of collision accidents through machine learning algorithms. Specifically, the driver behavior and vehicle driving data characteristics are mapped, for example, driving behavior characteristics such as sudden braking, sudden acceleration, and sharp turns are identified through maps. For a large number of new energy vehicles that actually collide, at known collision time points, such as 10 seconds before and 180 seconds after, the causes of collision accidents and factor weights are mined and analyzed based on the driver behavior characteristics and vehicle driving state characteristics. Then, the machine learning method is used for training to automatically detect and identify the time points that meet the driving behavior characteristics and vehicle driving characteristics during the collision accident process.

[0130] According to the above method of the present invention, the collision time point and related information of the vehicle collision accident can be accurately and effectively detected, which greatly improves the accuracy of the collision time point detection, so that the responsibility of the road traffic accident can be accurately identified, and the cost expenditure of insurance accident claims can be reduced through the electronic investigation of the accident site. In addition, the present invention can facilitate machine learning and enable the computer to recognize the same kind of maps by mapping the collision data (historical collision data and suspicious collision travel data). Compared with the recognition of simple digital data, the recognition of the map has better tolerance and redundancy, so that the collision time point can be more effectively identified. Moreover, the present invention can realize the automatic detection of collision accidents. When it is sensed that an extremely serious collision accident occurs (the high-voltage battery and electric drive of new energy vehicles have the characteristics of fast acceleration, high voltage, easy to catch fire, etc., so the degree of injury is large when a collision occurs), the back-end service personnel can make remote calls and tracking, and assist in on-site processing when the driver is incapable of acting, such as calling a medical ambulance and traffic police at the first time, thereby providing timely and effective accident processing.

[0131] Note that the present invention is not limited to the above-described embodiments, but includes other configurations that can achieve the objects of the present invention.

[0132] The above embodiment describes dividing the vehicle travel by gear information and door opening conditions. However, the present invention is not limited thereto. The vehicle travel can also be divided based on other travel data such as time and vehicle speed. However, the division by gear and / or door opening conditions in the embodiment is a preferred solution, which can help to more accurately detect the vehicle collision travel and collision time point.

[0133] In addition, the above embodiment also describes the filtering process of the sub-trips and the suspected collision time points. However, the present invention is not limited thereto, and such filtering process may not be performed. However, the filtering process in the embodiment of the present invention is a preferred solution, which can help to detect the collision time point more accurately and efficiently.

[0134] In addition, the mapped characteristic data of the driver's driving behavior and the characteristic data of the vehicle's driving state are not limited to the data shown in the above examples, but can also be other characteristic data of driving behavior and characteristic data of the vehicle's driving state, as long as the purpose of the present invention can be achieved.

[0135] In addition, the above embodiment describes the manual marking of historical collision accident cases. However, the present invention is not limited thereto. The marking can also be performed by machine marking, a combination of machine marking and manual verification, or other methods, as long as the purpose of the present invention can be achieved.

[0136] The embodiment of the present invention also provides a vehicle collision accident detection device, specifically Fig. 9 The vehicle collision accident detection device 200 of the present embodiment includes: a trip division unit 201, which obtains the driving data of the vehicle through a processor and divides the driving trip of the vehicle into multiple sub-trips according to the time axis; a suspicious collision time point calculation unit 202, which calculates the suspicious collision time point in the sub-trip based on the change of the acceleration of the vehicle in the lateral, longitudinal and vertical directions, and sets the sub-trip where the suspicious collision time point is located as the suspicious collision trip; a spectrum establishment unit 203, which establishes a spectrum for the characteristic data of the driver's driving behavior and the characteristic data of the vehicle's driving state in the suspicious collision trip; and a target collision time point identification unit 204, which calculates the spectrum of the characteristic data of the driver's driving behavior and the characteristic data of the vehicle's driving state in the suspicious collision trip based on the identification algorithm model established by performing spectrum analysis on the characteristic data of historical collision accidents and machine learning in advance, so as to identify the target collision time point that meets the characteristics of the driving behavior and driving state in the vehicle collision accident from the suspicious collision time point, and output information about the target collision time point.

[0137] The vehicle collision accident detection device 200 of the embodiment of the present invention can achieve the same technical effect as the vehicle collision accident detection method of the present invention. Although not described here, the vehicle collision accident detection device 200 can also perform various processes of the vehicle collision accident detection method of the present invention.

[0138] Fig.10 An exemplary system architecture 300 is shown to which a method for detecting a vehicle collision accident or an apparatus for detecting a vehicle collision accident according to an embodiment of the present invention may be applied.

[0139] like Fig.10 As shown, the system architecture 300 may include terminal devices 301, 302, 303, a network 304 and a server 305. The network 304 is used to provide a medium for communication links between the terminal devices 301, 302, 303 and the server 305. The network 304 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0140] Users can use terminal devices 301, 302, 303 to interact with server 305 through network 304 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 301, 302, 303, such as vehicle damage assessment applications, web browser applications, search applications, instant messaging tools, etc. (only as examples).

[0141] The terminal devices 301 , 302 , and 303 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers, etc.

[0142] The server 305 may be a server that provides various services, such as a background management server (only as an example) that provides data support for the vehicle frame number entered by the user using the terminal devices 301, 302, and 303. The background management server may analyze and process the received data such as the product information query request, and feed back the processing result (such as the time point of the vehicle collision accident - only as an example) to the terminal device.

[0143] It should be noted that the vehicle collision accident detection method provided in the embodiment of the present invention is generally executed by the server 305 , and accordingly, the vehicle collision accident detection device is generally disposed in the server 305 .

[0144] It should be understood that Fig.10 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0145] Reference below Fig.11 , which shows a schematic diagram of the structure of a computer system 400 of a terminal device suitable for implementing an embodiment of the present invention. Fig.11 The terminal 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 invention.

[0146] like Fig.11As shown, the computer system 400 includes a central processing unit (CPU) 401, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 402 or a program loaded from a storage part 408 into a random access memory (RAM) 403. In the RAM 403, various programs and data required for the operation of the system 400 are also stored. The CPU 401, the ROM 402, and the RAM 403 are connected to each other via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0147] 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 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.

[0148] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 409, and / or installed from the removable medium 411. When the computer program is executed by the central processing unit (CPU) 401, the above-mentioned functions defined in the system of the present invention are executed.

[0149] It should be noted that the computer-readable medium shown in the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0150] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and a part of the above-mentioned module, program segment, or code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of boxes in the block diagram or flow chart, can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0151] The units involved in the embodiments of the present invention may be implemented by software or hardware. The units described may also be set in a processor, for example, it may be described as: a processor including a trip division unit, a map establishment and judgment unit, a calculation unit, and an identification unit. The names of these units do not constitute a limitation on the units themselves in some cases. For example, the trip division unit may also be described as "a unit that obtains the driving data of a vehicle and divides the driving trip of the vehicle into multiple sub-trips according to the time axis".

[0152] As another aspect, the present invention also provides a computer-readable medium, which may be included in the device described in the above embodiment; or may exist independently and not assembled into the device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by a device, the device includes: obtaining the driving data of the vehicle through a processor, and dividing the driving trip of the vehicle into multiple sub-trips according to the time axis; calculating the suspicious collision time point in the sub-trip based on the change of the acceleration of the vehicle in the lateral, longitudinal and vertical directions, and setting the sub-trip where the suspicious collision time point is located as the suspicious collision trip; establishing a map of the characteristic data of the driver's driving behavior and the characteristic data of the vehicle's driving state in the suspicious collision trip; based on the recognition algorithm model established by performing map analysis on the characteristic data of historical collision accidents and machine learning in advance, calculating the map of the characteristic data of the driver's driving behavior and the characteristic data of the vehicle's driving state in the suspicious collision trip, so as to identify the target collision time point that meets the characteristics of the driving behavior and driving state in the vehicle collision accident from the suspicious collision time point, and output information about the target collision time point.

[0153] The present invention raises the level of automatic detection of automobile collision accidents to a new level, and provides a new method for intelligent automatic detection of collision accidents for new energy vehicles. The present invention uses, for example, AI artificial intelligence, industry big data standardization and big data mining and analysis methods to realize automatic detection technology of new energy vehicle collision accidents, which has great social value.

[0154] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for detecting a vehicle collision accident, comprising: Acquire the driving data of the vehicle through the processor, and divide the driving trip of the vehicle into a plurality of sub-trips according to the time axis; Calculating a suspected collision time point in the sub-trip based on changes in the acceleration of the vehicle in the lateral, longitudinal and vertical directions, and setting the sub-trip where the suspected collision time point is located as a suspected collision trip; Establishing a graph of the characteristic data of the driver's driving behavior and the characteristic data of the vehicle's driving state during the suspected collision trip; Based on a recognition algorithm model established in advance by performing a graph analysis on the characteristic data of historical collision accidents and performing machine learning, a graph of the characteristic data of the driver's driving behavior and the characteristic data of the vehicle's driving state of the suspected collision trip is calculated to identify a target collision time point that meets the characteristics of the driving behavior and driving state in the vehicle collision accident from the suspected collision time point, and output information about the target collision time point. The driver's driving behavior characteristic data mapped is the driver's driving behavior data that is different from the driving behavior data when the vehicle is driving normally; the driver's driving behavior characteristic data mapped is based on at least one of the following data: a change in the effective travel of the brake pedal, a change in the steering wheel angle, a rotation speed of the steering wheel angle, and a change in the accelerator pedal opening, Among them, the characteristic data of the vehicle's driving state that is mapped is vehicle driving state data that is different from the driving state data of the vehicle when it is driving normally; the characteristic data of the vehicle's driving state that is mapped is based on the change of the acceleration of the vehicle in at least one of the lateral, longitudinal and vertical directions.

2. The method for detecting a vehicle collision accident according to claim 1, wherein: The recognition algorithm model is established through the following steps: Based on the data of historical collision accidents, the characteristic data of the driver's driving behavior and vehicle driving status near the collision time point of the historical collision accidents are analyzed to obtain the causal factors of the collision accidents; According to the obtained collision accident causal factors, the characteristics of the collision time points of the historical collision accidents are marked with a map; as well as The marked historical collision accident data is used for model training using machine learning to establish the recognition algorithm model.

3. The method for detecting a vehicle collision accident according to claim 1, Before setting the subtrip where the suspected collision time point is located as a suspected collision trip, the method further includes: The sub-trip is filtered, wherein: In the sub-trip, determining whether the suspected collision time point exists within a predetermined time before the time point when the gear position is changed to the N gear or the P gear; If the suspected collision time point does not exist within the predetermined time, the subtrip is deleted.

4. The method for detecting a vehicle collision accident according to claim 2, Before setting the subtrip where the suspected collision time point is located as a suspected collision trip, the method further includes: The sub-trip is filtered, wherein: In the sub-trip, determining whether the suspected collision time point exists within a predetermined time before the time point when the gear position is changed to the N gear or the P gear; If the suspected collision time point does not exist within the predetermined time, the subtrip is deleted.

5. The method for detecting a vehicle collision accident according to claim 1, Before setting the subtrip where the suspected collision time point is located as a suspected collision trip, the method further includes: The calculated suspected collision time point is filtered, wherein: Based on the data of driving behavior and vehicle status near the suspected collision time point, determining whether the change in acceleration of the vehicle in the lateral, longitudinal and vertical directions at the suspected collision time point is an abnormal mutation of acceleration; In the case where the change in acceleration is an abnormal mutation, the suspected collision time point is marked on the map as a false suspected collision time point for identification based on the identification algorithm model.

6. The method for detecting a vehicle collision accident according to claim 2, Before setting the subtrip where the suspected collision time point is located as a suspected collision trip, the method further includes: The calculated suspected collision time point is filtered, wherein: Based on the data of driving behavior and vehicle status near the suspected collision time point, determining whether the change in acceleration of the vehicle in the lateral, longitudinal and vertical directions at the suspected collision time point is an abnormal mutation of acceleration; In the case where the change in acceleration is an abnormal mutation, the suspected collision time point is marked on the map as a false suspected collision time point for identification based on the identification algorithm model.

7. The method for detecting a vehicle collision accident according to claim 3, Before setting the subtrip where the suspected collision time point is located as a suspected collision trip, the method further includes: The calculated suspected collision time point is filtered, wherein: Based on the data of driving behavior and vehicle status near the suspected collision time point, determining whether the change in acceleration of the vehicle in the lateral, longitudinal and vertical directions at the suspected collision time point is an abnormal mutation of acceleration; In the case where the change in acceleration is an abnormal mutation, the suspected collision time point is marked on the map as a false suspected collision time point for identification based on the identification algorithm model.

8. The method for detecting a vehicle collision accident according to claim 4, Before setting the subtrip where the suspected collision time point is located as a suspected collision trip, the method further includes: The calculated suspected collision time point is filtered, wherein: Based on the data of driving behavior and vehicle status near the suspected collision time point, determining whether the change in acceleration of the vehicle in the lateral, longitudinal and vertical directions at the suspected collision time point is an abnormal mutation of acceleration; In the case where the change in acceleration is an abnormal mutation, the suspected collision time point is marked on the map as a false suspected collision time point for identification based on the identification algorithm model.

9. The method for detecting a vehicle collision accident according to any one of claims 1 to 8, wherein: The step of dividing the vehicle's travel journey into a plurality of sub-trips according to the time axis includes: The driving trip of the vehicle is divided into the plurality of sub-trips according to at least one of the gear information and the door opening condition in the driving data.

10. The method for detecting a vehicle collision accident according to any one of claims 1 to 8, wherein: The step of dividing the vehicle's travel journey into a plurality of sub-trips according to the time axis includes: determining whether there is a sub-trip in which the vehicle speed is always 0 among the plurality of sub-trips; If there is a sub-trip in which the vehicle speed is always 0, the sub-trip is deleted.

11. The method for detecting a vehicle collision accident according to claim 9, wherein: The step of dividing the vehicle's travel journey into a plurality of sub-trips according to the time axis includes: determining whether there is a sub-trip in which the vehicle speed is always 0 among the plurality of sub-trips; If there is a sub-trip in which the vehicle speed is always 0, the sub-trip is deleted.

12. A vehicle collision accident detection device, comprising: A trip division unit, which obtains the driving data of the vehicle through a processor, and divides the driving trip of the vehicle into a plurality of sub-trips according to a time axis; a suspicious collision time point calculation unit, which calculates the suspicious collision time point in the sub-trip based on the changes in the acceleration of the vehicle in the lateral, longitudinal and vertical directions, and sets the sub-trip where the suspicious collision time point is located as the suspicious collision trip; A map establishing unit, which establishes a map based on the characteristic data of the driver's driving behavior and the characteristic data of the vehicle's driving state in the suspected collision trip; as well as The target collision time point recognition unit calculates the graph of the characteristic data of the driver's driving behavior and the characteristic data of the vehicle's driving state of the suspected collision stroke based on the recognition algorithm model established in advance by performing graph analysis on the characteristic data of historical collision accidents and performing machine learning, so as to recognize the target collision time point that meets the characteristics of the driving behavior and the driving state in the vehicle collision accident from the suspected collision time point, and outputs information about the target collision time point. The driver's driving behavior characteristic data mapped is the driver's driving behavior data that is different from the driving behavior data when the vehicle is driving normally; the driver's driving behavior characteristic data mapped is based on at least one of the following data: a change in the effective travel of the brake pedal, a change in the steering wheel angle, a rotation speed of the steering wheel angle, and a change in the accelerator pedal opening, Among them, the characteristic data of the vehicle's driving state that is mapped is vehicle driving state data that is different from the driving state data of the vehicle when it is driving normally; the characteristic data of the vehicle's driving state that is mapped is based on the change of the acceleration of the vehicle in at least one of the lateral, longitudinal and vertical directions.

13. A vehicle collision accident detection system, comprising: processor; and A storage device for storing a program, The processor executes the program to implement the vehicle collision accident detection method according to any one of claims 1-11.

14. A computer readable medium having a computer program stored thereon, wherein the program is executed by a processor to implement the vehicle collision accident detection method according to any one of claims 1 to 11.

Images