Vehicle risk prediction method, device, equipment and readable storage medium

Abstract

The present application discloses a vehicle risk prediction method, device, equipment and readable storage medium, which relates to the field of vehicle safety technology, including: obtaining the driver status information of the vehicle to be detected, and determining the driver status score according to the driver status information; obtaining the vehicle status information of the vehicle to be detected, and determining the vehicle status score according to the vehicle status information; obtaining the load status information of the vehicle to be detected, and determining the load status score according to the load status information; obtaining the special situation score of the vehicle to be detected; obtaining each path of the vehicle to be detected, and obtaining the path weight and path score corresponding to each path; inputting the driver status score, vehicle status score, load status score, special situation score, each path weight and each path score into the accident risk prediction model obtained by pre-training, and outputting the risk prediction result corresponding to the vehicle to be detected. The present application enables the accident risk prediction model to have a clearer focus and more accurate prediction.

Description

Technical Field

[0001] The present application relates to the field of vehicle safety technology, and in particular to a vehicle risk prediction method, device, equipment and computer-readable storage medium. Background Art

[0002] Road traffic accidents are mainly caused by dangerous vehicles, which usually refer to vehicles that have higher risks due to their nature, state or driver behavior, such as large trucks, old vehicles, overloaded vehicles, fatigued driving vehicles, etc. Most road traffic accidents are caused by dangerous vehicles. Especially when these dangerous vehicles are driving on some dangerous roads, such as tunnels, bridges and other complex or poor road sections, the probability of accidents will increase significantly. Therefore, identifying dangerous vehicles on the road, predicting and controlling the path of dangerous vehicles is of great significance to reducing road traffic accidents and reducing the loss of life and property.

[0003] In summary, how to effectively solve the problem of being unable to predict vehicle risks is an issue that technical personnel in this field urgently need to solve. Summary of the invention

[0004] The purpose of this application is to provide a vehicle risk prediction method, which enables the accident risk prediction model to have a clearer focus, thereby achieving a more accurate prediction effect; another purpose of this application is to provide a vehicle risk prediction device, equipment and computer-readable storage medium.

[0005] In order to solve the above technical problems, this application provides the following technical solutions:

[0006] A vehicle risk prediction method, comprising:

[0007] Acquiring driver status information of the vehicle to be detected, and determining a driver status score according to the driver status information;

[0008] Acquiring vehicle status information of the vehicle to be detected, and determining a vehicle status score according to the vehicle status information;

[0009] Acquiring the loading status information of the vehicle to be detected, and determining the loading status score according to the loading status information;

[0010] Obtaining a special situation score of the vehicle to be detected;

[0011] Obtaining each path of the vehicle to be detected, and obtaining the path weight and path score corresponding to each path;

[0012] The driver status score, the vehicle status score, the cargo status score, the special situation score, each path weight and each path score are input into the accident risk prediction model obtained by pre-training, and the risk prediction result corresponding to the vehicle to be detected is output.

[0013] In a specific implementation of the present application, obtaining the driver status information of the vehicle to be detected, and determining the driver status score according to the driver status information, includes:

[0014] Acquiring the driving behavior of the driver in the vehicle to be detected, and determining a driving behavior score according to the driving behavior;

[0015] Acquiring the driving emotion of the driver in the vehicle to be detected, and determining a driving emotion score according to the driving emotion;

[0016] Obtaining preset driving behavior weights and driving emotion weights;

[0017] The driving behavior score, the driving emotion score, the driving behavior weight and the driving emotion weight are weighted to obtain the driver status score.

[0018] In a specific implementation of the present application, obtaining the vehicle status information of the vehicle to be detected, and determining the vehicle status score according to the vehicle status information, includes:

[0019] Acquire various vehicle status items of the vehicle to be detected; wherein each vehicle status item includes vehicle weight, vehicle length, vehicle width, vehicle height, vehicle speed, vehicle temperature and vehicle maintenance information;

[0020] Find the vehicle status item score corresponding to each vehicle status item from the vehicle status item score table;

[0021] The scores of each vehicle status item are summed up to obtain the vehicle status score.

[0022] In a specific implementation of the present application, obtaining the loading status information of the vehicle to be detected, and determining the loading status score according to the loading status information, includes:

[0023] Obtaining a content status item, a load assembly status item, and a load leakage status item of the vehicle to be detected;

[0024] Finding the load content status score corresponding to the load content status item, the load assembly status score corresponding to the load assembly status item, and the load leakage status score corresponding to the load leakage status item from the load status item scoring table;

[0025] The load content status score, the load assembly status score and the load spillage status score are summed up to obtain the load status score.

[0026] In a specific implementation of the present application, obtaining each path of the vehicle to be detected includes:

[0027] Acquire a historical driving trajectory set of the vehicle to be detected; wherein the historical driving trajectory set includes the historical driving trajectory of the vehicle to be detected or the historical driving trajectory of other vehicles of the same type as the vehicle to be detected;

[0028] Obtaining the current position of the vehicle to be detected;

[0029] Selecting a historical driving trajectory that coincides with the current position from the historical driving trajectory set, and determining the selected historical driving trajectory as the inertial trajectory of the vehicle to be detected;

[0030] Obtaining the end point of each inertial trajectory, and obtaining other feasible paths from the navigation system according to the current position and the end point;

[0031] Determine the remaining feasible paths as the regular trajectory of the vehicle to be detected;

[0032] Get the preset inertial path initial weight and regular trajectory initial weight;

[0033] Obtain a first traffic condition weight adjustment factor and a second traffic condition weight adjustment factor; wherein the first traffic condition weight adjustment factor is greater than the second traffic condition weight adjustment factor;

[0034] Obtaining the road condition information of each inertial trajectory and each conventional trajectory respectively;

[0035] Calculate the adjusted path weights corresponding to each inertial trajectory and each regular trajectory according to the road condition information of each inertial trajectory and each regular trajectory, the initial weight of the inertial path, the initial weight of the regular trajectory, the first road condition weight adjustment factor, and the second road condition weight adjustment factor;

[0036] The inertial trajectories and the conventional trajectories are sorted according to the weights of the paths after adjustment, and the paths of the to-be-detected vehicle are obtained by selecting the end with the larger weights after adjustment.

[0037] In a specific implementation of the present application, obtaining the path weight and path score corresponding to each path includes:

[0038] Determine the adjusted path weights corresponding to each path as the path weights corresponding to each path;

[0039] Obtain the road condition score, dangerous vehicle quantity score and environmental condition score corresponding to each path;

[0040] Obtain preset road condition weights, dangerous vehicle quantity weights and environmental condition weights;

[0041] The road condition score, the dangerous vehicle quantity score, the environmental condition score, the road condition weight, the dangerous vehicle quantity weight and the environmental condition weight are weighted to obtain the path score.

[0042] In a specific implementation of the present application, after outputting the risk prediction result corresponding to the vehicle to be detected, the method further includes:

[0043] When the risk prediction result indicates that the accident risk of the vehicle to be detected is high, preset control measures are taken to perform risk control; wherein the risk prediction result includes a high accident risk and a low accident risk.

[0044] A vehicle risk prediction device, comprising:

[0045] A driver status score determination module, used to obtain driver status information of a vehicle to be detected and determine a driver status score according to the driver status information;

[0046] A vehicle status score determination module, used to obtain vehicle status information of the vehicle to be detected, and determine a vehicle status score according to the vehicle status information;

[0047] A load status score determination module, used to obtain the load status information of the vehicle to be detected, and determine the load status score according to the load status information;

[0048] A special situation score determination module, used to obtain the special situation score of the vehicle to be detected;

[0049] A path weight and path score acquisition module is used to acquire each path of the vehicle to be detected, and acquire the path weight and path score corresponding to each path;

[0050] The risk prediction result output module is used to input the driver status score, the vehicle status score, the cargo status score, the special situation score, the weight of each path and the score of each path into the accident risk prediction model obtained by pre-training, and output the risk prediction result corresponding to the vehicle to be detected.

[0051] A vehicle risk prediction device, comprising:

[0052] Memory for storing computer programs;

[0053] A processor is used to implement the steps of the vehicle risk prediction method as described above when executing the computer program.

[0054] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the vehicle risk prediction method as described above are implemented.

[0055] The vehicle risk prediction method provided in the present application obtains the driver status information of the vehicle to be detected, and determines the driver status score according to the driver status information; obtains the vehicle status information of the vehicle to be detected, and determines the vehicle status score according to the vehicle status information; obtains the load status information of the vehicle to be detected, and determines the load status score according to the load status information; obtains the special situation score of the vehicle to be detected; obtains each path of the vehicle to be detected, and obtains the path weight and path score corresponding to each path; inputs the driver status score, vehicle status score, load status score, special situation score, each path weight and each path score into the accident risk prediction model obtained by pre-training, and outputs the risk prediction result corresponding to the vehicle to be detected.

[0056] It can be seen from the above technical scheme that the driver's status and driving status can be obtained in real time through the driver's status information, and a more accurate and comprehensive judgment can be achieved through the driver's status score, vehicle status score, cargo status score, and special situation score, thereby improving the integrity of vehicle identification factors. By introducing special situations for special factors, more accurate and reasonable dangerous vehicle identification can be achieved through manual adjustment. The introduction of special factors makes the proposed vehicle risk prediction method more adaptable to local conditions and in line with the actual situation of a specific area or location. It is considered that the risk of a vehicle accident is not only related to the degree of danger of the vehicle, but also to the degree of danger of the driving path. The weights of each path and the scores of each path are also input into the accident risk prediction model for vehicle risk prediction. By predicting and judging vehicle accident risks from all angles of driver, vehicle and path, and classifying factors, the accident risk prediction model has a clearer focus, thereby achieving a more accurate prediction effect.

[0057] Correspondingly, the present application also provides a vehicle risk prediction device, equipment and computer-readable storage medium corresponding to the above-mentioned vehicle risk prediction method, which have the above-mentioned technical effects and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0059] Figure 1 This is an implementation flow chart of the vehicle risk prediction method in the embodiment of the present application;

[0060] Figure 2 Another implementation flow chart of the vehicle risk prediction method in the embodiment of the present application;

[0061] Figure 3 This is a structural block diagram of a vehicle risk prediction device in an embodiment of the present application;

[0062] Figure 4 This is a structural block diagram of a vehicle risk prediction device in an embodiment of the present application;

[0063] Figure 5 A schematic diagram of the specific structure of a vehicle risk prediction device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0064] In order to enable those skilled in the art to better understand the present application, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0065] See also Figure 1 , Figure 1 This is an implementation flow chart of the vehicle risk prediction method in the embodiment of the present application. The method may include the following steps:

[0066] S101: Acquire driver status information of a vehicle to be detected, and determine a driver status score according to the driver status information.

[0067] Driver status scores corresponding to different driver statuses are preset. When risk prediction of a vehicle to be detected is required, the driver status information of the vehicle to be detected is obtained, and the driver status score is determined based on the driver status information.

[0068] Driver status information may include the driver's driving behavior and driving emotions during driving. Dangerous driving behavior and driving emotions may lead to a higher accident rate.

[0069] S102: Acquire vehicle status information of the vehicle to be detected, and determine a vehicle status score according to the vehicle status information.

[0070] The vehicle status scores corresponding to different vehicle statuses are preset, the vehicle status information of the vehicle to be detected is obtained, and the vehicle status score is determined according to the vehicle status information.

[0071] Vehicle status may include the vehicle's weight, length, width, height, speed, temperature, and maintenance condition.

[0072] S103: Obtaining the loading status information of the vehicle to be detected, and determining the loading status score according to the loading status information.

[0073] The load status scores corresponding to different load statuses are preset, the load status information of the vehicle to be detected is obtained, and the load status score is determined according to the load status information.

[0074] The cargo status may include cargo contents, cargo assembly, and cargo spillage.

[0075] S104: Obtain the special situation score of the vehicle to be detected.

[0076] The special situation score of the vehicle to be tested can be obtained through experience and historical data corresponding to the vehicle to be tested, which shows special situations where the hazard rate and accident rate are extremely high or extremely low.

[0077] For example, a certain high-risk vehicle type, a certain high-risk cargo transportation, an ambulance, etc., by setting scores (positive or negative) for these special situations, the accuracy of judging the danger level of the vehicle in this situation can be improved. Vehicles that meet a certain special situation are set to the corresponding score, otherwise they are set to 0. The sum of the scores of all special situation items is used as the special situation score W.

[0078] S105: Obtain each path of the vehicle to be detected, and obtain the path weight and path score corresponding to each path.

[0079] The risk level of a vehicle is not only related to the dangerous situation of the vehicle itself, but also to the path of the vehicle. Obtain each path of the vehicle to be detected, and obtain the path weight and path score corresponding to each path.

[0080] S106: Input the driver status score, vehicle status score, cargo status score, special situation score, path weights and path scores into the pre-trained accident risk prediction model, and output the risk prediction result corresponding to the vehicle to be detected.

[0081] The accident risk prediction model is pre-trained. After obtaining the scores of each path, the driver status score, vehicle status score, cargo status score, special situation score, each path weight and each path score are input into the accident risk prediction model obtained by pre-training, and the risk prediction result corresponding to the vehicle to be detected is output.

[0082] It can be seen from the above technical scheme that the driver's status and driving status can be obtained in real time through the driver's status information, and a more accurate and comprehensive judgment can be achieved through the driver's status score, vehicle status score, cargo status score, and special situation score, thereby improving the integrity of vehicle identification factors. By introducing special situations for special factors, more accurate and reasonable dangerous vehicle identification can be achieved through manual adjustment. The introduction of special factors makes the proposed vehicle risk prediction method more adaptable to local conditions and in line with the actual situation of a specific area or location. It is considered that the risk of a vehicle accident is not only related to the degree of danger of the vehicle, but also to the degree of danger of the driving path. The weights of each path and the scores of each path are also input into the accident risk prediction model for vehicle risk prediction. By predicting and judging vehicle accident risks from all angles of driver, vehicle and path, and classifying factors, the accident risk prediction model has a clearer focus, thereby achieving a more accurate prediction effect.

[0083] It should be noted that, based on the above embodiments, the embodiments of the present application also provide corresponding improved solutions. In the subsequent embodiments, the same steps or corresponding steps as those in the above embodiments can be referenced to each other, and the corresponding beneficial effects can also be referenced to each other, which will not be repeated one by one in the following improved embodiments.

[0084] See also Figure 2 , Figure 2 This is another implementation flow chart of the vehicle risk prediction method in the embodiment of the present application. The method may include the following steps:

[0085] S201: Acquire driver status information of a vehicle to be detected, and determine a driver status score according to the driver status information.

[0086] S202: Acquire vehicle status information of the vehicle to be detected, and determine a vehicle status score according to the vehicle status information.

[0087] S203: Obtaining the loading status information of the vehicle to be detected, and determining the loading status score according to the loading status information.

[0088] S204: Obtain the special situation score of the vehicle to be detected.

[0089] The driver status score can be recorded as X, the vehicle status score can be recorded as Y, the cargo status score can be recorded as Z, and the special situation score can be recorded as W. The final score of the vehicle's danger score is set to S=0.4X+0.3*Y+0.3*Z+W. During a driving process, the above states of the vehicle are likely to remain unchanged, so the above data does not need to be updated in real time, but only needs to be updated every once in a while. The score threshold of dangerous vehicles can be initially set to T=70, and the vehicle is considered to be a dangerous vehicle when the final score S>=T.

[0090] The final score S of the vehicle danger score can be set to be updated every 10 minutes, and the score threshold of dangerous vehicles can be dynamically adjusted according to the local accident situation and the danger score of the accident vehicles at each time. If the danger score S of a large number of vehicles involved in accidents is statistically significantly lower than the threshold T, it means that the dangerous vehicle has not been successfully identified and the threshold score T needs to be lowered. The vehicle status and cargo conditions also need to be dynamically adjusted according to the statistics of local vehicle accident causes to improve the accuracy of dangerous vehicle identification.

[0091] S205: Obtain a set of historical driving trajectories of the vehicle to be detected.

[0092] The historical driving trajectory set includes the historical driving trajectory of the vehicle to be detected or the historical driving trajectory of other vehicles of the same type as the vehicle to be detected.

[0093] A historical driving trajectory set of the vehicle to be detected is obtained, and the historical driving trajectory set includes the historical driving trajectory of the vehicle to be detected or the historical driving trajectory of other vehicles of the same type as the vehicle to be detected. For example, the historical driving trajectory set of the vehicle to be detected can be obtained from a traffic management system and a navigation system in descending order of driving frequency. If there is no historical driving trajectory of the vehicle to be detected, the historical driving trajectory of other vehicles of the same type as the vehicle to be detected is obtained because similar vehicles, such as muck trucks and muck trucks, ambulances and ambulances, have similar driving trajectories.

[0094] S206: Obtain the current position of the vehicle to be detected.

[0095] The current position of the vehicle to be detected is obtained. For example, the current position of the vehicle to be detected can be determined by a navigation system.

[0096] S207: Selecting a historical driving trajectory that coincides with the current position from the historical driving trajectory set, and determining the selected historical driving trajectory as the inertial trajectory of the vehicle to be detected.

[0097] After obtaining the historical driving trajectory set of the vehicle to be detected and the current position of the vehicle to be detected, the historical driving trajectory that coincides with the current position is selected from the historical driving trajectory set, and the selected historical driving trajectory is determined as the inertial trajectory of the vehicle to be detected, which can be the set M.

[0098] S208: Obtain the end point of each inertial trajectory, and obtain the remaining feasible paths from the navigation system according to the current position and the end point.

[0099] After the inertial trajectory of the vehicle to be detected is selected, the end point of each inertial trajectory is obtained, and the remaining feasible paths are obtained from the navigation system according to the current position and the end point.

[0100] S209: Determine the remaining feasible paths as regular trajectories of the vehicle to be detected.

[0101] After obtaining the remaining feasible paths, each of the remaining feasible paths is determined as a regular trajectory of the vehicle to be detected, which can be recorded as a set N.

[0102] S210: Obtaining a preset inertial path initial weight and a conventional trajectory initial weight.

[0103] The inertial path initial weight corresponding to the inertial trajectory and the regular trajectory initial weight corresponding to the regular trajectory are respectively pre-set to obtain the preset inertial path initial weight and regular trajectory initial weight. For example, the inertial path initial weight can be set to 1.2, and the regular trajectory initial weight can be set to 1.

[0104] S211: Obtain a first traffic condition weight adjustment factor and a second traffic condition weight adjustment factor.

[0105] Among them, the first road condition weight adjustment factor is greater than the second road condition weight adjustment factor.

[0106] The first road condition weight adjustment factor and the second road condition weight adjustment factor can be set in advance according to the road condition, such as the first road condition weight adjustment factor corresponds to good road conditions, the second road condition weight adjustment factor corresponds to bad road conditions, and the first road condition weight adjustment factor is greater than the second road condition weight adjustment factor. The first road condition weight adjustment factor and the second road condition weight adjustment factor are obtained.

[0107] And the corresponding second road condition weight adjustment factor can be set to be adjustable when the road condition is bad. The second road condition weight adjustment factor can be set to 1 when the road condition is good, and to 0.8 when the road condition is bad. The second road condition weight adjustment factor is initialized to 0.8 when the road condition is bad, and can be adjusted to 0.9 based on the analysis of subsequent operation data. For example, if a certain type of vehicle still maintains its original trajectory when the road condition is bad, it indicates that the inertia factor has a greater impact, and the second road condition weight adjustment factor is adjusted to 0.9. The path score in the historical driving records of the vehicle to be detected or other vehicles of the same type as the vehicle to be detected can be obtained from the historical driving data, and the road condition is determined to be poor based on the path score. If the proportion of poor road conditions in all driving records is greater than 50%, it is considered that the inertia factor has a greater impact, and the second road condition weight adjustment factor is set to 0.9, otherwise the second road condition weight adjustment factor is 0.8.

[0108] S212: Obtaining road condition information of each inertial trajectory and each conventional trajectory respectively.

[0109] After obtaining each inertial trajectory and each regular trajectory of the vehicle to be detected, the road condition information of each inertial trajectory and each regular trajectory is obtained respectively.

[0110] S213: Calculate the adjusted path weights corresponding to each inertial trajectory and each regular trajectory according to the road condition information of each inertial trajectory and each regular trajectory, the initial weight of the inertial path, the initial weight of the regular trajectory, the first road condition weight adjustment factor and the second road condition weight adjustment factor.

[0111] After obtaining the road condition information of each inertial trajectory and each regular trajectory, and obtaining the initial weight of the inertial path, the initial weight of the regular trajectory, the first road condition weight adjustment factor, and the second road condition weight adjustment factor, the adjusted path weights corresponding to each inertial trajectory and each regular trajectory are calculated according to the road condition information of each inertial trajectory and each regular trajectory, the initial weight of the inertial path, the initial weight of the regular trajectory, the first road condition weight adjustment factor, and the second road condition weight adjustment factor. For example, the adjusted path weights corresponding to each inertial trajectory and each regular trajectory can be obtained by multiplying the corresponding initial weights and the road condition weight adjustment factors according to the road condition information of each inertial trajectory and each regular trajectory.

[0112] Continuing with the example in step S212, when the second road condition weight adjustment factor is 0.9, the adjusted weights are: inertial path with good road conditions > inertial path with bad road conditions > conventional path with good road conditions > conventional path with bad road conditions; otherwise, when the second road condition weight adjustment factor is 0.8, the inertial path with good road conditions > conventional path with good road conditions > inertial path with bad road conditions > conventional path with bad road conditions.

[0113] S214: sorting the inertial trajectories and the conventional trajectories according to the adjusted weights of the paths, and selecting the paths of the to-be-detected vehicle from the end with the larger adjusted weights.

[0114] After calculating the adjusted path weights corresponding to each inertial trajectory and each conventional trajectory, the inertial trajectory and each conventional trajectory are sorted according to the size of the adjusted path weights, and the paths of the vehicles to be detected are selected from the end with the larger adjusted path weights.

[0115] S215: Determine the adjusted path weights corresponding to each path as the path weights corresponding to each path.

[0116] After each path of the to-be-detected vehicle is selected, the adjusted path weights corresponding to each path are determined as the path weights corresponding to each path.

[0117] Continuing with the above example, after weight adjustment, only the 10 paths with the highest weights in set M and set N are retained according to weight sorting. The possible weight values ​​after path adjustment are 1.2, 1.0, 0.96, 0.8 or 1.2, 1.08, 1.0, 0.9. There will be many paths with the same weight after adjustment, but because the paths are obtained from the traffic management system or navigation system, and the acquisition order includes information such as historical frequency and driving duration, when retaining the 10 paths with the highest weights, it is only necessary to retain them according to the size of the weights. Paths with the same weights have been sorted according to the acquisition order. If there are less than 10 paths, the actual number shall prevail. The set L of these 10 paths is the most likely trajectory set of the vehicle to be detected after comprehensive analysis, and the adjusted weights corresponding to the paths in the set are the probability of the vehicle to be detected traveling on the path.

[0118] S216: Obtain the road condition score, dangerous vehicle quantity score and environmental condition score corresponding to each path.

[0119] The road condition score, the number of dangerous vehicles score and the environmental condition score are scored for each path in advance, and after the path weights corresponding to each path are determined, the road condition score, the number of dangerous vehicles score and the environmental condition score corresponding to each path are obtained.

[0120] Road condition analysis is divided into three aspects: road, vehicle and environment. Each path in the path set L is analyzed to obtain the danger score of each path. In terms of roads, four factors are considered, including road conditions, such as flatness, potholes, ruggedness, straightness, whether dangerous sections such as bridges and tunnels are included, the probability of historical road accidents, and traffic volume. The score is I=100*(a+b+c+d / 600), and the upper limit of the score is 100, where a represents the road condition and takes a value of 0 or 1 (when the length of the road with potholes accounts for 30% or more of the total length, the value is 1, otherwise it is 0); b represents whether dangerous sections such as bridges and tunnels are included, which is obtained from the map navigation data and takes a value of 1 or 0, 1 represents the existence of dangerous sections such as bridges or tunnels, and 0 represents the absence of dangerous sections such as bridges or tunnels; c represents the probability of road accidents, which takes a value of [0, 1]. If the probability of road accidents is greater than the preset probability value, it takes 1, and if the probability of road accidents is less than or equal to the preset probability value, it takes 0; d represents the number of vehicles, and 600 can be set as the upper limit of the number of vehicles, that is, the number of vehicles exceeding 600 is also recorded as 600.

[0121] The vehicle aspect mainly focuses on the number of dangerous vehicles on the current path. The more dangerous vehicles there are on the same path, the more likely an accident will occur. The current vehicle information on each path is obtained from the traffic management system to calculate and obtain the number of dangerous vehicles. By analyzing the number of dangerous vehicles on a specific path, the degree of danger of the path is assessed in a targeted manner, which is also conducive to more reasonable path planning for vehicles, avoiding the situation where dangerous vehicles are concentrated on a certain path, resulting in a surge in the probability of accident risks. The score based on the number of dangerous vehicles is J=100*((x / D) / 0.1), with an upper limit of 100, where x represents the number of dangerous vehicles and D represents the total capacity of road vehicles. If the proportion of dangerous vehicles exceeds 10%, it is considered to be very dangerous.

[0122] The environment mainly focuses on four aspects: road water accumulation, snow accumulation, visibility and line of sight obstruction. The score is K=100(e+f+(200-g / 150)+(50-h / 40)) according to the severity of environmental factors, and the upper limit of the score is 100. Among them, e represents the water accumulation, with a value of 0 or 1, 1 represents obvious water accumulation on the road, and 0 represents no obvious water accumulation on the road; f represents the snow accumulation, with a value of 0 or 1, 1 represents obvious snow accumulation on the road, and 0 represents no obvious snow accumulation on the road; g represents visibility, with a value of 200 when it is greater than 200; h represents light intensity, with a value of 50 when it is greater than 50. The final score of the road condition of a certain path can be set to P=0.3*I+0.3*J+0.4*K, and the weight adjustment is dynamically adjusted according to the actual operation and accident occurrence to achieve more accurate and effective path analysis.

[0123] S217: Obtain preset road condition weights, dangerous vehicle quantity weights, and environmental condition weights.

[0124] The road condition weights, dangerous vehicle number weights and environmental condition weights are pre-set, and after obtaining the road condition scores, dangerous vehicle number scores and environmental condition scores corresponding to each path, the preset road condition weights, dangerous vehicle number weights and environmental condition weights are obtained.

[0125] S218: Perform weighted calculation on the road condition score, the dangerous vehicle quantity score, the environmental condition score, the road condition weight, the dangerous vehicle quantity weight and the environmental condition weight to obtain a path score.

[0126] After obtaining the road condition score, dangerous vehicle number score and environmental condition score, and obtaining the road condition weight, dangerous vehicle number weight and environmental condition weight, the road condition score, dangerous vehicle number score, environmental condition score, road condition weight, dangerous vehicle number weight and environmental condition weight are weighted to obtain the path score.

[0127] S219: Input the driver status score, vehicle status score, cargo status score, special situation score, path weights and path scores into the pre-trained accident risk prediction model, and output the risk prediction result corresponding to the vehicle to be detected.

[0128] The training process of the accident risk prediction model may include:

[0129] The XGBoost model uses Gradient Boosted Trees (gbtree) as the basic algorithm. The core idea of ​​Gradient Boosted Trees is to optimize the prediction ability of the model by gradually fitting the residuals, thereby generating a series of weak learners (decision trees), and finally combining these weak learners into a strong learner.

[0130] Data preprocessing mainly includes removing outliers and filling in blank values, normalizing the driver status score, vehicle status score, cargo status score, and special situation score. The collected historical data is divided into a 4:1 ratio of training set and validation set. The training set is used for model fitting, and the validation set is used for performance evaluation.

[0131] The key hyperparameters of the model are set to boost_type=gbtree, n_estimators=100, max_depth=5, gamma=0.01, and the rest of the parameters use the default values. The objective function L is composed of the loss function l and the regularization term constitute:

[0132] ;

[0133] Among them, n is the number of samples, K is the number of trees, and f is the final prediction model. is the output of the kth tree, is the true label of the i-th sample, is the predicted value of the ith sample.

[0134] The loss function is:

[0135] ;

[0136] The regularization term is:

[0137] ;

[0138] Among them, T is the number of leaf nodes in the tree, w j is the weight of the leaf node, γ and λ are regularization parameters.

[0139] The training process uses a distributed gradient boosting algorithm to gradually optimize the objective function, which mainly includes the following steps:

[0140] (1) Initialize the model , initialize all sample prediction values ​​to a constant.

[0141] (2) In each iteration, a new tree is generated based on the residual predicted by the current model , used to fit the residuals.

[0142] (3) Update the model:

[0143] ;

[0144] Where η is the learning rate, is the output of step t, is the output of step t-1, is the output of the new tree constructed in step t.

[0145] (4) Repeat steps (2) and (3) until the specified number of iterations is completed or the error converges.

[0146] Model Evaluation:

[0147] After training, the validation set is used to evaluate the model performance. The calculation indicators include: Accuracy, Precision, Recall and F1-Score. If the model performance does not meet expectations, adjust the hyperparameters or increase the input features for retraining.

[0148] By incorporating the vehicle's inertial path into the analysis and adjusting its weight according to the actual situation, the special situation for the current vehicle is implied, thereby improving the model's prediction effect for the current vehicle. Since the wrong prediction of a vehicle that is actually high-risk has a greater impact on reality, it is necessary to collect data on vehicles that are predicted by the model to be low-risk but eventually have an accident to iteratively optimize the accident risk prediction model, minimize the false negative rate of the model prediction, and avoid predicting vehicles that are actually high-risk as low-risk, which is more in line with actual needs. When the model prediction is true, it indicates that the current vehicle has a high risk of an accident during this driving process, and corresponding control measures need to be taken for such vehicles to reduce the accident rate. If the model prediction is false, it indicates that the current vehicle has a low risk of an accident during this driving process. Since the various influencing factors are constantly changing, it is still necessary to pay real-time attention to its driving trajectory and related data in the future to predict its accident probability.

[0149] By defining the concepts of inertial path and conventional path, and obtaining the most likely path for the vehicle based on the weight adjustment method, and then conducting a hazard analysis of the path from three aspects: road conditions, vehicles, and environment, the accident risk prediction model can integrate various possible paths and their hazard probabilities to achieve more accurate predictions.

[0150] S220: When the risk prediction result indicates that the accident risk of the vehicle to be detected is high, taking preset control measures to control the risk.

[0151] Among them, the risk prediction results include high accident risk and low accident risk.

[0152] The risk prediction results output by the accident risk prediction model include two situations: high accident risk and low accident risk. After the risk prediction result corresponding to the vehicle to be detected is predicted by the accident risk prediction model, when the risk prediction result indicates that the accident risk of the vehicle to be detected is high, the preset control measures are taken to control the risk.

[0153] Targeted risk control measures are formulated based on the dangerous situations of the driver, vehicle and road. A low level of danger indicates that it is not the root cause of the vehicle accident, thereby reducing the investment in manpower and material resources in this area, and implementing emergency methods mainly for high-risk areas.

[0154] When the driver status score X>=70, it indicates that the driver is in a dangerous driving state. Targeted voice reminders and warnings can be given for dangerous driving behaviors, or the nearest traffic police can be notified to intercept and remind the driver, and corresponding emotional management methods can be provided for dangerous driving emotions to reduce the impact of emotions on the driver.

[0155] For the case where the vehicle danger score S>=70, it indicates that the current vehicle is relatively dangerous. It can be handled from three aspects: traffic control, vehicle control and emergency system linkage. Traffic control timely notifies the vehicle management department and the transportation department of the predicted situation based on the current dangerous situation of the vehicle. After further confirmation, the traffic management department dispatches traffic police to guide the dangerous vehicle on the path, guide nearby normal vehicles to detour to avoid danger, or directly intercept dangerous vehicles in serious situations, and traffic police are required to open the road for special situations. Vehicle control mainly intercepts the vehicle while driving and handles it, reducing the factors that may cause accidents hidden in the vehicle, such as controlling the vehicle speed, reducing the length, height and width, and reducing the weight of the load, which can be controlled. Emergency system linkage is mainly aimed at the handling of dangerous vehicles before and after accidents. Before the accident occurs, the emergency system of the bridge or tunnel is promptly linked, including tunnel ventilation, tunnel fire protection, tunnel lighting, tunnel warning and bridge warning, etc. When the dangerous vehicle enters the bridge or tunnel, the emergency system is ready to start the emergency plan in real time, and at the same time, the dangerous vehicle warning is displayed in real time at the entrance and exit to remind the vehicles entering to be vigilant. At the same time, before an accident occurs, the emergency rescue department will be contacted to inform the dangerous vehicle of its real-time situation. When an accident occurs with a dangerous vehicle, the emergency rescue department will be notified immediately to reduce the response time, so that people can be rescued immediately and the loss of life and property caused by the accident can be reduced.

[0156] For the paths with road condition scores P>=70 in the most likely path set L, the path guidance is mainly used to change the vehicle's path to a safer road to reduce the impact of the road on the risk of vehicle accidents. Path guidance mainly adjusts the vehicle's path, and changes the navigation route by linking the vehicle's navigation system to avoid driving on dangerous roads, so that the vehicle's route avoids dangerous sections such as bridges or tunnels, and avoids roads with a large number of dangerous vehicles, reducing the probability of accidents.

[0157] The settings of the above score thresholds are dynamically adjusted according to actual conditions, and the emphasis may be different in each period of time to achieve more accurate and regionally targeted traffic control methods.

[0158] In a specific implementation of the present application, obtaining the driver status information of the vehicle to be detected and determining the driver status score according to the driver status information may include the following steps:

[0159] Step 1: Obtain the driving behavior of the driver in the vehicle to be detected, and determine the driving behavior score based on the driving behavior;

[0160] Step 2: Obtain the driving emotion of the driver in the vehicle to be detected, and determine the driving emotion score according to the driving emotion;

[0161] Step 3: Obtaining preset driving behavior weights and driving emotion weights;

[0162] Step 4: Perform weighted calculation on the driving behavior score, driving emotion score, driving behavior weight and driving emotion weight to obtain the driver status score.

[0163] For the convenience of description, the above four steps can be combined for explanation.

[0164] The driving behavior of the driver in the vehicle to be detected is obtained, and a driving behavior score is determined based on the driving behavior. The driving emotion of the driver in the vehicle to be detected is obtained, and a driving emotion score is determined based on the driving emotion. The preset driving behavior weight and driving emotion weight are obtained, and the driving behavior score, driving emotion score, driving behavior weight and driving emotion weight are weightedly calculated to obtain a driver status score.

[0165] The camera can be used to obtain the cab video image data and audio data in real time, and the human posture estimation algorithm can be used to identify the driver's driving action changes in real time, such as focusing on the local action changes of the hands and head, frequent head turning, driving posture, etc. The risk factor of these dangerous action changes can be set to w1=0.5. The target detection algorithm and speech recognition algorithm are used to identify the driver's driving performance. For example, the target detection algorithm is used to focus on identifying behaviors such as not wearing a seat belt, playing with a mobile phone, and eating. The speech recognition algorithm is used to identify the driver's conversation content and extract key information to analyze the cab situation. The risk factor of the above dangerous performance can be set to w2=0.5. Incorrect driving behavior and driving performance imply problems such as fatigue driving and distracted driving, which greatly increases the probability of accidents.

[0166] Use facial expression recognition algorithms to identify the driver's emotional changes during driving, judge the driver's negative emotions such as tension, anxiety and anger, and promptly discover the driver's emotional driving. Different dangerous factors w3 are set for different negative emotions. The above human posture estimation model, target detection model, speech recognition model and facial recognition model are specifically constructed, mainly for the unique and specific scene of the cab, focusing on the local information of the hands and head, focusing on the recognition accuracy of common driving items such as food, electronic equipment and seat belts, focusing on the high-frequency communication vocabulary that may appear in the cab, and focusing on common facial emotions during driving. In the process of model recognition, the acquired data is used to further iterate and optimize the recognition effect and pertinence of the model, and improve the reliability of the model in the current specified scene.

[0167] The risk scores of the driving behavior and driving emotions of the above driver status are set to 60 (30 + 30) and 60 respectively. When the driver of the vehicle has dangerous behavior and dangerous driving emotions, the corresponding score t is set respectively, otherwise the score is set to 0. The above scores are multiplied by the sum of the risk coefficients, and the case of more than 100 points is counted as 100, as the driver status score X = t1*w1+t2*w2+t3*w3, where t1 is the driving action change score, t2 is the driving performance score, and t3 is the driving emotion score. The higher the score, the more dangerous the driver's driving situation. Since the driver's status will change frequently, the above data needs to be acquired, identified and updated in real time.

[0168] The innovative addition of artificial intelligence methods to judge the driver's posture, expression and cab conditions can obtain the driver's condition and driving status in real time, and continuously iterate based on operating data to achieve high-precision recognition of the predictive model in the specific environment of the cab.

[0169] In a specific implementation of the present application, obtaining vehicle status information of a vehicle to be detected and determining a vehicle status score according to the vehicle status information may include the following steps:

[0170] Step 1: Obtain various vehicle status items of the vehicle to be detected; wherein each vehicle status item includes vehicle weight, vehicle length, vehicle width, vehicle height, vehicle speed, vehicle temperature and vehicle maintenance information;

[0171] Step 2: Find the vehicle status item scores corresponding to each vehicle status item from the vehicle status item score table;

[0172] Step 3: Sum the scores of each vehicle status item to obtain the vehicle status score.

[0173] For the convenience of description, the above three steps can be combined for explanation.

[0174] Each vehicle status item includes vehicle weight, vehicle length, vehicle width, vehicle height, vehicle speed, vehicle temperature and vehicle maintenance information. Each vehicle status item of the vehicle to be detected is obtained, and the vehicle status item score corresponding to each vehicle status item is found from the vehicle status item score table. The vehicle status score is summed up and calculated. By calculating the vehicle status score according to the vehicle status item score corresponding to each vehicle status item, the accuracy of the vehicle status score is improved.

[0175] The vehicle weight is obtained through the road toll station scale and other means, the vehicle length, height and width are obtained through traffic cameras and other means, the vehicle speed is obtained through cameras and speedometers and other means, the vehicle temperature is obtained through cameras and vehicle thermometers, and the vehicle maintenance status is obtained through cameras, mainly including the damage, deformation, rust and damage of the vehicle body. The scores of the above seven aspects of vehicle status are set to 30, 30, 40, 40, 40, 30, 30 respectively. When the vehicle is overweight, overlong, overhigh, overwide, speeding, high temperature, and poorly maintained, the corresponding scores are set respectively, otherwise the score is set to 0. The sum of the above seven scores is counted as 100 if it exceeds 100 points, which is used as the vehicle status score Y. The higher the score, the higher the probability that the vehicle is a dangerous vehicle.

[0176] In a specific implementation of the present application, obtaining the loading status information of the vehicle to be detected and determining the loading status score according to the loading status information may include the following steps:

[0177] Step 1: Obtain the cargo content status item, cargo assembly status item and cargo leakage status item of the vehicle to be detected;

[0178] Step 2: Find the load content status score corresponding to the load content status item, the load assembly status score corresponding to the load assembly status item, and the load leakage status score corresponding to the load leakage status item from the load status item score table;

[0179] Step 3: Sum the load content status score, load assembly status score and load spillage status score to obtain the load status score.

[0180] For the convenience of description, the above four steps can be combined for explanation.

[0181] Obtain the load content status item, load assembly status item and load spillage status item of the vehicle to be inspected, find the load content status score corresponding to the load content status item, the load assembly status score corresponding to the load assembly status item and the load spillage status score corresponding to the load spillage status item from the load status item scoring table, sum and calculate the load content status score, load assembly status score and load spillage status score to obtain the load status score. The final load status score is calculated by comprehensively calculating the load content status score, load assembly status score and load spillage status score, which greatly improves the accuracy of the calculated load status score.

[0182] The contents of the load can be obtained through the camera by obtaining obvious signs on the vehicle, including various chemicals and dangerous goods, flammable, explosive, toxic and harmful goods. The load assembly is obtained through the camera, mainly including the load fixation status and container damage and deformation. The load leakage is also obtained through the camera, mainly including liquid spillage, cargo falling and other situations. The scores of the above three aspects of the load situation are set to 40, 40, and 40 respectively. When the vehicle carries dangerous goods, the cargo assembly is poor, and the cargo is spilled or dropped, the corresponding scores are set respectively, otherwise the score is set to 0. The sum of the above three scores (the case of more than 100 points is counted as 100) is used as the load situation score Z. The higher the score, the more dangerous the load is and the higher the degree of danger of the vehicle.

[0183] Corresponding to the above method embodiment, the present application also provides a vehicle risk prediction device. The vehicle risk prediction device described below and the vehicle risk prediction method described above can refer to each other.

[0184] See also Figure 3 , Figure 3 This is a structural block diagram of a vehicle risk prediction device in an embodiment of the present application, and the device may include:

[0185] A driver status score determination module 31 is used to obtain driver status information of the vehicle to be detected and determine a driver status score according to the driver status information;

[0186] The vehicle status score determination module 32 is used to obtain the vehicle status information of the vehicle to be detected and determine the vehicle status score according to the vehicle status information;

[0187] The load status score determination module 33 is used to obtain the load status information of the vehicle to be detected and determine the load status score according to the load status information;

[0188] A special case score determination module 34 is used to obtain a special case score of the vehicle to be detected;

[0189] A path weight and path score acquisition module is used to acquire each path of the vehicle to be detected, and acquire the path weight and path score corresponding to each path;

[0190] The risk prediction result output module 35 is used to input the driver status score, vehicle status score, cargo status score, special situation score, each path weight and each path score into the pre-trained accident risk prediction model, and output the risk prediction result corresponding to the vehicle to be detected.

[0191] It can be seen from the above technical scheme that the driver's status and driving status can be obtained in real time through the driver's status information, and a more accurate and comprehensive judgment can be achieved through the driver's status score, vehicle status score, cargo status score, and special situation score, thereby improving the integrity of vehicle identification factors. By introducing special situations for special factors, more accurate and reasonable dangerous vehicle identification can be achieved through manual adjustment. The introduction of special factors makes the proposed vehicle risk prediction method more adaptable to local conditions and in line with the actual situation of a specific area or location. It is considered that the risk of a vehicle accident is not only related to the degree of danger of the vehicle, but also to the degree of danger of the driving path. The weights of each path and the scores of each path are also input into the accident risk prediction model for vehicle risk prediction. By predicting and judging vehicle accident risks from all angles of driver, vehicle and path, and classifying factors, the accident risk prediction model has a clearer focus, thereby achieving a more accurate prediction effect.

[0192] In a specific implementation of the present application, the driver status score determination module includes:

[0193] A driving behavior score determination submodule is used to obtain the driving behavior of the driver in the vehicle to be detected and determine the driving behavior score according to the driving behavior;

[0194] A driving emotion score determination submodule is used to obtain the driving emotion of the driver in the vehicle to be detected and determine the driving emotion score according to the driving emotion;

[0195] A behavior weight and emotion weight acquisition submodule, used to acquire preset driving behavior weight and driving emotion weight;

[0196] The driver status score acquisition submodule is used to perform weighted calculation on the driving behavior score, driving emotion score, driving behavior weight and driving emotion weight to obtain the driver status score.

[0197] In a specific implementation of the present application, the vehicle status score determination module includes:

[0198] The vehicle status item acquisition submodule is used to acquire various vehicle status items of the vehicle to be detected; wherein each vehicle status item includes vehicle weight, vehicle length, vehicle width, vehicle height, vehicle speed, vehicle temperature and vehicle maintenance information;

[0199] The vehicle status item score search submodule is used to search the vehicle status item score corresponding to each vehicle status item from the vehicle status item score table;

[0200] The vehicle status score obtaining submodule is used to sum the scores of each vehicle status item to obtain the vehicle status score.

[0201] In a specific implementation of the present application, the load status score determination module includes:

[0202] A cargo status item acquisition submodule is used to acquire cargo content status items, cargo assembly status items and cargo leakage status items of the vehicle to be detected;

[0203] The load spillage status item score search submodule is used to search the load content status score corresponding to the load content status item, the load assembly status score corresponding to the load assembly status item, and the load spillage status score corresponding to the load spillage status item from the load status item score table;

[0204] The cargo status score obtaining submodule is used to sum and calculate the cargo content status score, the cargo assembly status score and the cargo leakage status score to obtain the cargo status score.

[0205] In a specific implementation of the present application, the path weight and path score acquisition module includes:

[0206] A historical driving trajectory set acquisition submodule is used to acquire a historical driving trajectory set of the vehicle to be detected; wherein the historical driving trajectory set includes the historical driving trajectory of the vehicle to be detected or the historical driving trajectory of other vehicles of the same type as the vehicle to be detected;

[0207] The current position acquisition submodule is used to obtain the current position of the vehicle to be detected;

[0208] An inertial trajectory determination submodule is used to select a historical driving trajectory that coincides with the current position from a historical driving trajectory set, and determine the selected historical driving trajectory as the inertial trajectory of the vehicle to be detected;

[0209] The feasible path acquisition submodule is used to obtain the end point of each inertial trajectory and obtain the remaining feasible paths from the navigation system according to the current position and the end point;

[0210] A conventional trajectory determination submodule, used to determine the remaining feasible paths as the conventional trajectory of the vehicle to be detected;

[0211] The initial weight acquisition submodule is used to obtain the preset inertial path initial weight and the regular trajectory initial weight;

[0212] A traffic weight adjustment factor acquisition submodule, used to acquire a first traffic weight adjustment factor and a second traffic weight adjustment factor; wherein the first traffic weight adjustment factor is greater than the second traffic weight adjustment factor;

[0213] A road condition information acquisition submodule, used to respectively acquire the road condition information of each inertial trajectory and each conventional trajectory;

[0214] A path-adjusted weight calculation submodule, for calculating the path-adjusted weights corresponding to each inertial trajectory and each conventional trajectory respectively according to the road condition information of each inertial trajectory and each conventional trajectory, the initial weight of the inertial path, the initial weight of the conventional trajectory, the first road condition weight adjustment factor and the second road condition weight adjustment factor;

[0215] The path selection submodule is used to sort the inertial trajectories and the conventional trajectories according to the size of the adjusted weights of the paths, and select the paths of the vehicles to be detected from the end with the larger weights after the paths are adjusted.

[0216] In a specific implementation of the present application, the path weight and path score acquisition module includes:

[0217] A path weight determination submodule, used to determine the adjusted path weights corresponding to each path as the path weights corresponding to each path;

[0218] The path condition score acquisition submodule is used to obtain the road condition score, dangerous vehicle quantity score and environmental condition score corresponding to each path;

[0219] The path condition weight acquisition submodule is used to obtain the preset road condition weight, dangerous vehicle quantity weight and environmental condition weight;

[0220] The path score acquisition submodule is used to perform weighted calculation on the road condition score, the dangerous vehicle quantity score, the environmental condition score, the road condition weight, the dangerous vehicle quantity weight and the environmental condition weight to obtain the path score.

[0221] In a specific embodiment of the present application, the device may further include:

[0222] The risk control module is used to take preset control measures to perform risk control after outputting the risk prediction result corresponding to the vehicle to be detected, when the risk prediction result is that the accident risk of the vehicle to be detected is high; wherein the risk prediction result includes high accident risk and low accident risk.

[0223] Corresponding to the above method embodiment, see Figure 4 , Figure 4 This is a schematic diagram of a vehicle risk prediction device provided in this application, which may include:

[0224] A memory 332, for storing computer programs;

[0225] The processor 322 is used to implement the steps of the vehicle risk prediction method of the above method embodiment when executing the computer program.

[0226] For details, please refer to Figure 5 , Figure 5 A specific structural diagram of a vehicle risk prediction device provided for this embodiment, the vehicle risk prediction device may have relatively large differences due to different configurations or performances, and may include a processor (central processing units, CPU) 322 (for example, one or more processors) and a memory 332, and the memory 332 stores one or more computer programs 342 or data 344. Among them, the memory 332 can be a temporary storage or a permanent storage. The program stored in the memory 332 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the data processing device. Furthermore, the processor 322 can be configured to communicate with the memory 332 to execute a series of instruction operations in the memory 332 on the vehicle risk prediction device 301.

[0227] The vehicle risk prediction device 301 may further include one or more power supplies 326 , one or more wired or wireless network interfaces 350 , one or more input and output interfaces 358 , and / or one or more operating systems 341 .

[0228] The steps in the vehicle risk prediction method described above can be implemented by the structure of the vehicle risk prediction device.

[0229] Corresponding to the above method embodiment, the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps can be implemented:

[0230] Obtain driver status information of the vehicle to be detected, and determine the driver status score based on the driver status information; obtain vehicle status information of the vehicle to be detected, and determine the vehicle status score based on the vehicle status information; obtain cargo status information of the vehicle to be detected, and determine the cargo status score based on the cargo status information; obtain special situation score of the vehicle to be detected; obtain each path of the vehicle to be detected, and obtain the path weight and path score corresponding to each path; input the driver status score, vehicle status score, cargo status score, special situation score, each path weight and each path score into the accident risk prediction model obtained by pre-training, and output the risk prediction result corresponding to the vehicle to be detected.

[0231] The computer-readable storage medium may include: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0232] For an introduction to the computer-readable storage medium provided in this application, please refer to the above method embodiment, and this application will not go into details here.

[0233] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the devices, equipment and computer-readable storage media disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0234] Specific examples are used herein to illustrate the principles and implementation methods of the present application, and the description of the above embodiments is only used to help understand the technical solution and core ideas of the present application. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the scope of protection of the present application.

Claims

1. A vehicle risk prediction method, characterized in that: include: Acquiring driver status information of the vehicle to be detected, and determining a driver status score according to the driver status information; Acquiring vehicle status information of the vehicle to be detected, and determining a vehicle status score according to the vehicle status information; Acquiring the loading status information of the vehicle to be detected, and determining the loading status score according to the loading status information; Obtaining a special situation score of the vehicle to be detected; Acquire a historical driving trajectory set of the vehicle to be detected; wherein the historical driving trajectory set includes the historical driving trajectory of the vehicle to be detected or the historical driving trajectory of other vehicles of the same type as the vehicle to be detected; Obtaining the current position of the vehicle to be detected; Selecting a historical driving trajectory that coincides with the current position from the historical driving trajectory set, and determining the selected historical driving trajectory as the inertial trajectory of the vehicle to be detected; Obtaining the end point of each inertial trajectory, and obtaining other feasible paths from the navigation system according to the current position and the end point; Determine the remaining feasible paths as the regular trajectory of the vehicle to be detected; Get the preset inertial path initial weight and regular trajectory initial weight; Obtain a first traffic condition weight adjustment factor and a second traffic condition weight adjustment factor; wherein the first traffic condition weight adjustment factor is greater than the second traffic condition weight adjustment factor; Obtaining the road condition information of each inertial trajectory and each conventional trajectory respectively; Calculate the adjusted path weights corresponding to each inertial trajectory and each regular trajectory according to the road condition information of each inertial trajectory and each regular trajectory, the initial weight of the inertial path, the initial weight of the regular trajectory, the first road condition weight adjustment factor, and the second road condition weight adjustment factor; The inertial trajectories and the conventional trajectories are sorted according to the weights of the paths after adjustment, and the paths of the vehicle to be detected are obtained by selecting the end with the larger weight after adjustment; Determine the adjusted path weights corresponding to each path as the path weights corresponding to each path; Obtain the road condition score, dangerous vehicle quantity score and environmental condition score corresponding to each path; Obtain preset road condition weights, dangerous vehicle quantity weights, and environmental condition weights; Performing weighted calculation on the road condition score, the dangerous vehicle quantity score, the environmental condition score, the road condition weight, the dangerous vehicle quantity weight and the environmental condition weight to obtain the path score; The driver status score, the vehicle status score, the cargo status score, the special situation score, each path weight and each path score are input into the accident risk prediction model obtained by pre-training, and the risk prediction result corresponding to the vehicle to be detected is output.

2. The vehicle risk prediction method according to claim 1, characterized in that: Acquiring driver status information of the vehicle to be detected, and determining a driver status score according to the driver status information, including: Acquiring the driving behavior of the driver in the vehicle to be detected, and determining a driving behavior score according to the driving behavior; Acquiring the driving emotion of the driver in the vehicle to be detected, and determining a driving emotion score according to the driving emotion; Obtaining preset driving behavior weights and driving emotion weights; The driving behavior score, the driving emotion score, the driving behavior weight and the driving emotion weight are weighted to obtain the driver status score.

3. The vehicle risk prediction method according to claim 1, characterized in that: Acquiring vehicle status information of the vehicle to be detected, and determining a vehicle status score according to the vehicle status information, including: Acquire various vehicle status items of the vehicle to be detected; wherein each vehicle status item includes vehicle weight, vehicle length, vehicle width, vehicle height, vehicle speed, vehicle temperature and vehicle maintenance information; Find the vehicle status item score corresponding to each vehicle status item from the vehicle status item score table; The scores of each vehicle status item are summed up to obtain the vehicle status score.

4. The vehicle risk prediction method according to claim 1, characterized in that: Acquiring the loading status information of the vehicle to be detected, and determining the loading status score according to the loading status information, including: Obtaining a content status item, a load assembly status item, and a load leakage status item of the vehicle to be detected; Finding the load content status score corresponding to the load content status item, the load assembly status score corresponding to the load assembly status item, and the load leakage status score corresponding to the load leakage status item from the load status item scoring table; The load content status score, the load assembly status score and the load spillage status score are summed up to obtain the load status score.

5. The vehicle risk prediction method according to any one of claims 1 to 4, characterized in that: After outputting the risk prediction result corresponding to the vehicle to be detected, the method further includes: When the risk prediction result indicates that the accident risk of the vehicle to be detected is high, preset control measures are taken to perform risk control; wherein the risk prediction result includes a high accident risk and a low accident risk.

6. A vehicle risk prediction device, characterized in that: include: A driver status score determination module, used to obtain driver status information of a vehicle to be detected and determine a driver status score according to the driver status information; A vehicle status score determination module, used to obtain vehicle status information of the vehicle to be detected, and determine a vehicle status score according to the vehicle status information; A load status score determination module, used to obtain the load status information of the vehicle to be detected, and determine the load status score according to the load status information; A special situation score determination module, used to obtain the special situation score of the vehicle to be detected; The path weight and path score acquisition module includes: A historical driving trajectory set acquisition submodule is used to acquire a historical driving trajectory set of the vehicle to be detected; wherein the historical driving trajectory set includes the historical driving trajectory of the vehicle to be detected or the historical driving trajectory of other vehicles of the same type as the vehicle to be detected; The current position acquisition submodule is used to obtain the current position of the vehicle to be detected; An inertial trajectory determination submodule is used to select a historical driving trajectory that coincides with the current position from a historical driving trajectory set, and determine the selected historical driving trajectory as the inertial trajectory of the vehicle to be detected; The feasible path acquisition submodule is used to obtain the end point of each inertial trajectory and obtain the remaining feasible paths from the navigation system according to the current position and the end point; A conventional trajectory determination submodule, used to determine each remaining feasible path as a conventional trajectory of the vehicle to be detected; The initial weight acquisition submodule is used to obtain the preset inertial path initial weight and the regular trajectory initial weight; A traffic weight adjustment factor acquisition submodule, used to acquire a first traffic weight adjustment factor and a second traffic weight adjustment factor; wherein the first traffic weight adjustment factor is greater than the second traffic weight adjustment factor; A road condition information acquisition submodule, used to respectively acquire the road condition information of each inertial trajectory and each conventional trajectory; A path-adjusted weight calculation submodule, for calculating the path-adjusted weights corresponding to each inertial trajectory and each conventional trajectory respectively according to the road condition information of each inertial trajectory and each conventional trajectory, the initial weight of the inertial path, the initial weight of the conventional trajectory, the first road condition weight adjustment factor and the second road condition weight adjustment factor; The path selection submodule is used to sort the inertial trajectories and the conventional trajectories according to the weights of the paths after adjustment, and select the paths of the vehicles to be detected from the end with the larger weights after adjustment; A path weight determination submodule, used to determine the adjusted path weights corresponding to each path as the path weights corresponding to each path; The path condition score acquisition submodule is used to obtain the road condition score, dangerous vehicle quantity score and environmental condition score corresponding to each path; The path condition weight acquisition submodule is used to obtain the preset road condition weight, dangerous vehicle quantity weight and environmental condition weight; The path score obtaining submodule is used to perform weighted calculation on the road condition score, the dangerous vehicle quantity score, the environmental condition score, the road condition weight, the dangerous vehicle quantity weight and the environmental condition weight to obtain the path score; The risk prediction result output module is used to input the driver status score, the vehicle status score, the cargo status score, the special situation score, the weight of each path and the score of each path into the accident risk prediction model obtained by pre-training, and output the risk prediction result corresponding to the vehicle to be detected.

7. A vehicle risk prediction device, characterized in that: include: Memory for storing computer programs; A processor, configured to implement the steps of the vehicle risk prediction method as claimed in any one of claims 1 to 5 when executing the computer program.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the vehicle risk prediction method according to any one of claims 1 to 5 are implemented.

Images