Driving risk assessment method and device based on driving behaviors

Abstract

The invention discloses a driving risk assessment method and device based on driving behaviors. The method comprises the following steps: S1, acquiring parameter data and risk assessment indexes which influence the driving behaviors of a driver; s2, performing feature processing on the driving behavior parameter data of the driver in the step S1; s3, obtaining a logistic regression function; s4, establishing a driving risk assessment model; s5, performing model training on the driving risk assessment model based on a data set formed by the driver driving behavior parameter data after feature processing; and S6, performing driving behavior risk assessment based on the trained driving risk assessment model, and performing optimization suggestion on the driving behavior based on the importance parameters. According to the driving risk assessment method and device based on the driving behavior, the driving behavior of the driver and the risk caused by other surrounding environment factors to the fine-grained driving behavior can be comprehensively considered, the risk of the driving behavior of the driver is analyzed more finely, and optimization suggestions are given.

Description

Technical Field

[0001] The present invention relates to the technical field of risk assessment, and in particular to a driving risk assessment method and device based on driving behavior. Background Art

[0002] In recent years, the road transport industry has grown rapidly to adapt to the circulation needs and production methods of the modern economy, playing an increasingly important role in social and economic life. However, while road transport promotes economic development, promotes social production, and improves people's living standards, it also brings with it increasingly significant road traffic safety issues. Currently, road traffic accidents have become one of the most serious threats to public safety.

[0003] Data shows that driver error is one of the main causes of traffic accidents, followed by traffic accidents caused by the impact of surrounding road conditions on driving behavior. Therefore, in-depth research on differences in drivers' driving behavior and driving abilities as well as changes in surrounding road conditions is of great significance to reducing traffic accidents and ensuring traffic safety.

[0004] Existing driving behavior risk assessment methods only analyze and evaluate the driver's driving behavior data, without comprehensively considering the risks caused by other factors to fine-grained driving behavior. Therefore, there is an urgent need for an assessment method and device that can more finely analyze the risk characteristics of the driver's driving behavior and identify driving behavior risks. Summary of the Invention

[0005] The purpose of the present invention is to provide a driving risk assessment method and device based on driving behavior, which can comprehensively consider the risks caused by the driver's own driving behavior and other surrounding environmental factors to fine-grained driving behavior, so as to solve the problems of existing driving risk assessment indicators being single, the reliability of assessment results being poor, and the assessment effect needing to be improved.

[0006] To achieve the above objectives, the present invention provides a driving risk assessment method based on driving behavior, comprising the following steps:

[0007] S1. Obtain parameter data and risk assessment indicators that affect the driver's driving behavior;

[0008] S2, performing feature processing on the driver's driving behavior parameter data in step S1;

[0009] S3, obtaining a logistic regression function based on the driver's driving behavior parameter data after feature processing in step S2;

[0010] S4. Establish a driving risk assessment model;

[0011] S5. Based on the data set formed by the driver's driving behavior parameter data after feature processing in step S2, a driving risk assessment model is trained;

[0012] S6. Perform driving behavior risk assessment based on the trained driving risk assessment model and make optimization suggestions for driving behavior based on importance parameters.

[0013] Preferably, in step S1, the parameter data affecting the driver's driving behavior includes the in-vehicle driver's driving behavior data, the to-be-measured vehicle's own motion state driving data, and the in-vehicle moving vehicle's motion state driving data, and the driver's driving behavior includes head movements, hand movements, eye movements, and foot movements;

[0014] The motion state driving data of the vehicle to be tested includes the straight state data, turning state data and lane change state data of the vehicle to be tested, and the motion state driving data of the mobile vehicle outside the vehicle includes the distance change data between the vehicle to be tested and the mobile vehicle and the speed data of the mobile vehicle.

[0015] Preferably, in step S1, the driver's driving behavior is quantified based on the parameter data affecting the driver's driving behavior to obtain a driving behavior risk assessment index, including head-down driving, driving with one hand on the steering wheel, fatigue driving, driving while making a phone call, abnormal braking and abnormal vehicle route.

[0016] Preferably, in step S2, the feature processing of the driver's driving behavior parameter data in step S1 includes eliminating abnormal and invalid data.

[0017] Preferably, in step S3, the logistic regression function is:

[0018]

[0019] Where x=(x1,x2,x3...x n ) is the vector of the driver’s driving behavior data after feature processing, Y is the probability of driving behavior risk, g(x)=α0+α1x1+α2x2+…+α n x n ;

[0020] Based on the regression function Obtain a probability function, which is:

[0021]

[0022] Preferably, in step S4, based on the probability function in step S3 The driving risk assessment model is established as follows:

[0023]

[0024] Among them, α0, α1, α2 + … + α n are all calculated by statistical software.

[0025] Preferably, in step S5, the driving risk assessment model is trained based on the random forest classification method, which specifically includes the following steps:

[0026] S51. Randomly and with replacement, draw N sample units from the training set to generate a large number of decision trees;

[0027] S52. Randomly draw m < M variables at each node and use them as candidate variables for splitting the node;

[0028] S53. Completely generate all decision trees without pruning;

[0029] S54. The category of the terminal node is determined by the mode category corresponding to the node;

[0030] S55. For a new observation point, classify it with all the trees, and its category is generated by the majority decision principle.

[0031] Preferably, in step S6, the driving behavior risk assessment specifically includes:

[0032] By analyzing the driving amplitude and the head-down holding time when the driver bows their head, identify whether the driver has the behavior of driving with their head down;

[0033] By analyzing the time when the driver holds the steering wheel with one hand, identify whether the driver has the behavior of driving with one hand on the steering wheel;

[0034] By analyzing the number of times the driver closes their eyes and the time of closing their eyes, identify whether the driver has the behavior of driving while fatigued;

[0035] By analyzing the driver's hand-raising action and the time of holding the hand-raising action, identify whether the driver has the behavior of driving while making a phone call;

[0036] By analyzing the driver's deceleration and emergency braking time, identify whether the driver has the behavior of abnormal braking while driving;

[0037] Collect the driving data of the vehicle to be measured and the moving target vehicle's motion state through the vehicle to be measured and moving target vehicle driving state acquisition module, analyze the angle and number of changes in the vehicle's driving route, and identify whether the driver has the behavior of driving when the vehicle route is abnormal.

[0038] To achieve the above-mentioned objectives, the present invention further provides a driving risk assessment device based on driving behavior, comprising a driver motion capture module, a driving state acquisition module for a vehicle to be measured and a moving target vehicle, a driving behavior analysis module, and a driving behavior feedback module. The driver motion capture module is used to capture and obtain the driver's motions during driving, and the driving state acquisition module for a vehicle to be measured and a moving target vehicle is used to obtain driving state information of the vehicle to be measured and the driving state information of surrounding moving vehicles during driving.

[0039] The driving behavior analysis module determines whether the driver has driving risks within a certain period of time based on the acquired driver action information, the driving status information of the vehicle to be measured during driving, and the driving status information of surrounding moving vehicles, and intervenes and controls the vehicle to be measured based on the judgment result. The driving behavior feedback module generates a corresponding driving behavior feedback report and driving behavior optimization suggestions based on the driving behavior analysis module.

[0040] Preferably, the driver motion capture module includes a hand motion acquisition module, an eye motion acquisition module, a head motion acquisition module, and a foot motion acquisition module;

[0041] The driving state acquisition module of the vehicle to be measured and the moving target vehicle includes a motion state acquisition module of the vehicle to be measured, a distance acquisition module between the vehicle to be measured and the moving vehicle, and a speed acquisition module of the moving vehicle.

[0042] Therefore, the present invention adopts the above-mentioned driving risk assessment method and device based on driving behavior, and the beneficial effects are as follows:

[0043] (1) The present invention can comprehensively consider the risks of the driver's own driving behavior and other surrounding environmental factors on fine-grained driving behavior, analyze the risks of the driver's driving behavior in a more refined manner and provide optimization suggestions.

[0044] (2) The present invention solves the problems that the existing driving risk assessment indicators are single, the reliability of the assessment results is poor, and the assessment effect needs to be improved.

[0045] (3) The present invention can well identify the driver's driving behavior preferences during driving, as well as the impact of surrounding environmental factors on driving behavior, so as to remind the driver to adjust and improve driving behavior in a timely manner and reduce the risk of traffic accidents.

[0046] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 This is an overall flow chart of an embodiment of a driving risk assessment method and device based on driving behavior of the present invention.

[0048] Figure 2 It is a driving behavior parameter data type of an embodiment of a driving risk assessment method and device based on driving behavior of the present invention. DETAILED DESCRIPTION

[0049] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0050] Unless otherwise defined, technical or scientific terms used in the present invention shall have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs.

[0051] like Figure 1 As shown, a driving risk assessment method based on driving behavior includes the following steps:

[0052] S1. Obtain parameter data and risk assessment data that affect the driver's driving behavior;

[0053] The present invention comprehensively considers that parameter data affecting the driver's driving behavior include the driver's driving behavior data inside the vehicle, the vehicle's own motion state driving data to be measured, and the vehicle's motion state driving data outside the vehicle.

[0054] The above data can be obtained through devices such as vehicle-mounted sensors, sensors on smartphones, and cameras installed inside the vehicle. Among them, vehicle-mounted sensors are used to obtain driver operating behavior data and vehicle driving status data; sensors on smartphones such as gyroscopes and accelerometers are used to capture vehicle steering, braking, acceleration and other data; cameras record non-quantitative data such as the movements of various parts of the driver's body.

[0055] The driver's driving behavior includes head movements, hand movements, eye movements and foot movements; the driving data of the vehicle to be tested includes the straight-line state data, turning state data and lane change state data of the vehicle to be tested; the driving data of the moving vehicle outside the vehicle includes the distance change data between the vehicle to be tested and the moving vehicle and the speed data of the moving vehicle.

[0056] Based on the above parameter data affecting the driver's driving behavior, the driver's driving behavior is quantified to obtain driving behavior risk assessment indicators, including driving with head down, driving with one hand on the steering wheel, fatigue driving, driving with a phone call, abnormal braking and abnormal vehicle route.

[0057] S2. Perform feature processing on the driver's driving behavior parameter data in step S1, including eliminating abnormal and invalid data, eliminating useless data and attributes to improve data quality and reliability, and converting the acquired unstructured data such as images into unified structured data. The principal component analysis method is used to reduce the dimension of the data to reduce the amount of calculation and complexity.

[0058] Feature extraction includes temporal features, frequency features, and spatial features. Temporal features include extracting data such as the vehicle speed and acceleration of the vehicle to be measured and moving vehicles from continuous time series data. Frequency features include performing frequency analysis on discrete data, such as frequency, period, and other information. Spatial features mainly consider spatial information such as the position and direction of the vehicle to be measured and moving vehicles.

[0059] S3. Obtain a logistic regression function based on the driver's driving behavior parameter data after feature processing in step S2;

[0060] The logistic regression function is:

[0061]

[0062] where \(x=(x_1,x_2,x_3...x\) n ) is a vector composed of the driver's driving behavior data after feature processing, \(Y\) is the probability of the occurrence of driving behavior risks, and \(g(x)=\alpha_0+\alpha_1x_1+\alpha_2x_2+\cdots+\alpha\) n x n ; <00,00147>

[0063] Based on the regression function ; [[ID=..4]]Obtain a probability function, and the probability function is:

[0064]

[0065] S4. Establish a driving risk assessment model;

[0066] Specifically, based on the probability function in step S3 Establish a driving risk assessment model as follows:

[0067]

[0068] where \(\alpha_0,\alpha_1,\alpha_2+\cdots+\alpha\) n are all calculated by statistical software.

[0069] S5. Based on the data set formed by the driver's driving behavior parameter data after feature processing in step S2, train the driving risk assessment model;

[0070] Train the driving risk assessment model based on the random forest classification method, which specifically includes the following steps: <0000:67>

[0071] S51. Randomly and with replacement draw \(N\) sample units from the training set to generate a large number of decision trees;

[0072] [[ID=:5]]S52. Randomly draw \(m < M\) variables at each node and use them as candidate variables for splitting the node (the number of variables at each node should be the same);

[0073] S53, completely generate all decision trees without pruning (the minimum node is 1);

[0074] S54, the category to which the terminal node belongs is determined by the majority category corresponding to the node;

[0075] S55. For new observation points, use all trees to classify them, and their categories are generated by the majority decision principle.

[0076] The random forest algorithm used in the present invention is to make predictions by constructing multiple decision trees and taking the average of their outputs, which can improve the accuracy and stability of the predictions.

[0077] S6. Perform driving behavior risk assessment based on the trained driving risk assessment model, make optimization suggestions for driving behavior based on importance parameters, and remind drivers to adjust and improve their driving behavior in a timely manner to reduce the risk of traffic accidents.

[0078] The driving behavior risk assessment specifically includes:

[0079] By analyzing the driving range and the time the driver keeps his head down, it can be identified whether the driver has the behavior of lowering his head;

[0080] By analyzing the time the driver holds the steering wheel with one hand, it can be identified whether the driver has the behavior of driving with one hand on the steering wheel;

[0081] By analyzing the number of times and duration of the driver's eyes closing, it can be used to identify whether the driver is driving fatigued.

[0082] By analyzing the driver's hand-raising action and the duration of the hand-raising action, it is possible to identify whether the driver is driving while making a phone call.

[0083] By analyzing the driver's deceleration and sudden braking time, it can identify whether the driver has abnormal braking behavior;

[0084] The driving state acquisition module of the vehicle to be measured and the moving target vehicle collects the driving state data of the vehicle to be measured and the driving state data of the moving vehicle, analyzes the angle and number of vehicle route changes, and identifies whether the driver has any driving behavior when the vehicle route is abnormal, thereby classifying the driving risks caused by the driver's driving behavior into risk levels, such as low risk, medium risk and high risk.

[0085] Specifically, we use evaluation indicators such as accuracy, recall rate, and F1 value to evaluate model performance, optimize the model structure and parameters based on the model's shortcomings, and optimize the model training process by adjusting important parameters such as learning rate and number of iterations.

[0086] A driving risk assessment device based on driving behavior includes a driver motion capture module, a driving state acquisition module for a vehicle to be measured and a moving target vehicle, a driving behavior analysis module, and a driving behavior feedback module. The driver motion capture module is used to capture and obtain the driver's movements during driving. The driving state acquisition module for a vehicle to be measured and a moving target vehicle is used to obtain driving state information of the vehicle to be measured and surrounding moving vehicles during driving.

[0087] The driving behavior analysis module determines whether the driver has driving risks within a certain period of time based on the driver's action information, the driving status information of the vehicle to be measured during driving, and the driving status information of surrounding moving vehicles, and intervenes and controls the vehicle to be measured based on the judgment results. The driving behavior feedback module generates corresponding driving behavior feedback reports and driving behavior optimization suggestions based on the driving behavior analysis module.

[0088] Among them, the driver motion capture module includes a hand motion acquisition module, an eye motion acquisition module, a head motion acquisition module, and a foot motion acquisition module; the driving status acquisition module of the vehicle to be measured and the moving target vehicle includes a motion status acquisition module of the vehicle to be measured, a distance acquisition module between the vehicle to be measured and the moving vehicle, and a speed acquisition module of the moving vehicle.

[0089] The head movement data captured by the driver motion capture module is analyzed to obtain the driving amplitude and the time the driver keeps his head down, so as to identify whether the driver has the behavior of lowering his head while driving;

[0090] The hand motion data captured by the driver motion capture module is analyzed to obtain the time the driver holds the steering wheel with one hand, and to identify whether the driver is driving with one hand on the steering wheel;

[0091] The eye movement data captured by the driver motion capture module is analyzed to obtain the number of times and duration of the driver's eye closure, and to identify whether the driver is driving fatigued.

[0092] Note that if the driver's eyes are too small and there is little difference between opening and closing, the algorithm will have difficulty identifying them. You can also add a mouth movement acquisition module to the driver motion capture module to analyze whether the driver yawns, thereby identifying whether the driver is fatigued driving behavior.

[0093] The hand motion data captured by the driver motion capture module is analyzed to obtain the driver's hand raising action and the duration of the hand raising action, so as to identify whether the driver is driving while making a phone call;

[0094] The foot movement data captured by the driver motion capture module is analyzed to obtain the driver's deceleration and sudden braking time, such as sudden deceleration in a short period of time, sudden speed reduction from high speed to low speed, etc., to identify whether the driver has abnormal braking driving behavior.

[0095] The driving state acquisition module of the vehicle to be measured and the moving target vehicle collects the driving state data of the vehicle to be measured and the driving state data of the moving vehicle, analyzes the angle and number of vehicle route changes, identifies whether the driver has any driving behavior when the vehicle route is abnormal, and thus evaluates the driving risk caused by the driver's driving behavior.

[0096] Therefore, the present invention adopts the above-mentioned driving risk assessment method and device based on driving behavior, comprehensively considering the driver's own driving behavior, the operating status of the vehicle to be tested and the influence of moving vehicles, analyzing the changing trend of the driver's driving status during driving, and analyzing the driving behavior risk in a more granular manner, which is of great significance for preventing traffic accidents and ensuring road safety.

[0097] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A driving risk assessment method based on driving behavior, characterized in that: It includes the following steps: S1. Obtain the parameter data and risk assessment indicators that affect the driving behavior of the driver; S2. Perform feature processing on the driver's driving behavior parameter data in step S1; S3. Obtain the logistic regression function according to the driver's driving behavior parameter data after feature processing in step S2; S4. Establish a driving risk assessment model; S5. Based on the data set formed by the driver's driving behavior parameter data after feature processing in step S2, perform model training on the driving risk assessment model; S6. Perform driving behavior risk assessment based on the trained driving risk assessment model, and give optimization suggestions for driving behavior based on importance parameters.

2. The driving risk assessment method based on driving behavior according to claim 1, characterized in that: In step S1, the parameter data that affect the driving behavior of the driver include the driving behavior data of the driver in the vehicle, the driving data of the待测 vehicle's own motion state, and the driving data of the moving vehicle's motion state outside the vehicle. The driver's driving behavior includes head movements, hand movements, eye movements, and foot movements; The driving data of the待测 vehicle's own motion state includes the straight-line state data, turning state data, and lane-changing state data of the待测 vehicle. The driving data of the moving vehicle's motion state outside the vehicle includes the vehicle distance transformation data between the待测 vehicle and the moving vehicle and the vehicle speed data of the moving vehicle.

3. The driving risk assessment method based on driving behavior according to claim 2, characterized in that: In step S1, based on the parameter data that affect the driving behavior of the driver, the driving behavior of the driver is quantitatively processed to obtain driving behavior risk assessment indicators, including driving with head down, driving with one hand on the steering wheel, fatigue driving, driving while making a call, abnormal braking driving, and abnormal vehicle route.

4. The driving risk assessment method based on driving behavior according to claim 3 is characterized in that: In step S2, the feature processing of the driver's driving behavior parameter data in step S1 includes removing abnormal and invalid data.

5. The driving risk assessment method based on driving behavior according to claim 4 is characterized in that: In step S3, the logistic regression function is: Where x=(x1,x2,x3...x n ) is the vector of the driver’s driving behavior data after feature processing, Y is the probability of driving behavior risk, g(x)=α0+α1x1+α2x2+…+α n x n ; Based on regression function A probability function is obtained, which is:

6. The driving risk assessment method based on driving behavior according to claim 5, characterized in that: In step S4, based on the probability function in step S3 The driving risk assessment model is established as follows: Among them, α0, α1, α2+…+α n All were calculated by statistical software.

7. The driving risk assessment method based on driving behavior according to claim 6, characterized in that: In step S5, the driving risk assessment model is trained based on the random forest classification method, which specifically includes the following steps: S51. Randomly and with replacement draw N sample units from the training set to generate a large number of decision trees; S52. Randomly draw m < M variables at each node and use them as candidate variables for splitting the node; S53. Completely generate all decision trees without pruning; S54. The category of the terminal node is determined by the mode category corresponding to the node; S55. For a new observation point, classify it with all the trees, and its category is generated by the majority decision principle.

8. The driving risk assessment method based on driving behavior according to claim 7 is characterized in that: In step S6, the driving behavior risk assessment specifically includes: By analyzing the driving amplitude and the duration of keeping the head down when the driver lowers the head, identify whether the driver has the behavior of driving with the head down; By analyzing the time when the driver holds the steering wheel with one hand, identify whether the driver has the behavior of driving with one hand on the steering wheel; By analyzing the number of times the driver closes the eyes and the duration of closing the eyes, identify whether the driver has the behavior of fatigue driving; By analyzing the raising hand movement of the driver and the duration of keeping the raising hand movement, identify whether the driver has the behavior of driving while making a call; By analyzing the deceleration and emergency braking time of the driver, identify whether the driver has the behavior of abnormal braking driving; The driving state acquisition module of the vehicle to be measured and the moving target vehicle collects the driving state data of the vehicle to be measured and the driving state data of the moving vehicle, analyzes the angle and number of vehicle route changes, and identifies whether the driver has abnormal driving behavior when the vehicle route is abnormal.

9. A driving risk assessment device based on driving behavior, characterized in that: It includes a driver motion capture module, a driving state acquisition module for the vehicle to be measured and the moving target vehicle, a driving behavior analysis module and a driving behavior feedback module. The driver motion capture module is used to capture and obtain the driver's actions during driving. The driving state acquisition module for the vehicle to be measured and the moving target vehicle is used to obtain the driving state information of the vehicle to be measured and the driving state information of the surrounding moving vehicles during driving. The driving behavior analysis module determines whether the driver has driving risks within a certain period of time based on the acquired driver action information, the driving state information of the vehicle to be measured during driving, and the driving state information of surrounding moving vehicles, and intervenes and controls the vehicle to be measured based on the judgment result. The driving behavior feedback module generates a corresponding driving behavior feedback report and driving behavior optimization suggestions based on the driving behavior analysis module.

10. The driving risk assessment device based on driving behavior according to claim 9, characterized in that: The driver motion capture module includes a hand motion acquisition module, an eye motion acquisition module, a head motion acquisition module, and a foot motion acquisition module; The driving state acquisition module of the vehicle to be measured and the moving target vehicle includes a motion state acquisition module of the vehicle to be measured, a distance acquisition module between the vehicle to be measured and the moving vehicle, and a speed acquisition module of the moving vehicle.

Images