A method for constructing a driving behavior safety evaluation model based on real vehicle testing

Abstract

The present invention relates to the field of automobile driving safety, and discloses a method for constructing a driving behavior safety evaluation model based on real vehicle testing, and the implementation steps are: collecting the driver's eye movement information, automobile operation information, radar information and lane line information during the driving of the automobile on the test road section; calculating the numerical value of each driving behavior safety evaluation index characterization quantity; then using the calculated characterization quantity numerical value as an index, calculating its weight, and finally obtaining a driving behavior safety evaluation model. The present invention uses real vehicle test data to numerically evaluate the driver's driving behavior safety, and the evaluation result has the advantages of visualization and comparability, and can better reflect the driving behavior safety of the driver's regular driving process. At the same time, according to the numerical value of each evaluation index characterization quantity, the common dangerous driving behavior of the driver during the driving process can be known, which is of great significance to the driver to improve driving habits and reduce the occurrence of traffic accidents and traffic violations.

Description

Technical Field

[0001] The present invention relates to the field of automobile driving safety, and in particular to a method for constructing a driving behavior safety evaluation model based on real vehicle testing. Background Art

[0002] In recent years, with the popularization of vehicles, traffic safety issues have become increasingly serious. According to statistics, the number of deaths caused by traffic accidents worldwide exceeds 1.2 million each year. Relevant studies have shown that dangerous driving behavior is one of the main causes of traffic accidents. Among the factors causing traffic accidents issued by domestic and foreign transportation departments, traffic accidents caused by drivers account for more than 70%, and the driving behavior of drivers has a significant impact on the safety and reliability of urban transportation systems. Therefore, it is very necessary to carry out research on the construction of a driving behavior safety evaluation model, which can help drivers discover their own dangerous driving habits and formulate corresponding safe driving habit training programs, thereby improving driving quality and reducing the occurrence of traffic accidents and traffic violations.

[0003] At present, there is no unified and clear standard for the evaluation of driving behavior safety at home and abroad, and the evaluation method has not formed a complete theoretical system. In previous studies, due to the immaturity of driving behavior analysis technology and the difficulty in collecting driver status information, the research on driving behavior safety evaluation methods mainly focused on specific driving conditions such as fatigue driving, emergency conditions and secondary driving. In these studies, the characterization methods of driving behavior safety are relatively simple, the characterization indicators are not comprehensive and systematic, and there is a lack of analysis of driver environmental perception data. In addition, most of the research data are obtained through driving simulators, and there are few analyses and evaluations of driving behavior safety using data obtained when the vehicle is driving in a real road environment, which cannot truly reflect the safety of conventional driving behavior in an actual driving environment. Summary of the invention

[0004] In order to solve the deficiencies mentioned in the above background technology, the purpose of the present invention is to provide a method for constructing a driving behavior safety evaluation model based on real vehicle testing, which can be achieved through the following technical solutions:

[0005] A method for constructing a driving behavior safety evaluation model based on real vehicle testing, the method comprising the following steps:

[0006] Step 1: Selection of experimental subjects and test sections;

[0007] Step 2: Real vehicle data collection and evaluation index calculation;

[0008] Step 3: Calculate the weight of evaluation index;

[0009] Step 4: Construction of driving behavior safety evaluation model.

[0010] Furthermore, it is characterized in that the method for collecting real vehicle data and calculating evaluation indicators in step 2 is:

[0011] Step 2.1, calculation of effective fixation area percentage:

[0012] The areas where the driver's eyes are focused during driving are divided into three categories: the front area: refers to the area that the driver can observe through the front windshield of the car; the rearview mirror area: refers to the area that can be observed through the rearview mirror, left and right rearview mirrors, and the windows on both sides near them; other areas: other areas excluding the front area and the rearview mirror area;

[0013] The basis for judging the driver's gaze area is as follows:

[0014] Gaze angle deviation threshold: 60°;

[0015] Gaze point movement speed: V eye ≤15° / s;

[0016] Fixation duration: t eye ≥0.2s;

[0017] When the driver's eye movement data meets the above conditions, it is considered that the driver's eyes are looking at this area;

[0018] The effective gaze area percentage is represented by X1, which refers to the time T that the driver's eyes focus on the front area and the rearview mirror area during driving. eye The ratio of X1 to the total driving time T is calculated as follows:

[0019]

[0020] X1 is dimensionless; and 0<T sye ≤T, we can see that X1∈(0,1].

[0021] Step 2.2, calculation of braking distance estimation capability:

[0022] The characterization quantity of the braking distance estimation ability is X2, and its calculation process is as follows: the non-contact eye tracker collects the driver's eye movement information to determine the driver's gaze angle α, that is, the angle between the driver's line of sight and the horizontal plane; β represents the angle between the driver's gaze direction and the longitudinal direction; at the same time, the vertical height h of the driver's eyes from the horizontal road surface in a relaxed state is measured, in meters; the longitudinal speed V (km / h) and longitudinal acceleration a (m / s) of the vehicle are obtained according to the vehicle CAN bus and the longitudinal acceleration sensor 2 );

[0023] The calculation formula for the driver's gaze point is as follows:

[0024]

[0025] Here, α, β∈[0, 90°); S * Indicates the longitudinal distance between the driver's gaze point and the vehicle, unit: m:

[0026] If S * If it is too small, it means that the driver does not have enough control over the braking distance of the car at different speeds;

[0027] The calculation formula of the theoretical braking distance of a car is:

[0028]

[0029] S represents the longitudinal distance between the theoretical parking position of the car and the self-vehicle, unit: (m); Δt is the driver's reaction time, here Δt = 1.2s; V0 represents the speed of the self-vehicle during braking (km / h);

[0030] Count the time T when the distance between the driver's gaze point and the vehicle is greater than the theoretical braking distance L ; The time T when the distance between the driver's gaze point and the vehicle is less than the theoretical braking distance S ; The calculation formula of the braking distance estimation capability characterization quantity is:

[0031]

[0032] X2 is dimensionless, X2∈(0,1], the larger X2 is, the greater T L The larger the proportion, the safer the driver's driving behavior.

[0033] Step 2.3, calculation of brake rationality:

[0034] The braking acceleration of the car's rapid deceleration process is set to A, and the deceleration duration is set to t2; when the car's deceleration process meets the following conditions, it can be determined as a braking operation;

[0035] A≤-3m / s 2

[0036] t2≥3s

[0037] The distance between the vehicle and surrounding vehicles is detected by millimeter-wave radar. The rationality of the vehicle's braking is judged by a high-definition camera installed under the front windshield of the vehicle and facing the front. X3 is the representation of the rationality of braking, and the calculation formula is as follows:

[0038]

[0039] Here, X3∈(0,1], dimensionless; B Ris the reasonable number of braking times, unit (times); B A is the actual number of braking times, unit (times); the closer X3 is to 1, the closer the driver's reasonable braking times are to the total number of braking times, and the safer the driving behavior is; conversely, the closer X3 is to 0, the fewer reasonable braking times the driver has, and the worse the driving behavior safety is.

[0040] Step 2.4, longitudinal safety distance maintenance calculation:

[0041] The millimeter-wave radar monitors the distance LC between the vehicle and surrounding vehicles, and the distance between the vehicle and the vehicle in front is less than D R Duration t q and greater than D R Duration t w ; The safety distance maintenance characterization quantity is represented by X4, and its calculation formula is as follows:

[0042]

[0043] X4∈(0, 1], dimensionless; the closer X4 is to 1, the better the driver maintains the safe distance.

[0044] Step 2.5, turn to security:

[0045] The high-definition camera in the car collects the driver's actions to determine whether the driver has turned on the turn signal; the lane line monitoring system determines whether the car has changed lanes; the number of times the driver turns on the turn signal is recorded as K1; the number of times the car changes lanes is K2;

[0046] The indicator of steering safety is X5, and the calculation formula of X5 is:

[0047]

[0048] X5 is dimensionless and does not consider the situation where the driver turns on the turn signal but fails to change lanes. It can be seen that K1≤K2, X5∈(0,1]; the closer X5 is to 1, the more standardized the driver's operation is and the higher the safety is.

[0049] Furthermore, the evaluation index weight calculation method in step 3 is:

[0050] Step 3.1, determine the evaluation index attributes:

[0051] The properties of each evaluation can be known from the analysis and calculation of the representative values ​​of the previous evaluation indicators;

[0052] Step 3.2: Construct the original data matrix, standardize the data, and eliminate the dimension effect. The formula is as follows:

[0053]

[0054] Positive indicators:

[0055]

[0056] Reverse indicators:

[0057]

[0058] Among them, (i=1, 2, 3, ..., 30), (j=1, 2, 3, 4, 5);

[0059] Z ij is the standardized value, x i，j is the value of the characterization quantity of the jth evaluation index of the i-th driver, They represent the maximum and minimum values ​​of the j-th index respectively. The positive and negative indexes are processed according to formulas (9) and (10) respectively.

[0060] Step 3.3, calculate the weight of the i-th driver evaluation index representation quantity under the index:

[0061]

[0062] Thus, the weight matrix Y = {y ij} 30*5 ;

[0063] Step 3.4, calculate the information entropy value e and information utility value d of the indicator:

[0064]

[0065] d j =1-e j (13)

[0066] Step 3.5, calculate the weight of the indicator:

[0067]

[0068] Furthermore, the method for constructing the driving behavior safety evaluation model in step 4 is:

[0069] According to the weights of the evaluation index representation quantities calculated previously, the calculation formula of the driving behavior safety evaluation model can be obtained as follows:

[0070]

[0071] Here, ω1~ω5 are weights of each safety evaluation index representation quantity, ω1~ω5∈(0,1); Z1~Z5 are dimensionless normalized values ​​of each evaluation index representation quantity of the driver;

[0072] The driving behavior safety evaluation score PΔ Expressed in percent:

[0073]

[0074] here It is the maximum value of the standardized value of each evaluation index;

[0075] The final driving behavior safety evaluation model can be obtained:

[0076]

[0077] Furthermore, the process of the driving behavior safety evaluation is as follows:

[0078] The driver needs to drive through the test road. During the collection process, the driver's eye movement information, vehicle operation information, radar information and lane line information are collected. According to formulas (1)-(7), the numerical representation values ​​of the driver's driving behavior safety evaluation index X'1~X'5 are calculated. Then X'1~X'5 are standardized:

[0079]

[0080] Finally, the driver's driving behavior safety score is obtained:

[0081]

[0082] Beneficial effects of the present invention:

[0083] 1. The technical solution of the present invention collects the driver's eye movement information through a non-contact eye tracker to confirm the driver's eye gaze area and gaze angle during driving; collects steering wheel angle and vehicle speed information through the CAN bus; collects the longitudinal acceleration information of the car in real time through the longitudinal acceleration sensor installed on the B-pillar of the car; confirms the relative distance between the car and the surrounding vehicles through the millimeter wave radar; collects the image information of the lane line through the high-definition camera installed on both sides of the front windshield of the car facing the front, and then determines whether the car crosses the lane line through the lane line monitoring system. The information required for this scheme is easy to collect and has good feasibility. The weights of the characterization quantities of each evaluation index in the driving behavior safety evaluation model are calculated using the entropy method. The male-female ratio of the volunteers is 2:1, and they all have good eyesight or good eyesight after wearing glasses. Compared with expert scoring, the weights calculated using the entropy method are more scientific and accurate, which also makes the driving behavior safety evaluation model of the present invention have a wide range of applicability.

[0084] 2. The present invention is applied to the field of driving safety, and uses the data of the driver's actual vehicle test to numerically evaluate the safety of his driving behavior. The evaluation results have the advantages of visualization and comparability, and can better reflect the safety of conventional driving behavior in the actual driving environment. The present invention can be applied to driving tests and safe driving training. Drivers can know their own dangerous driving behaviors according to the numerical values ​​of the characterization quantities of each evaluation index, and then improve their driving habits in a targeted manner and reduce the occurrence of traffic accidents and traffic violations. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] The present invention will be further described below in conjunction with the accompanying drawings.

[0086] Figure 1 : Schematic diagram of the process of building a driver's driving behavior safety assessment model;

[0087] Figure 2 : Analysis diagram of driver’s gaze point and distance from the vehicle. DETAILED DESCRIPTION

[0088] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0089] like Figure 1-2 As shown, a method for constructing a driving behavior safety evaluation model based on real vehicle testing comprises the following steps:

[0090] Step 1: Selection of experimental subjects and test sections;

[0091] Experimental subjects: 30 driver volunteers were recruited as experimental subjects, including 20 males and 10 females. All volunteers held C1 driver's licenses, aged between 18 and 60, had a driving experience of more than 1 year, and had driven more than 3,000 kilometers. All of the above experimental subjects had good vision or good vision after wearing glasses, were in good spirits during the test, and were not affected by alcohol or drugs.

[0092] Test sections: In order to comprehensively examine the safety of drivers' driving behavior under different traffic road environments and speeds, the test sections should include urban conditions, ring roads, highways, and intersections with red lights. During the test, the peak traffic period should be avoided, and the period with smoother traffic flow should be selected for data collection. In order to ensure the accuracy of the test, the entire test section should be no less than 50 km. At the same time, in order to ensure that the test results are closer to the natural driving state, during the driving process of the car, in addition to normal driving needs, the driver must also complete lane changes, U-turns, turns, etc. according to voice prompts. During the test, the driver shall not be informed of the content of the test and what information needs to be collected.

[0093] Step 2: Real vehicle data collection and evaluation index calculation;

[0094] The method for collecting real vehicle data and calculating evaluation indicators in step 2 is:

[0095] Step 2.1, calculation of effective fixation area percentage:

[0096] The areas where the driver's eyes are focused during driving are divided into three categories: the front area: refers to the area that the driver can observe through the front windshield of the car; the rearview mirror area: refers to the area that can be observed through the rearview mirror, left and right rearview mirrors, and the windows on both sides near them; other areas: other areas excluding the front area and the rearview mirror area;

[0097] The basis for judging the driver's gaze area is as follows:

[0098] Gaze angle deviation threshold: 60°;

[0099] Gaze point movement speed: V eye ≤15° / s;

[0100] Fixation duration: t eye ≥0.2s;

[0101] When the driver's eye movement data meets the above conditions, it is considered that the driver's eyes are looking at this area;

[0102] The effective gaze area percentage is represented by X1, which refers to the time T that the driver's eyes focus on the front area and the rearview mirror area during driving. eye The ratio of X1 to the total driving time T is calculated as follows:

[0103]

[0104] X1 is dimensionless; and 0<T eye ≤T, we can see that X1∈(0,1].

[0105] Step 2.2, calculation of braking distance estimation capability:

[0106] The braking distance prediction ability represents the deviation between the driver's gaze point and the theoretical braking stopping position when the car is driving in a straight line. The higher the prediction ability value, the better the driver's grasp of the vehicle speed and following distance, which also means that the driver's driving behavior is safer.

[0107] The characterization quantity of the braking distance estimation ability is X2, and its calculation process is as follows: the non-contact eye tracker collects the driver's eye movement information to determine the driver's gaze angle α, that is, the angle between the driver's line of sight and the horizontal plane; β represents the angle between the driver's gaze direction and the longitudinal direction; at the same time, the vertical height h of the driver's eyes from the horizontal road surface in a relaxed state is measured, in meters; the longitudinal speed V (km / h) and longitudinal acceleration a (m / s) of the vehicle are obtained according to the vehicle CAN bus and the longitudinal acceleration sensor 2 );

[0108] The calculation formula of the driver's gaze point is as follows:

[0109]

[0110] Here, α, β∈[0, 90°); S * Indicates the longitudinal distance between the driver's gaze point and the vehicle, unit: m:

[0111] If S * If it is too small, it means that the driver does not have enough control over the braking distance of the car at different speeds;

[0112] The calculation formula of the theoretical braking distance of a car is:

[0113]

[0114] S represents the longitudinal distance between the theoretical parking position of the car and the self-vehicle, unit: (m); Δt is the driver's reaction time, here Δt = 1.2s; V0 represents the speed of the self-vehicle during braking (km / h);

[0115] Count the time T when the distance between the driver's gaze point and the vehicle is greater than the theoretical braking distance L ; The time T when the distance between the driver's gaze point and the vehicle is less than the theoretical braking distance S ; The calculation formula of the braking distance estimation capability characterization quantity is:

[0116]

[0117] X2 is dimensionless, X2∈(0,1], the larger X2 is, the greater T L The larger the proportion, the safer the driver's driving behavior.

[0118] Step 2.3, calculation of brake rationality:

[0119] The braking acceleration of the car's rapid deceleration process is set to A, and the deceleration duration is set to t2; when the car's deceleration process meets the following conditions, it can be determined as a braking operation;

[0120] A≤-3m / s 2

[0121] t2≥3s

[0122] The distance between the vehicle and surrounding vehicles is detected by millimeter-wave radar. The rationality of the vehicle's braking is judged by a high-definition camera installed under the front windshield of the vehicle and facing the front. X3 is the representation of the rationality of braking, and the calculation formula is as follows:

[0123]

[0124] Here, X3∈(0,1], dimensionless; B R is the reasonable number of braking times, unit (times); B A is the actual number of braking times, unit (times); the closer X3 is to 1, the closer the driver's reasonable braking times are to the total number of braking times, and the safer the driving behavior is; conversely, the closer X3 is to 0, the fewer reasonable braking times the driver has, and the worse the driving behavior safety is.

[0125] Step 2.4, calculation of longitudinal safety distance maintenance:

[0126] The longitudinal safety distance maintenance mainly reflects the driver's control of the distance to the longitudinal vehicle in front during driving. The driver can maintain the distance within a safe range, which fully reflects that the driver has rich driving experience and sufficient safety awareness.

[0127] The distance LC between the vehicle and surrounding vehicles is monitored by millimeter-wave radar. The minimum safe driving distance DR (unit: m) of the vehicle is stipulated as follows:

[0128] ①On the highway

[0129] When V>100km / h, D R =100m;

[0130] When 60km / h≤V<100km / h, D R It is numerically equal to the speed of the car;

[0131] ② Under urban conditions, the minimum safe driving distance is as follows:

[0132] When 40km / h≤V<60km / h, D R It is numerically equal to the speed of the car;

[0133] When 30km / h≤V<40km / h, D R =30m;

[0134] When the car is in the speed range of V<30m / s, it may be affected by external factors such as traffic lights, traffic volume, road conditions, etc., so they are not taken into consideration.

[0135] Here, the distance between the car and the front car is less than D R Duration t q and greater than D R Duration t w ; The safety distance maintenance characterization quantity is represented by X4, and its calculation formula is as follows:

[0136]

[0137] X4∈(0, 1], dimensionless; the closer X4 is to 1, the better the driver maintains the safe distance.

[0138] Step 2.5, turn to security:

[0139] Steering safety indicates whether the driver turns on the turn signal when changing lanes, turning, making a U-turn, etc. The common scenarios where turn signals are required are shown in Table 1 below;

[0140] Table 1 Driving scenarios that normally require the use of turn signals

[0141]

[0142] The high-definition camera in the car collects the driver's actions to determine whether the driver has turned on the turn signal; the lane line monitoring system determines whether the car has changed lanes; the number of times the driver turns on the turn signal is recorded as K1; the number of times the car changes lanes is K2;

[0143] The indicator of steering safety is X5, and the calculation formula of X5 is:

[0144]

[0145] X5 is dimensionless and does not consider the situation where the driver turns on the turn signal but fails to change lanes. It can be seen that K1≤K2, X5∈(0,1]; the closer X5 is to 1, the more standardized the driver's operation is and the higher the safety is.

[0146] Step 3: Calculate the weight of evaluation index;

[0147] The present invention adopts the entropy method to determine the weight of each evaluation index characterization quantity. The entropy method is an objective weighting method that uses the variation degree of the evaluation index to calculate the information entropy. It avoids the defects of the subjective assignment method to a certain extent and can effectively solve the problem of multi-index comprehensive evaluation. The specific calculation method is as follows:

[0148] The percentage of effective gaze area, braking distance estimation ability, braking rationality, longitudinal safety distance maintenance and steering safety of 30 drivers are used as evaluation indicators of the entropy method.

[0149] The calculation method of the evaluation index weight in step 3 is:

[0150] Step 3.1, determine the evaluation index attributes:

[0151] From the analysis and calculation of the characterization values ​​of the previous evaluation indicators, we can know the attributes of each evaluation, as shown in Table 2;

[0152] Table 2 Attribute analysis of evaluation indicators

[0153]

[0154] Step 3.2: Construct the original data matrix, standardize the data, and eliminate the dimension effect. The formula is as follows:

[0155]

[0156] Positive indicators:

[0157]

[0158] Reverse indicators:

[0159]

[0160] Among them, (i=1, 2, 3, ..., 30), (j=1, 2, 3, 4, 5);

[0161] Z ij is the standardized value, x i，j is the value of the characterization quantity of the jth evaluation index of the i-th driver, They represent the maximum and minimum values ​​of the j-th index respectively. The positive and negative indexes are processed according to formulas (9) and (10) respectively.

[0162] Step 3.3, calculate the weight of the i-th driver evaluation index representation quantity under the index:

[0163]

[0164] Thus, the weight matrix Y = {y ij} 30*5 ;

[0165] Step 3.4, calculate the information entropy value e and information utility value d of the indicator:

[0166]

[0167] d j =1-e j (13)

[0168] Step 3.5, calculate the weight of the indicator:

[0169]

[0170] Step 4: Construction of driving behavior safety evaluation model.

[0171] The method for constructing the driving behavior safety evaluation model in step 4 is:

[0172] According to the weights of the evaluation index representation quantities calculated previously, the calculation formula of the driving behavior safety evaluation model can be obtained as follows:

[0173]

[0174] Here, ω1~ω5 are weights of each safety evaluation index representation quantity, ω1~ω5∈(0,1); Z1~Z5 are dimensionless normalized values ​​of each evaluation index representation quantity of the driver;

[0175] The driving behavior safety evaluation score P Δ Expressed in percent:

[0176]

[0177] here It is the maximum value of the standardized value of each evaluation index;

[0178] The final driving behavior safety evaluation model can be obtained:

[0179]

[0180] The process of driving behavior safety evaluation is as follows:

[0181] The driver needs to drive through the test road. During the collection process, the driver's eye movement information, vehicle operation information, radar information and lane line information are collected. According to formulas (1)-(7), the numerical representation values ​​of the driver's driving behavior safety evaluation index X'1~X'5 are calculated. Then X'1~X'5 are standardized:

[0182]

[0183] Finally, the driver's driving behavior safety score is obtained:

[0184]

[0185] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0186] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected.

Claims

1. A method for constructing a driving behavior safety evaluation model based on real vehicle testing, characterized in that: The method comprises the following steps: Step 1: Selection of experimental subjects and test sections; Step 2: Real vehicle data collection and evaluation index calculation; Step 2.1, calculation of effective fixation area percentage: The areas where the driver's eyes are focused during driving are divided into three categories: the front area: refers to the area that the driver can see through the front windshield of the car; the rearview mirror area: refers to the area that the driver can see through the rearview mirror, left and right rearview mirrors, and the windows on both sides near them; other areas: other areas excluding the front area and the rearview mirror area; The basis for judging the driver's gaze area is as follows: Gaze angle deviation threshold: 60°; Gaze point movement speed: V eye ≤15° / s; Fixation duration: t eye ≥0.2s; When the driver's eye movement data meets the above conditions, it is considered that the driver's eyes are looking at this area; The effective gaze area percentage is represented by X1, which refers to the time T that the driver's eyes focus on the front area and the rearview mirror area during driving. eye The ratio of X1 to the total driving time T is calculated as follows: X1 is dimensionless; and 0<T eye ≤T, we know that X1∈(0,1]; Step 2.2, calculation of braking distance estimation capability: The characterization quantity of the braking distance estimation ability is X2, and its calculation process is as follows: the driver's eye movement information is collected by a non-contact eye tracker to determine the driver's gaze angle α, that is, the angle between the driver's line of sight and the horizontal plane; β represents the angle between the driver's gaze direction and the longitudinal direction; at the same time, the vertical height h of the driver's eyes from the horizontal road surface in a relaxed state is measured, unit: m; the longitudinal speed V and longitudinal acceleration a of the vehicle are obtained according to the vehicle CAN bus and the longitudinal acceleration sensor, where the unit of the longitudinal speed V of the vehicle is: km / h, and the unit of the longitudinal acceleration a is: m / s 2 ; The calculation formula of the driver's gaze point is as follows: Here, α, β∈[0,90°); S * Indicates the longitudinal distance between the driver's gaze point and the vehicle, unit: m; If S * If it is too small, it means that the driver does not have enough control over the braking distance of the car at different speeds; The calculation formula of the theoretical braking distance of a car is: S represents the longitudinal distance between the theoretical parking position of the car and the self-vehicle, unit: m; Δt is the driver's reaction time, here Δt = 12.s; V0 represents the speed of the self-vehicle during braking, unit: km / h; Count the time T when the distance between the driver's gaze point and the vehicle is greater than the theoretical braking distance L ; The time T when the distance between the driver's gaze point and the vehicle is less than the theoretical braking distance S ; The calculation formula of the braking distance estimation capability characterization quantity is: X2 is dimensionless, X2∈(0,1], the larger X2 is, the greater T L The larger the proportion, the safer the driver's driving behavior; Step 2.3, calculation of brake rationality: The braking acceleration of the car's rapid deceleration process is set to A, and the deceleration duration is set to t2; when the car's deceleration process meets the following conditions, it is determined to be a braking operation; <h2 style=";text-align:left;direction:ltr">A≤-3m / s<h2 style=";text-align:left;direction:ltr"> 2 t2≥3s The distance between the vehicle and surrounding vehicles is detected by millimeter-wave radar. The rationality of the vehicle's braking is judged by a high-definition camera installed under the front windshield of the vehicle and facing the front. X3 is the representation of the rationality of braking, and the calculation formula is as follows: Here, X3∈(0,1], dimensionless; B R B is the reasonable number of braking times, unit: times; A is the actual number of braking times, unit: times; the closer X3 is to 1, the closer the driver's reasonable braking times are to the total number of braking times, and the safer the driving behavior is; conversely, the closer X3 is to 0, the fewer reasonable braking times the driver has, and the worse the driving behavior safety is; Step 2.4, longitudinal safety distance maintenance calculation: The millimeter-wave radar monitors the distance LC between the vehicle and surrounding vehicles, and the distance between the vehicle and the vehicle in front is less than D R Duration t q and greater than D R Duration t w ; The safety distance maintenance characterization quantity is represented by X4, and its calculation formula is as follows: X4∈(0,1], dimensionless; the closer X4 is to 1, the better the driver maintains the safe distance; Step 2.5, turn to security: The high-definition camera in the car collects the driver's actions to determine whether the driver has turned on the turn signal; the lane line monitoring system determines whether the car has changed lanes; the number of times the driver turns on the turn signal is recorded as K1; the number of times the car changes lanes is K2; The indicator of steering safety is X5, and the calculation formula of X5 is: X5 is dimensionless, and does not consider the situation where the driver turns on the turn signal but fails to change lanes. It can be seen that K1≤K2, X5∈(0,1]; the closer X5 is to 1, the more standardized the driver's operation is and the higher the safety is. Step 3: Calculate the weight of evaluation index; Step 3.1, determine the evaluation index attributes: The properties of each evaluation can be known from the analysis and calculation of the representative values ​​of the previous evaluation indicators; Step 3.2: Construct the original data matrix, standardize the data, and eliminate the dimension effect. The formula is as follows: Positive indicators: Reverse indicators: Where, i = 1, 2, 3, ..., 30, j = 1, 2, 3, 4, 5; Z ij is the standardized value, x i，j is the value of the characterization quantity of the jth evaluation index of the i-th driver, They represent the maximum and minimum values ​​of the j-th index respectively. The positive and negative indexes are processed according to formulas (9) and (10) respectively. Step 3.3, calculate the weight of the i-th driver evaluation index representation quantity under the index: Thus, the weight matrix Y = {y ij } 30*5 ; Step 3.4, calculate the information entropy value e and information utility value d of the indicator: d j =1-e j (13) Step 3.5, calculate the weight of the indicator: Step 4: Construction of driving behavior safety evaluation model; According to the weights of the evaluation index representation quantities calculated previously, the calculation formula of the driving behavior safety evaluation model is as follows: Here, ω1~ω5 are weights of each safety evaluation index representation quantity, ω1~ω5∈(0,1); Z1~Z5 are dimensionless normalized values ​​of each evaluation index representation quantity of the driver; The driving behavior safety evaluation score P Δ Expressed in percent: here It is the maximum value of the standardized value of each evaluation index; The final driving behavior safety evaluation model is obtained:

2. The method for constructing a driving behavior safety evaluation model based on real vehicle testing according to claim 1, characterized in that: The process of driving behavior safety evaluation is as follows: The driver needs to drive through the test road. During the collection process, the driver's eye movement information, vehicle operation information, radar information and lane line information are collected. According to formulas (1)-(7), the numerical representation values ​​of the driver's driving behavior safety evaluation index X′1~X′5 are calculated; then X′1~X′5 are standardized: Finally, the driver's driving behavior safety score is obtained:

Images