Traffic accident risk evaluation method based on urban checkpoint license plate data

Abstract

The invention discloses a traffic accident risk evaluation method based on urban checkpoint license plate data, belongs to the field of traffic safety, and particularly relates to the traffic accident risk evaluation method based on the urban checkpoint license plate data. The objective of the invention is to solve the problems that the data acquisition cost is high, the risk analysis is hysteretic, and the license plate big data information cannot be fully mined in the traditional traffic accident risk analysis. The method comprises the following steps: 1, obtaining traffic gate license plate data, and obtaining processed traffic gate license plate data; 2, constructing a vehicle track database; 3, determining longitude and latitude coordinates of a bayonet, and converting the track database into a track coordinate database; 4, constructing a risk analysis initial matrix, and performing normalization processing on risk evaluation index values in the matrix to obtain processed risk evaluation index values; calculating an evaluation index weight value by adopting an entropy weight method; and based on the risk evaluation index value and the evaluation index weight value, evaluating the grade of the road section by adopting a rank sum ratio comprehensive evaluation method.

Description

Technical Field

[0001] The present invention belongs to the field of traffic safety, and particularly relates to a traffic accident risk assessment method based on license plate data of urban checkpoints. Background Art

[0002] With the rapid development of urban informatization construction and intelligent transportation systems, traffic monitoring systems have become increasingly mature. Urban traffic checkpoint data collected based on license plate recognition technology can provide sample data that tends to be full-scale. Improving the utilization rate of license plate data, deeply mining the spatio-temporal information of vehicle trips, and conducting accident risk analysis on different sections can provide a scientific basis for urban traffic safety decision-making.

[0003] Traffic accident risk analysis is an important part of the field of traffic safety. At present, most traffic accident analysis methods involve selecting influencing factors based on historical accident data and constructing prediction models. However, the randomness and probability of traffic accidents make the acquisition of accident data require a long period, and the specific road conditions vary greatly, which greatly reduces the timeliness and availability of accident analysis, making traffic accident prevention lag behind.

[0004] Obtaining road traffic data based on traffic detection technology and unmanned aerial vehicle video acquisition technology requires separately obtaining data for each road to be analyzed, which has disadvantages such as high investment and long time consumption, so the applicability is low. The present invention conducts accident risk analysis based on urban checkpoint license plate data. The data acquisition cost is low, and the samples tend to be full-scale data, which can be applied to the main roads between all intersections in the city. The method proposed by the present invention is an alternative safety analysis technology that only analyzes based on road traffic flow characteristics and does not rely on historical accident data, so the timeliness is good. In summary, the traffic accident risk analysis method based on urban checkpoint license plate data proposed by the present invention aims to overcome the limitations of the prior art and provide a low-cost and applicable traffic safety solution to ensure urban road traffic safety. Summary of the Invention

[0005] The purpose of the present invention is to solve the problems of high data acquisition cost, lag in risk analysis, and failure to fully mine license plate big data information in traditional traffic accident risk analysis, and then proposes a traffic accident risk assessment method based on urban checkpoint license plate data.

[0006] The specific process of a traffic accident risk assessment method based on urban checkpoint license plate data is as follows:

[0007] Step 1: Obtain traffic checkpoint license plate data, process the traffic checkpoint license plate data, and obtain the processed traffic checkpoint license plate data;

[0008] Step 2: Construct a vehicle trajectory database based on the processed traffic checkpoint license plate data obtained in Step 1;

[0009] Step 3: Determine the longitude and latitude coordinates of the card ports, and convert the trajectory database into a trajectory coordinate database;

[0010] Step 4: Construct an initial matrix for risk analysis, normalize the risk evaluation index values in the initial matrix for risk analysis to obtain the normalized risk evaluation index values; Based on the normalized risk evaluation index values, use the entropy weight method to calculate the evaluation index weight values; Based on the normalized risk evaluation index values and the evaluation index weight values, use the rank sum ratio comprehensive evaluation method to evaluate the road section level.

[0011] The beneficial effects of the present invention are as follows:

[0012] The present invention first processes the traffic card port data to construct a vehicle trajectory coordinate database, then obtains the traffic flow characteristic indicators of the road sections to be analyzed during peak hours, and further constructs a risk evaluation model using the RSR-entropy weight method to obtain the accident risk assessment index of each road section.

[0013] The present invention discloses a traffic accident risk analysis method based on urban card port license plate data, which relates to a method for obtaining vehicle trajectories in an urban road network and a substitution evaluation method for traffic accident risks. The purpose of the present invention is to solve the problems that the big data information of license plates in the urban road network has not been fully mined, traffic accidents are random and probabilistic, and the traditional traffic accident risk prediction model has lag. The process is as follows: 1. Process the card port license plate data; 2. Construct a vehicle trajectory database; 3. Determine the longitude and latitude coordinates of the card ports, and convert the trajectory database into a trajectory coordinate database; 4. Establish an accident risk evaluation system, select representative risk evaluation indicators, and construct a substitution evaluation model for traffic accident risks based on the RSR-entropy weight method.

[0014] Based on large-scale traffic card port data, aiming at the problems of lag in traffic accident risk analysis and high cost of obtaining road data, the present invention proposes a traffic accident risk analysis method based on urban card port license plate data. After collecting and processing the card port license plate data, use MYSQL to construct a vehicle trajectory coordinate database, obtain traffic flow characteristic indicators from it, construct a substitution evaluation model for accident risks based on the RSR-entropy weight method, and give early warnings to potential accident road sections in a timely manner, which is of great significance for ensuring the safety and smoothness of urban road traffic. Description of the Drawings

[0015] Figure 1 It is a flow chart of the present invention;

[0016] Figure 2 It is a fitting effect diagram of the RSR estimated value for the accident risk evaluation of six random road sections in the present invention. Detailed Embodiment

[0017] Specific Embodiment 1: The specific process of this embodiment is as follows:

[0018] Step 1: Obtain the license plate data of traffic checkpoints, process the license plate data of traffic checkpoints, and obtain the processed license plate data of traffic checkpoints;

[0019] Step 2: Build a vehicle trajectory database based on the processed license plate data of traffic checkpoints obtained in Step 1;

[0020] Step 3: Determine the longitude and latitude coordinates of the checkpoints, and convert the trajectory database into a trajectory coordinate database;

[0021] Step 4: Build an initial risk analysis matrix, normalize the risk evaluation index values in the initial risk analysis matrix to obtain the normalized risk evaluation index values; based on the normalized risk evaluation index values, calculate the evaluation index weight values using the entropy weight method; based on the normalized risk evaluation index values and the evaluation index weight values, evaluate the road section level using the rank sum ratio comprehensive evaluation method.

[0022] Specific Embodiment 2: The difference between this embodiment and Specific Embodiment 1 is that in Step 1, the license plate data of traffic checkpoints is obtained, the license plate data of traffic checkpoints is processed, and the processed license plate data of traffic checkpoints is obtained;

[0023] The specific process is as follows:

[0024] Step 11: Obtain the license plate data of traffic checkpoints through the electronic police equipment at urban intersections, record the license plate data of traffic checkpoints by row (1 row for the license plate data of 1 traffic checkpoint), screen the license plate data of traffic checkpoints, screen out the missing data and error data, and directly delete the missing and error data to obtain the screened license plate data of traffic checkpoints;

[0025] The license plate data of traffic checkpoints includes the following attribute fields: license plate number, license plate type, capture time, affiliated organization, capture location, driving direction, vehicle type;

[0026] Step 12: Group the screened license plate data of traffic checkpoints according to the license plate number, with the same license plate number in one group, and sort each group of license plate data in ascending order according to the capture time;

[0027] If the license plate numbers of adjacent A (several) pieces of data in a group are the same, the intersection positions corresponding to the license plate numbers of adjacent A (several) pieces of data are the same, and the capture time interval of adjacent A (several) pieces of data is less than the set threshold, that is, all of (1)-(3) are satisfied, then it is determined as redundant data, and only the first valid piece of data is retained for the A (several) pieces of redundant data;

[0028] ID i = ID i+1(1)

[0029] Sid i = Sid i+1 (2)

[0030] t i+1 -t i <λ (3)

[0031] Wherein,

[0032] ID i is the license plate number of the data in the i-th row, and ID i+1 is the license plate number of the data in the (i + 1)-th row;

[0033] Sid i is the intersection number where the data in the i-th row is located, and Sid i+1 is the intersection number where the data in the (i + 1)-th row is located;

[0034] t i is the capture time of the data in the i-th row, and t i+1 is the capture time of the data in the (i + 1)-th row;

[0035] λ is the time interval threshold;

[0036] A is a positive integer;

[0037] Step One Three:

[0038] Number the intersections in the data obtained in Step One Two, and convert the time format. Uniformly take 0:00:00 in the early morning of the same day as the time starting point, and convert the time into a format in seconds;

[0039] Discard the data of the date part, that is, delete the "year / month / day" part. Take 0:00:00 in the early morning of the same day as the time starting point, and convert the time field:

[0040] T = a×3600 + b×3600 + c (4)

[0041] Wherein: T is the converted time; a is the hour digit in the original time; b is the minute digit; c is the second digit.

[0042] Other steps and parameters are the same as those in the specific implementation method one.

[0043] Specific implementation method three: The difference between this implementation method and the specific implementation method one or two is that in step two, a vehicle trajectory database is constructed based on the processed traffic checkpoint license plate data obtained in step one; the specific process is as follows:

[0044] Step Two One: Group the processed traffic checkpoint license plate data obtained in step one according to the license plate number;

[0045] Step Two: Arrange each group of license plate numbers in ascending order of time;

[0046] Step Two: Concatenate the locations of each group of license plate numbers at each traffic checkpoint and separate them with commas (location (13, 12, 1...));

[0047] Concatenate the time field values of each group of license plate numbers at each traffic checkpoint and separate them with commas (time (0, 1, 12...));

[0048] Step Two: Save the result obtained in Step Two as a new trajectory table to form a vehicle trajectory database.

[0049] Other steps and parameters are the same as those in the first or second specific implementation manner.

[0050] Specific implementation manner four: The difference between this implementation manner and one of the first to third specific implementation manners is that in step four, a risk analysis initial matrix is constructed, and the risk evaluation index values in the risk analysis initial matrix are normalized to obtain the normalized risk evaluation index values; based on the normalized risk evaluation index values, the entropy weight method is used to calculate the evaluation index weight values; based on the normalized risk evaluation index values and the evaluation index weight values, the rank sum ratio comprehensive evaluation method is used to evaluate the road section level;

[0051] The specific process is as follows:

[0052] Step Four: Construct a risk analysis initial matrix, and normalize the risk evaluation index values in the risk analysis initial matrix to obtain the normalized risk evaluation index values;

[0053] Step Four: Based on the normalized risk evaluation index values, use the entropy weight method to calculate the evaluation index weight values;

[0054] Step Four: Based on the normalized risk evaluation index values and the evaluation index weight values, use the rank sum ratio comprehensive evaluation method to evaluate the road section level.

[0055] Other steps and parameters are the same as those in one of the first to third specific implementation manners.

[0056] Specific implementation manner five: The difference between this implementation manner and one of the first to fourth specific implementation manners is that in step four, a risk analysis initial matrix is constructed, and the risk evaluation index values in the risk analysis initial matrix are normalized to obtain the normalized risk evaluation index values; the specific process is as follows:

[0057] Step Four: Determine the road section between adjacent traffic checkpoints as the risk evaluation road section OD;

[0058] Screen out the vehicle data of all OD pairs passing through the risk assessment sections;

[0059] Set a selected time period, filter the time field of the vehicle data of all OD pairs passing through the risk assessment sections that have been screened out, and obtain the vehicle data passing through the risk assessment sections within the selected time period;

[0060] Step 4.1.2: Group the vehicle data obtained in Step 4.1.1 according to the "vehicle type" field and divide it into three vehicle types: cars, buses, and trucks;

[0061] Step 4.1.3: Statistically analyze the vehicle data obtained in Step 4.1.2, and calculate the traffic flow index Q of the i-th type of vehicle according to (5); i ;

[0062]

[0063] In the formula: Q i is the traffic flow index of the i-th type of vehicle, where i = 1, 2, 3;

[0064] Q1 is the traffic flow of the 1st type of vehicle on the section, Q2 is the traffic flow of the 2nd type of vehicle on the section, and Q3 is the traffic flow of the 3rd type of vehicle on the section;

[0065] The 1st type of vehicle is a car; the 2nd type of vehicle is a bus; the 3rd type of vehicle is a truck;

[0066] N i is the total number of vehicles of the i-th type passing through the OD of the risk assessment section, where i = 1, 2, 3;

[0067] N1 is the total number of vehicles of the 1st type passing through the OD of the risk assessment section, N2 is the total number of vehicles of the 2nd type passing through the OD of the risk assessment section, and N3 is the total number of vehicles of the 3rd type passing through the OD of the risk assessment section;

[0068] T is the duration within the selected time period;

[0069] Step 4.1.4: Based on the vehicle data obtained in Step 4.1.2, calculate the difference between the arrival and departure times of each vehicle in the OD of the risk assessment section to obtain the running time of each vehicle;

[0070] Based on the running time of each vehicle, calculate according to (6) to obtain the average speed index V of the i-th type of vehicle; i , where i = 1, 2, 3;

[0071]

[0072] In the formula: V i is the average speed of the i-th type of vehicle;

[0073] V1 is the average vehicle speed of the first type of vehicle; V2 is the average vehicle speed of the second type of vehicle; V3 is the average vehicle speed of the third type of vehicle;

[0074] S is the total length of the OD of the risk assessment section;

[0075] t n is the running time of the t n th vehicle in the i-th type of vehicle, i = 1, 2, 3;

[0076] Step Four-Fifteen.

[0077] Based on the traffic flow Q1 of the first type of vehicle on the section, the traffic flow Q2 of the second type of vehicle on the section, the traffic flow Q3 of the third type of vehicle on the section, the average vehicle speed V1 of the first type of vehicle, the average vehicle speed V2 of the second type of vehicle, and the average vehicle speed V3 of the third type of vehicle, construct an initial risk analysis matrix;

[0078] The initial risk analysis matrix is defined as follows:

[0079]

[0080] In the formula: M is the model input matrix; m is the number of evaluation sections; x st is the value of the t-th risk evaluation index of the s-th section to be evaluated; s = 1, 2,..., m;

[0081] x s1 is the value of the first risk evaluation index of the s-th section to be evaluated, and the first risk evaluation index value is the traffic flow Q1 of the first type of vehicle on the section, s = 1, 2,..., m;

[0082] x s2 is the value of the second risk evaluation index of the s-th section to be evaluated, and the second risk evaluation index value is the traffic flow Q2 of the second type of vehicle on the section, s = 1, 2,..., m;

[0083] x s3 is the value of the third risk evaluation index of the s-th section to be evaluated, and the third risk evaluation index value is the traffic flow Q3 of the third type of vehicle on the section, s = 1, 2,..., m;

[0084] x s4 is the value of the fourth risk evaluation index of the s-th section to be evaluated, and the fourth risk evaluation index value is the average vehicle speed V1 of the first type of vehicle, s = 1, 2,..., m;

[0085] x s5 is the value of the fifth risk evaluation index of the s-th section to be evaluated, and the fifth risk evaluation index value is the average vehicle speed V2 of the second type of vehicle, s = 1, 2,..., m;

[0086] x s6is the value of the 6th risk assessment index for the sth road section to be evaluated. The value of the 6th risk assessment index is the average speed V3 of the 3rd type of vehicle, where s = 1, 2, …, m;

[0087] Step 416. For the value x of the tth risk assessment index of the sth road section to be evaluated in the initial matrix of risk analysis in Step 415 st perform normalization processing to obtain the value y of the tth risk assessment index of the sth road section to be evaluated after normalization processing st , where t = 1, 2, …, 6; It is expressed as:

[0088]

[0089]

[0090] In the formula:

[0091] y s1 is the value of the 1st risk assessment index of the sth road section to be evaluated after normalization processing, where s = 1, 2, …, m;

[0092] y s2 is the value of the 2nd risk assessment index of the sth road section to be evaluated after normalization processing, where s = 1, 2, …, m;

[0093] y s3 is the value of the 3rd risk assessment index of the sth road section to be evaluated after normalization processing, where s = 1, 2, …, m;

[0094] y s4 is the value of the 4th risk assessment index of the sth road section to be evaluated after normalization processing, where s = 1, 2, …, m;

[0095] y s5 is the value of the 5th risk assessment index of the sth road section to be evaluated after normalization processing, where s = 1, 2, …, m;

[0096] y s6 is the value of the 6th risk assessment index of the sth road section to be evaluated after normalization processing, where s = 1, 2, …, m;

[0097] max(x s1 ) is the maximum value of the 1st risk assessment index of the sth road section to be evaluated, where s = 1, 2, …, m, max(x s1 ) = max(x 11 , x 21 , …, x m1 );

[0098] max(x s2 ) is the maximum value of the 2nd risk assessment index of the sth road section to be evaluated, where s = 1, 2, …, m, max(xs2 ) = max(x 12 , x 22 , …, x m2 );

[0099] max(x s3 ) is the maximum value of the 3rd risk assessment indicator for the sth road section to be evaluated, s = 1, 2, …, m, max(x s3 ) = max(x 13 , x 23 , …, x m3 );

[0100] max(x s4 ) is the maximum value of the 4th risk assessment indicator for the sth road section to be evaluated, s = 1, 2, …, m, max(x s4 ) = max(x 14 , x 24 , …, x m4 );

[0101] max(x s5 ) is the maximum value of the 5th risk assessment indicator for the sth road section to be evaluated, s = 1, 2, …, m, max(x s5 ) = max(x 15 , x 25 , …, x m5 );

[0102] max(x s6 ) is the maximum value of the 6th risk assessment indicator for the sth road section to be evaluated, s = 1, 2, …, m, max(x s6 ) = max(x 16 , x 26 , …, x m6 );

[0103] min(x s1 ) is the minimum value of the 1st risk assessment indicator for the sth road section to be evaluated, s = 1, 2, …, m, min(x s1 ) = min(x 11 , x 21 , …, x m1 );

[0104] min(x s2 ) is the minimum value of the 2nd risk assessment indicator for the sth road section to be evaluated, s = 1, 2, …, m, min(x s2 ) = min(x 12 , x 22 , …, x m2 );

[0105] min(xs3 ) is the minimum value of the 3rd risk evaluation index for the sth road section to be evaluated, s = 1, 2, …, m, min(x s3 ) = min(x 13 , x 23 , …, x m3 );

[0106] min(x s4 ) is the minimum value of the 4th risk evaluation index for the sth road section to be evaluated, s = 1, 2, …, m, min(x s4 ) = min(x 14 , x 24 , …, x m4 );

[0107] min(x s5 ) is the minimum value of the 5th risk evaluation index for the sth road section to be evaluated, s = 1, 2, …, m, min(x s5 ) = min(x 15 , x 25 , …, x m5 );

[0108] min(x s6 ) is the minimum value of the 6th risk evaluation index for the sth road section to be evaluated, s = 1, 2, …, m, min(x s6 ) = min(x 16 , x 26 , …, x m6 ).

[0109] Other steps and parameters are the same as those in any one of the specific embodiments one to four.

[0110] Specific embodiment six: The difference between this embodiment and any one of the specific embodiments one to five is that based on the risk evaluation index values after normalization processing in step four two, the entropy weight method is used to calculate the evaluation index weight values; the specific process is as follows:

[0111] Step four two one,

[0112] According to the value y st of the tth risk evaluation index of the sth road section to be evaluated after normalization processing, calculate the proportion p st of the tth risk evaluation index of the sth road section to be evaluated after normalization processing, t = 1, 2, …, 6;

[0113] It is expressed as:

[0114]

[0115]

[0116] In the formula:

[0117] p s1 is the proportion of the first risk evaluation index of the sth road section to be evaluated after normalization, s = 1, 2, …, m;

[0118] p s2 is the proportion of the second risk evaluation index of the sth road section to be evaluated after normalization, s = 1, 2, …, m;

[0119] p s3 is the proportion of the third risk evaluation index of the sth road section to be evaluated after normalization, s = 1, 2, …, m;

[0120] p s4 is the proportion of the fourth risk evaluation index of the sth road section to be evaluated after normalization, s = 1, 2, …, m;

[0121] p s5 is the proportion of the fifth risk evaluation index of the sth road section to be evaluated after normalization, s = 1, 2, …, m;

[0122] p s6 is the proportion of the sixth risk evaluation index of the sth road section to be evaluated after normalization, s = 1, 2, …, m;

[0123] Step 4.2.2: Based on the proportion p st of the tth risk evaluation index of the sth road section to be evaluated after normalization, calculate the entropy value e t of the tth risk evaluation index of the sth road section to be evaluated after normalization, where t = 1, 2, 3, 4, 5, 6;

[0124] It is expressed as:

[0125]

[0126]

[0127] In the formula:

[0128] e1 is the entropy value of the first risk evaluation index of the sth road section to be evaluated after normalization, s = 1, 2, …, m;

[0129] e2 is the entropy value of the second risk evaluation index of the sth road section to be evaluated after normalization, s = 1, 2, …, m;

[0130] e3 is the entropy value of the third risk evaluation index of the sth road section to be evaluated after normalization, s = 1, 2, …, m;

[0131] $e_4$ is the entropy value of the 4th risk assessment index of the $s$-th road section to be evaluated after normalization, where $s = 1, 2, \ldots, m$;

[0132] $e_5$ is the entropy value of the 5th risk assessment index of the $s$-th road section to be evaluated after normalization, where $s = 1, 2, \ldots, m$;

[0133] $e_6$ is the entropy value of the 6th risk assessment index of the $s$-th road section to be evaluated after normalization, where $s = 1, 2, \ldots, m$;

[0134] $m$ is the number of road sections to be evaluated;

[0135] Step Four Two Three: According to the entropy value $e$ obtained in Step Four Two Two t , calculate the weight value $w$ of the $t$-th risk assessment index t , where $t = 1, 2, 3, 4, 5, 6$; expressed as:

[0136]

[0137] In the formula: $w_1$ is the weight value of the 1st risk assessment index; $w_2$ is the weight value of the 2nd risk assessment index; $w_3$ is the weight value of the 3rd risk assessment index; $w_4$ is the weight value of the 4th risk assessment index; $w_5$ is the weight value of the 5th risk assessment index; $w_6$ is the weight value of the 6th risk assessment index;

[0138] $n$ is the number of risk assessment indexes.

[0139] Other steps and parameters are the same as those in any one of the specific implementation manners One to Five.

[0140] Specific Implementation Manner Seven: The difference between this implementation manner and any one of the specific implementation manners One to Six is that in Step Four Three, based on the risk assessment index values and evaluation index weight values after normalization, the rank sum ratio comprehensive evaluation method is used to evaluate the road section grade; the specific process is as follows:

[0141] Step Four Three One: Use the non-integer rank sum ratio method to rank the initial risk analysis matrix:

[0142]

[0143] In the formula:

[0144] $R$ s1 is the rank of the element in the $s$-th row and the 1st column, $R$ s2 is the rank of the element in the $s$-th row and the 2nd column, $R$ s3 is the rank of the element in the $s$-th row and the 3rd column, $R$ s4 is the rank of the element in the $s$-th row and the 4th column, $R$ s5 is the rank of the element in the $s$-th row and the 5th column, $R$ s6is the rank of the element in the 6th column of the s-th row, where s = 1, 2, …, m;

[0145] Step Four Thirty-Two: Based on the weight value w t of the t-th risk assessment indicator and the rank R st of the element in the s-th row and t-th column, calculate the weighted rank sum ratio WRSR s of the s-th row; expressed as:

[0146]

[0147] where: w t is the weight value of the t-th risk assessment indicator; R st is the rank of the element in the s-th row and t-th column;

[0148] Step Four Thirty-Three: Sort the weighted rank sum ratio WRSR s values from smallest to largest to obtain the sorted WRSR1, WRSR2, …, WRSR n ;

[0149] Compile the frequency f of the weighted rank sum ratio of each row after sorting, and calculate the cumulative frequency ∑f of the weighted rank sum ratio of each row; determine the rank R and the average rank of the weighted rank sum ratio of each row

[0150] Calculate the downward cumulative frequency of the weighted rank sum ratio of each row

[0151] where m is the number of evaluation sections;

[0152] Convert the downward cumulative frequency p of the weighted rank sum ratio of each row into the Probit value p r of the probability unit;

[0153] Step Four Thirty-Four: Using the probability unit value p r corresponding to the cumulative frequency as the independent variable and the corresponding weighted rank sum ratio WRSR s value as the dependent variable, calculate the regression equation to obtain the parameters a and b to be estimated:

[0154] WRSR s = a + bP r , s = 1, 2, …, m

[0155] where:

[0156] p r is the probability unit value corresponding to the cumulative frequency, and a and b are the parameters to be estimated;

[0157] Step Four Thirty-Five:

[0158] Classify the accident risks of the sections to be evaluated into three levels: high-risk sections, medium-risk sections, and low-risk sections;

[0159] The percentile critical value L of the low-risk section is L < 15.886, and the probit value p r < 4;

[0160] The percentile critical value of the medium-risk section is 15.886 ≤ L < 84.134, and the probit value 4 ≤ p r < 6;

[0161] The percentile critical value L of the high-risk section is L ≥ 84.134, and the probit value p r ≥ 6;

[0162] Substitute the probit value p r = 4 into the regression equation WRSR for determining the parameters a and b to be estimated s = a + bP r , and obtain the estimated value A of the weighted rank sum ratio WRSR s ;

[0163] Substitute the probit value p r = 6 into the regression equation WRSR for determining the parameters a and b to be estimated s = a + bP r , and obtain the estimated value B of the weighted rank sum ratio WRSR s ;

[0164] The WRSR critical value A' of the low-risk section is A' < A;

[0165] The WRSR critical value of the medium-risk section is A ≤ A' < B

[0166] The WRSR critical value of the high-risk section is B ≤ A';

[0167] Step Four Three Six

[0168] Substitute the probit value p obtained in Step Four Three Four r into the regression equation for determining the parameters a and b to be estimated, and obtain n estimated values of WRSR;

[0169] Judge the range of the WRSR critical value to which the estimated value of each WRSR s belongs, and determine whether each WRSR s is a low-risk section, a medium-risk section or a high-risk section.

[0170] Other steps and parameters are the same as those in any one of Embodiments 1 to 6.

[0171] Embodiment 8: The difference between this embodiment and any one of Embodiments 1 to 7 is that in Step Four Three Three, for the weighted rank sum ratio WRSR sSort the values from smallest to largest to obtain the sorted WRSR1, WRSR2, …, WRSR n ;

[0172] Compile the frequency f of the weighted rank sum ratio for each row after sorting, and calculate the cumulative frequency ∑f of the weighted rank sum ratio for each row; determine the rank R and average rank of the weighted rank sum ratio for each row

[0173] Calculate the downward cumulative frequency of the weighted rank sum ratio for each row

[0174] where m is the number of evaluation sections;

[0175] Convert the downward cumulative frequency p of the weighted rank sum ratio for each row to the Probit value p of the probability unit r ;

[0176] The specific process is as follows:

[0177] Step 4.3.1. Sort the weighted rank sum ratio WRSR s Sort the values from smallest to largest to obtain the sorted WRSR1, WRSR2, …, WRSR n ;

[0178] Compile the frequency f of the weighted rank sum ratio for each row after sorting, and calculate the cumulative frequency ∑f for each row of the weighted rank sum ratio;

[0179] The specific process is as follows:

[0180] Sort the weighted rank sum ratio values from smallest to largest. The same weighted rank sum ratio values are in the same row, and different weighted rank sum ratio values are in different rows, for a total of n rows; n ≤ m

[0181] The frequency f1 of the weighted rank sum ratio in the first row after sorting is the number of times the weighted rank sum ratio appears in the first row;

[0182] The frequency f2 of the weighted rank sum ratio in the second row after sorting is the number of times the weighted rank sum ratio appears in the second row;

[0183] Until

[0184] The frequency f of the weighted rank sum ratio in the nth row after sorting n is the number of times the weighted rank sum ratio appears in the nth row;

[0185] The frequency is the number of times the WRSR s value appears. In the embodiment, WRSR s Because the number of sections is small, WRSR s Each section is different, but the number of sections is large. The first WRSR sIt is 0.28. If another RSR value calculated for other sections later is also 0.28, then for the row with 0.28 in the RSR distribution table, its frequency is 2.

[0186] The cumulative frequency ∑f1 of the weighted rank sum ratio in the first row is the number of times the weighted rank sum ratio appears in the first row;

[0187] The cumulative frequency ∑f2 of the weighted rank sum ratio in the second row is the sum from the number of times the weighted rank sum ratio appears in the first row to the number of times the weighted rank sum ratio appears in the second row;

[0188] Until

[0189] The cumulative frequency ∑f of the weighted rank sum ratio in the nth row n is the sum from the number of times the weighted rank sum ratio appears in the first row to the number of times the weighted rank sum ratio appears in the nth row;

[0190] Step Four Three Two: Determine the rank R and average rank of the weighted rank sum ratio for each row

[0191] The specific process is as follows:

[0192] The rank R of the weighted rank sum ratio in the first row is all the integers from smallest to largest in the cumulative frequency ∑f1 of the weighted rank sum ratio in the first row;

[0193] The rank R of the weighted rank sum ratio in the second row is all the integers from smallest to largest from the cumulative frequency ∑f1 + 1 of the weighted rank sum ratio in the first row to the cumulative frequency ∑f2 of the weighted rank sum ratio in the second row;

[0194] Until

[0195] The rank R of the weighted rank sum ratio in the nth row is the cumulative frequency Σf of the weighted rank sum ratio in the (n - 1)th row n-1 + 1 to the cumulative frequency Σf of the weighted rank sum ratio in the nth row n and all the integers from smallest to largest among them;

[0196] The average rank of the weighted rank sum ratio in the first row is the sum of all the integers from smallest to largest in the cumulative frequency ∑f1 of the weighted rank sum ratio in the first row divided by the frequency f1;

[0197] The average rank of the weighted rank sum ratio in the second row is the sum of all the integers from smallest to largest from the cumulative frequency ∑f1 + 1 of the weighted rank sum ratio in the first row to the cumulative frequency Σf2 of the weighted rank sum ratio in the second row divided by the frequency f2;

[0198] Until

[0199] The average rank of the weighted rank sum ratio in the nth row is the cumulative frequency of the weighted rank sum ratio in the (n - 1)th row

[0200] ∑f n-1 The cumulative frequency ∑f of the weighted rank sum ratio from the 1st to the nth row n The sum of all integers from smallest to largest divided by the frequency f n value;

[0201] Step Four Three Three Three: Calculate the downward cumulative frequency of the weighted rank sum ratio for each row where m is the number of evaluation sections;

[0202] Step Four Three Three Four: Convert the downward cumulative frequency p of the weighted rank sum ratio for each row into the value p of the probit r .

[0203] Other steps and parameters are the same as those in the first to seventh specific embodiments

[0204] Specific Embodiment Nine: The difference between this embodiment and any one of the first to eighth specific embodiments is that in the above-mentioned Step Four Three Three Three, when calculating the downward cumulative frequency of the weighted rank sum ratio for each row where m is the number of evaluation sections;

[0205] The specific process is as follows:

[0206] The downward cumulative frequency of the weighted rank sum ratio of the first row is

[0207] The downward cumulative frequency of the weighted rank sum ratio of the second row is

[0208] The downward cumulative frequency of the weighted rank sum ratio of the nth row is

[0209] For perform correction; the process is as follows:

[0210] The downward cumulative frequency of the weighted rank sum ratio of the nth row is

[0211] Other steps and parameters are the same as those in the first to eighth specific embodiments

[0212] Specific Embodiment Ten: The difference between this embodiment and any one of the first to ninth specific embodiments is that in the above-mentioned Step Four Three Three Four, when converting the downward cumulative frequency p of the weighted rank sum ratio for each row into the value p of the probit r ;

[0213] The specific process is as follows:

[0214] Retrieve the downward cumulative frequency of the weighted rank sum ratio of the first row in the percentage - probit conversion table to obtain the corresponding value p of the probit r ;

[0215] The downward cumulative frequency of the weighted rank sum ratio of the second row Retrieve in the percentage and probit conversion table to obtain the value p of the corresponding probit r ;

[0216] The downward cumulative frequency of the weighted rank sum ratio of the nth row Retrieve in the percentage and probit conversion table to obtain the value p of the corresponding probit r ;

[0217] The rows of the percentage and probit conversion table are 0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9;

[0218] The columns of the percentage and probit conversion table are 1, 2, 3, 4, 5, 6,..., 100;

[0219] That is, the percentage and probit conversion table ranges from 0 to 99.9. If p is 20.1, first find 20 in the column and then 0.1 in the row, and the cell corresponding to the row and column is the p corresponding to 20.1 r .

[0220] Other steps and parameters are the same as those in the first to ninth specific embodiments

[0221] The following details the embodiments of the present invention, and the examples of the embodiments are shown in the drawings

[0222] A traffic accident risk analysis method based on urban checkpoint license plate data provided by this embodiment includes the following steps:

[0223] I. Processing of traffic checkpoint license plate data

[0224] 1) First, parse the traffic checkpoint data involved in the present invention. The traffic checkpoint data is the real-time captured data of the electronic police system installed at intersections, mainly including field attributes such as license plate number, license plate type, capture time, and capture location. The meanings of each field are shown in Table 1

[0225] Table 1 Main fields of traffic checkpoint data

[0226]

[0227] 2) Due to the limitations of the device's own sampling rate and reasons such as license plate occlusion, unlicensed vehicles, bad weather, or vehicle lane changes, there is a certain amount of data noise in the originally collected traffic checkpoint data. The abnormal situations of the data are mainly divided into data missing, data error, and data redundancy

[0228] 3) The license plate data used in the present invention has sensitivity issues. While protecting vehicle privacy, it is necessary to ensure that each vehicle has a unique identifiable number.

[0229] 4) The content of the "affiliated organization" field in the vehicle license plate data is the Chinese name of all intersections, which is stored as a string type in the database, occupying a large amount of storage space, and there will be a problem that the string is too long to display the trajectory during trajectory extraction. The same applies to the "capture time" field.

[0230] The processing process is as follows:

[0231] First step, according to the license plate number character length limit and the condition of not allowing illegal characters, directly delete the missing and incorrect data;

[0232] Second step, after sorting the traffic checkpoint data in ascending order according to the license plate number and capture time, if the license plate numbers and capture locations of adjacent data are the same and the capture time interval is less than the set threshold, that is, when all of the following formulas are satisfied, it is determined as redundant data. Traverse all the data, delete the redundant data, and only keep the first valid data.

[0233] ID i =ID i+1 (1)

[0234] Sid i =Sid i+1 (2)

[0235] t i+1 -t i <λ (3)

[0236] In the formula,

[0237] ID i is the license plate number of the i-th row of data, ID i+1 is the license plate number of the (i + 1)-th row of data;

[0238] Sid i is the intersection number where the i-th row of data is located, Sid i+1 is the intersection number where the (i + 1)-th row of data is located;

[0239] t i is the capture time of the i-th row of data, t i+1 is the capture time of the (i + 1)-th row of data;

[0240] λ is the time interval threshold, which is taken as 10s in this embodiment;

[0241] A is a positive integer;

[0242] Step 3: First, process the license plate numbers in the MYSQL database, then sort the data in ascending order by intersection name, and then traverse and group the intersection names for numbering.

[0243] Step 4: Discard the data of the date part, that is, delete the "year / month / day" part. Taking 0:00:00 in the early morning of the same day as the time starting point, convert the time field as follows:

[0244] T = a × 3600 + b × 3600 + c

[0245] In the formula: T is the converted time; a is the hour digit in the original time; b is the minute digit; c is the second digit.

[0246] By processing the original traffic checkpoint data, delete the missing and incorrect data, and identify and delete the redundant data. Taking the traffic checkpoint data of a certain day as an example, the original data contains a total of 1,378,883 records. After screening out the vehicles without license plates, the license plates containing incorrect characters, and the duplicate data, there are 1,227,222 records left, and the proportion of valid data accounts for 89% of the total data. After performing license plate numbering, intersection numbering, and time format conversion on the data, the data format is shown in Table 2.

[0247] Table 2 Data format after processing

[0248]

[0249] II. Construction of trajectory coordinate database

[0250] Construct a vehicle trajectory database. Taking each intersection as a travel trajectory point, first group the license plate number field, screen out all vehicles, and then splice the "location" and "time" fields in ascending order of time respectively to establish a vehicle trajectory data table, thereby constructing a vehicle trajectory database.

[0251] Step 1: Group the license plate recognition data by license plate number;

[0252] Step 2: Sort each group of license plate numbers in ascending order of time;

[0253] Step 3: Splice the location and time attribute fields of each group of license plate numbers respectively, separated by ",";

[0254] Step 4: Save the obtained result as a new trajectory table to form a trajectory database;

[0255] Step 5: Determine the longitude and latitude coordinates of each intersection and construct a trajectory point coordinate table;

[0256] Step 6: Combine the longitude and latitude coordinates corresponding to each intersection with the license plate recognition data to establish a trajectory coordinate database, as shown in Table 3.

[0257] Using the MYSQL database, the "latitude and longitude coordinates" are concatenated by license plate number grouping. After the execution is completed, as shown in Table 9, after extracting the trajectory routes of each vehicle by the method of grouping character concatenation, where the trajectory routes are represented by latitude and longitude coordinates, it shows that the correspondence and extraction of trajectory coordinates are realized.

[0258] Table 3 Trajectory Coordinate Data Table

[0259]

[0260] III. Obtaining Traffic Flow Characteristic Indexes

[0261] The accident risk alternative evaluation indexes are divided into indexes based on traffic conflicts and indexes based on traffic flow characteristics from two levels: macro and micro. Since the vehicle travel trajectories obtained in the present invention are point trajectories, it is difficult to obtain the conflict behaviors of vehicles approaching in the same time and space. Therefore, the present invention adopts traffic flow characteristic indexes such as traffic flow, average speed, and vehicle type, which focus on showing the dynamic operation status of the overall road traffic flow. The specific steps for obtaining traffic flow characteristic indexes are as follows:

[0262] The first step: Determine the checkpoint numbers of the OD points of the risk evaluation section, screen out all vehicle data passing through this section, and then filter the time field to obtain the vehicle data passing through this section during the peak period;

[0263] The second step: Group the vehicle data passing through the same section according to the "vehicle type" field, and divide them into three categories: cars, buses, and trucks. The vehicle type division is shown in Table 4;

[0264] Table 4 Vehicle Type Division Table

[0265]

[0266] The third step: Statistically analyze the grouped vehicle data, and calculate the traffic flow index of each vehicle type according to the following formula;

[0267]

[0268] In the formula:

[0269] Q i is the traffic flow index of the i-th type of vehicle on the section, i = 1, 2, 3;

[0270] Q1 is the traffic flow of the first type of vehicle on the section, Q2 is the traffic flow of the second type of vehicle on the section, and Q3 is the traffic flow of the third type of vehicle on the section;

[0271] The first type of vehicle is a car; the second type of vehicle is a bus; the third type of vehicle is a truck;

[0272] N i The total number of vehicles of the \(i\)-th vehicle type passing through the OD of the risk assessment section, where \(i = 1, 2, 3\);

[0273] \(N_1\) is the total number of vehicles of the 1st vehicle type passing through the OD of the risk assessment section, \(N_2\) is the total number of vehicles of the 2nd vehicle type passing through the OD of the risk assessment section, and \(N_3\) is the total number of vehicles of the 3rd vehicle type passing through the OD of the risk assessment section;

[0274] \(T\) is the duration within the selected time period;

[0275] Step 4: For the grouped vehicle data, calculate the difference between the arrival and departure times of each vehicle to obtain the vehicle running time, and calculate the average speed index of each vehicle type according to the following formula;

[0276]

[0277] In the formula: \(V\) i is the average speed of the \(i\)-th vehicle type;

[0278] \(V_1\) is the average speed of the 1st vehicle type; \(V_2\) is the average speed of the 2nd vehicle type; \(V_3\) is the average speed of the 3rd vehicle type;

[0279] \(S\) is the total length of the OD of the risk assessment section;

[0280] \(t\) n is the running time of the \(t\)-th n vehicle among the \(i\)-th vehicle type, where \(i = 1, 2, 3\);

[0281] IV. Construction of the risk assessment model

[0282] For the obtained risk assessment indicators, it is necessary to first establish an initial matrix for risk analysis and normalize the indicators. After normalization, the entropy weight method is used to calculate the weight values of the evaluation indicators, the rank sum ratio comprehensive evaluation method is used to calculate the comprehensive evaluation value of the section, determine the distribution of RSR and calculate the regression equation. Finally, substitute the Probit value of the accident occurring on the section into the regression equation to derive the RSR estimated value to judge the accident risk degree of the section. The specific steps are as follows:

[0283] Step 1: For the six traffic flow characteristic indicators, the initial matrix for risk analysis is defined as follows:

[0284]

[0285] In the formula: \(M\) is the model input matrix; \(m\) is the number of evaluation sections; \(x\) st is the value of the \(t\)-th risk assessment indicator of the \(s\)-th section to be evaluated.

[0286] Step 2: For the obtained initial matrix for risk analysis, normalize the indicator data according to the following formula:

[0287]

[0288] In the formula: x st is the value of the t-th risk evaluation index of the s-th road section to be evaluated, where s = 1, 2, …, m;

[0289] max(x st ) is the maximum value of the t-th risk evaluation index of the s-th road section to be evaluated, max(x st ) = max(x 1t , x 2t , …, x mt );

[0290] min(x st ) is the minimum value of the t-th risk evaluation index of the s-th road section to be evaluated, min(x st ) = min(x 1t , x 2t , …, x mt );

[0291] Step 3: Calculate the entropy value of each index according to the following formula:

[0292]

[0293] In the formula: p st is the proportion of the t-th risk evaluation index of the s-th road section to be evaluated after normalization, where s = 1, 2, …, m;

[0294] e t is the entropy value of the t-th risk evaluation index of the s-th road section to be evaluated after normalization, where s = 1, 2, …, m;

[0295] m is the number of evaluated road sections.

[0296] Step 4: According to the obtained entropy value e t , calculate the weight value of each index according to the following formula:

[0297]

[0298] In the formula: w t is the weight value of the t-th risk evaluation index value; n is the number of risk evaluation indexes.

[0299] Step 5: Use the non-integer rank sum ratio method to rank the initial matrix of risk analysis:

[0300]

[0301] In the formula: R st is the rank of the element in the s-th row and t-th column,

[0302] Step 6: Calculate the weighted rank sum ratio:

[0303]

[0304] Step 7: Sort the weighted rank sum ratio WRSR s values from smallest to largest to obtain the sorted WRSR1, WRSR2, …, WRSR n ;

[0305] Compile the frequency f of the weighted rank sum ratio for each row after sorting, and calculate the cumulative frequency ∑f of the weighted rank sum ratio for each row; determine the rank R and the average rank of the weighted rank sum ratio for each row

[0306] Calculate the downward cumulative frequency of the row WRSR s where m is the number of evaluation road sections;

[0307] Convert the downward cumulative frequency p of the weighted rank sum ratio for each row into the probit value p of the probability unit

[0308] ; r ;

[0309] Step 8: Using the probit value p corresponding to the cumulative frequency r as the independent variable and the corresponding weighted rank sum ratio WRSR s value as the dependent variable, calculate the regression equation to obtain the parameters to be estimated a, b:

[0310] WRSR s = a + bP r , s = 1, 2, …, m

[0311] In the formula:

[0312] p r is the probit value corresponding to the cumulative frequency, and a, b are the parameters to be estimated;

[0313] Step 9: Classify the accident risks of the road sections to be evaluated into three levels: high-risk road sections, medium-risk road sections, and low-risk road sections;

[0314] The percentile critical value L of the low-risk road section < 15.886, and the probit value p r < 4;

[0315] The percentile critical value of the medium-risk road section 15.886 ≤ L < 84.134, and the probit value 4 ≤ p r < 6;

[0316] The percentile critical value of the high-risk road section L ≥ 84.134, and the probit value p r ≥ 6;

[0317] Substitute the probit value p r = 4 into the regression equation WRSR for determining the parameters a and b to be estimated s = a + bP r , and obtain the estimated value A of the weighted rank sum ratio WRSR s ;

[0318] Substitute the probit value p r = 6 into the regression equation WRSR for determining the parameters a and b to be estimated s = a + bP r , and obtain the estimated value B of the weighted rank sum ratio WRSR s ;

[0319] The WRSR critical value A' for the low - risk section < A;

[0320] The WRSR critical value for the medium - risk section A ≤ A' < B

[0321] The WRSR critical value for the high - risk section B ≤ A';

[0322] Step 10: Substitute the probit value p r of the seventh step into the regression equation for determining the parameters a and b to be estimated, and obtain the estimated values of n WRSR;

[0323] Judge the range of the WRSR critical value to which the estimated value of each WRSR s belongs, and determine whether each WRSR s is a low - risk section, a medium - risk section or a high - risk section.

[0324] The risk level classification is shown in Table 5.

[0325] Table 5 Risk Level Classification Table

[0326]

[0327] The present invention can also have many other embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art can make various corresponding changes and deformations according to the present invention, but these corresponding changes and deformations should all fall within the protection scope of the appended claims of the present invention.

Claims

1. A traffic accident risk assessment method based on license plate data of urban checkpoints, characterized in that: The specific process of the method is as follows: Step 1: Obtain the license plate data of traffic checkpoints, process the license plate data of traffic checkpoints, and obtain the processed license plate data of traffic checkpoints; Step 2: Build a vehicle trajectory database based on the processed license plate data of traffic checkpoints obtained in Step 1; Step 3: Determine the longitude and latitude coordinates of the checkpoints, and convert the trajectory database into a trajectory coordinate database; Step 4: Build an initial risk analysis matrix, normalize the risk evaluation index values in the initial risk analysis matrix, and obtain the normalized risk evaluation index values; Based on the normalized risk evaluation index values, use the entropy weight method to calculate the evaluation index weight values; based on the normalized risk evaluation index values and the evaluation index weight values, use the rank sum ratio comprehensive evaluation method to evaluate the road section level.

2. The traffic accident risk assessment method based on urban checkpoint license plate data according to claim 1, wherein: In Step 1, obtain the license plate data of traffic checkpoints, process the license plate data of traffic checkpoints, and obtain the processed license plate data of traffic checkpoints; The specific process is as follows: Step 1-1: Obtain the license plate data of traffic checkpoints through the electronic police equipment at urban intersections, record the license plate data of traffic checkpoints row by row, screen the license plate data of traffic checkpoints, screen out the missing data and error data, directly delete the missing and error data, and obtain the screened license plate data of traffic checkpoints; The license plate data of traffic checkpoints includes the following attribute fields: license plate number, license plate type, capture time, affiliated organization, capture location, driving direction, vehicle type; Step 1-2: Group the screened license plate data of traffic checkpoints according to the license plate number, with the same license plate number in one group, and sort each group of license plate data in ascending order according to the capture time; If the license plate numbers of adjacent A pieces of data in a group are the same, the intersection positions corresponding to the license plate numbers of adjacent A pieces of data are the same, and the capture time interval of adjacent A pieces of data is less than the set threshold, that is, all of the formulas (1)-(3) are satisfied, then it is determined as redundant data, and only the first valid piece of data is retained for the A pieces of redundant data; ID i = ID i+1 (1) Sid i = Sid i+1 (2) t i+1 -t i <λ (3) In the formula, ID i is the license plate number ID for the data of the i-th row, ID i+1 is the license plate number ID for the data of the (i + …1)-th row; Sid i is the intersection ID where the i-th row of data is located, Sid i+1 is the intersection ID where the (i + 1)-th row of data is located; t i is the capture time of the data in the i-th row, t i+1 is the capture time of the data in the (i + 1)-th row; λ is the time interval threshold; A is a positive integer; Step 1-3: Number the intersections in the data obtained in Step 1-2, and convert the time format, uniformly with 0:00:00 in the early morning of the same day as the time starting point, and convert the time into a format in seconds; T = a×3600 + b×3600 + c (4) In the formula: T is the converted time; a is the hour digit in the original time; b is the minute digit; c is the second digit.

3. The traffic accident risk assessment method based on urban checkpoint license plate data according to claim 2, characterized in that: In Step 2, build a vehicle trajectory database based on the processed license plate data of traffic checkpoints obtained in Step 1; the specific process is as follows: Step 2-1: Group the processed license plate data of traffic checkpoints obtained in Step 1 according to the license plate number; Step 2-2: Sort each group of license plate numbers in ascending order according to the time; Step 2-3: Concatenate the locations of each group of license plate numbers at each traffic checkpoint, and separate them with ","; Concatenate the time field values of each group of license plate numbers at each traffic checkpoint, and separate them with ","; Step 2-4: Save the result obtained in Step 2-3 as a new trajectory table to form a vehicle trajectory database.

4. The traffic accident risk assessment method based on urban checkpoint license plate data according to claim 3, characterized in that: In Step 4, an initial risk analysis matrix is constructed, and the risk evaluation index values in the initial risk analysis matrix are normalized to obtain the normalized risk evaluation index values; Based on the normalized risk evaluation index values, the entropy weight method is used to calculate the evaluation index weight values; based on the normalized risk evaluation index values and the evaluation index weight values, the rank sum ratio comprehensive evaluation method is used to evaluate the road section grade; The specific process is as follows: Step 4-1: Construct an initial risk analysis matrix, and normalize the risk evaluation index values in the initial risk analysis matrix to obtain the normalized risk evaluation index values; Step 4-2: Based on the normalized risk evaluation index values, use the entropy weight method to calculate the evaluation index weight values; Step 4-3: Based on the normalized risk evaluation index values and the evaluation index weight values, use the rank sum ratio comprehensive evaluation method to evaluate the road section grade.

5. The traffic accident risk assessment method based on urban checkpoint license plate data according to claim 4, characterized in that: In Step 4-1 described above, an initial risk analysis matrix is constructed, and the risk evaluation index values in the initial risk analysis matrix are normalized to obtain the normalized risk evaluation index values; The specific process is as follows: Step 4-1-1: Determine the road section between adjacent traffic checkpoints as the risk evaluation road section OD; Filter out all vehicle data passing through the risk evaluation road section OD; Set the selected time period, and filter the time field of all the filtered vehicle data passing through the risk evaluation road section OD to obtain the vehicle data passing through the risk evaluation road section within the selected time period; Step 4-1-2: Group the vehicle data obtained in Step 4-1-1 according to the "vehicle type" field and divide it into three vehicle types: cars, buses, and trucks; Step 413: Statistically analyze the vehicle data obtained in Step 412, and calculate the traffic flow index Q of the i-th type of vehicle according to (5). i ; Where: Q i is the traffic flow index of the i-th type of vehicle, where i = 1, 2, 3; Q1 is the road section traffic flow of the first type of vehicle, Q2 is the road section traffic flow of the second type of vehicle, and Q3 is the road section traffic flow of the third type of vehicle; The first type of vehicle is a car; the second type of vehicle is a bus; the third type of vehicle is a truck; N i is the total number of vehicles passing through the OD of the risk assessment section for the i-th type of vehicle, where i = 1, 2, 3; N1 is the total number of vehicles of the first type passing through the risk evaluation road section OD, N2 is the total number of vehicles of the second type passing through the risk evaluation road section OD, and N3 is the total number of vehicles of the third type passing through the risk evaluation road section OD; T is the duration within the selected time period; Step 4-1-4: Based on the vehicle data obtained in Step 4-1-2, calculate the difference between the arrival and departure times of each vehicle on the risk evaluation road section OD to obtain the running time of each vehicle; Based on the running time of each vehicle, calculate according to (6) to obtain the average vehicle speed index V of the i-th type of vehicle i , where i = 1, 2, 3; Where: V i is the average vehicle speed of the i-th type of vehicle; V1 is the average speed of the first type of vehicle, V2 is the average speed of the second type of vehicle, and V3 is the average speed of the third type of vehicle; S is the total length of the risk evaluation road section OD; t n is the running time of the t-th vehicle in the i-th type of vehicle, where i = 1, 2, 3; n ​ Step 4-1-5: Based on the road section traffic flow Q1 of the first type of vehicle, the road section traffic flow Q2 of the second type of vehicle, the road section traffic flow Q3 of the third type of vehicle, the average speed V1 of the first type of vehicle, the average speed V2 of the second type of vehicle, and the average speed V3 of the third type of vehicle, construct an initial risk analysis matrix; The initial risk analysis matrix is defined as follows: Where: M is the initial matrix of risk analysis; m is the number of evaluation sections; x st is the value of the t-th risk evaluation index of the s-th section to be evaluated; s = 1, 2, …, m; x s1 It is the value of the first risk assessment index for the sth section to be evaluated. The value of the first risk assessment index is the traffic flow Q1 of the first type of vehicle on the section, where s = 1, 2, …, m; x s2 It is the value of the second risk assessment index for the sth road section to be evaluated. The value of the second risk assessment index is the traffic flow Q2 of the second type of vehicle on the road section, where s = 1, 2, …, m; x s3 It is the value of the third risk assessment index for the sth road section to be evaluated. The value of the third risk assessment index is the traffic flow Q3 of the third type of vehicle on the road section, where s = 1, 2, …, m; x s4 It is the value of the 4th risk assessment index for the sth road section to be evaluated. The value of the 4th risk assessment index is the average vehicle speed V1 of the 1st type of vehicle, where s = 1, 2, …, m; x s5 It is the value of the 5th risk assessment index for the sth road section to be evaluated. The value of the 5th risk assessment index is the average vehicle speed V2 of the 2nd type of vehicle, where s = 1, 2, …, m; x s6 is the value of the 6th risk assessment index for the sth road section to be evaluated. The value of the 6th risk assessment index is the average speed V3 of the 3rd type of vehicle, where s = 1, 2, …, m; Step 416: Normalize the value x of the t-th risk evaluation index for the s-th section to be evaluated in the initial risk analysis matrix in Step 415 st to obtain the value y of the t-th risk evaluation index for the s-th section to be evaluated after normalization st , where t = 1, 2, …, 6; which is expressed as: In the formula: y s1 It is the value of the first risk assessment index of the sth road section to be evaluated after normalization, where s = 1, 2, …, m; y s2 It is the second risk assessment index value of the sth road section to be evaluated after normalization, where s = 1, 2, …, m; y s3 It is the value of the 3rd risk assessment index for the sth road section to be evaluated after normalization, where s = 1, 2, …, m; y s4 It is the value of the 4th risk assessment index for the sth road section to be evaluated after normalization, where s = 1, 2, …, m; y s5 It is the value of the 5th risk evaluation index for the sth road section to be evaluated after normalization, where s = 1, 2, …, m; y s6 It is the value of the 6th risk evaluation index of the sth section to be evaluated after normalization, where s = 1, 2, …, m; max(x s1 ) is the maximum value of the first risk assessment index for the sth road section to be evaluated, where s = 1, 2, …, m, and max(x s1 ) = max(x 11 , x 21 , …, x m1 ); max(x s2 ) is the maximum value of the second risk assessment index for the s-th road section to be evaluated, where s = 1, 2, …, m, and max(x s2 ) = max(x 12 , x 22 , …, x m2 ); max(x s3 ) is the maximum value of the third risk assessment index for the s-th road section to be evaluated, where s = 1, 2, …, m, and max(x s3 ) = max(x 13 , x 23 , …, x m3 ); max(x s4 ) is the maximum value of the 4th risk evaluation index for the sth road section to be evaluated, where s = 1, 2, …, m, and max(x s4 ) = max(x 14 , x 24 , …, x m4 ); max(x s5 ) is the maximum value of the 5th risk assessment index for the sth road section to be evaluated, where s = 1, 2, …, m, and max(x s5 ) = max(x 15 , x 25 , …, x m5 ); max(x s6 ) is the maximum value of the 6th risk assessment index for the sth road section to be evaluated, where s = 1, 2, …, m, and max(x s6 ) = max(x 16 , x 26 , …, x m6 ); min(x s1 ) is the minimum value of the first risk assessment index for the sth road section to be evaluated, where s = 1, 2, …, m, and min(x s1 ) = min(x 11 , x 21 , …, x m1 ); min(x s2 ) is the minimum value of the second risk assessment index for the sth road section to be evaluated, where s = 1, 2, …, m, min(x s2 ) = min(x 12 , x 22 , …, x m2 ); min(x s3 ) is the minimum value of the third risk assessment indicator for the sth section to be evaluated, where s = 1, 2, …, m, min(x s3 ) = min(x 13 , x 23 , …, x m3 ); min(x s4 ) is the minimum value of the 4th risk assessment index for the sth road section to be evaluated, where s = 1, 2, …, m, min(x s4 ) = min(x 14 , x 24 , …, x m4 ); min(x s5 ) is the minimum value of the 5th risk assessment index for the sth road section to be evaluated, where s = 1, 2, …, m, min(x s5 ) = min(x 15 , x 25 , …, x m5 ); min(x s6 ) is the minimum value of the 6th risk assessment index for the sth road section to be evaluated, where s = 1, 2, …, m, min(x s6 ) = min(x 16 , x 26 , …, x m6 ).

6. The traffic accident risk assessment method based on urban checkpoint license plate data according to claim 5, characterized in that: In Step 4-2 described above, based on the normalized risk evaluation index values, the entropy weight method is used to calculate the evaluation index weight values; the specific process is as follows: Step 4-2-1: According to the value y of the t-th risk evaluation index of the s-th road section to be evaluated after normalization st , calculate the proportion p of the t-th risk evaluation index of the s-th road section to be evaluated after normalization st , where t = 1, 2, …, 6; It is expressed as: In the formula: p s1 It is the proportion of the first risk assessment index of the sth road section to be evaluated after normalization processing, where s = 1, 2, …, m; p s2 It is the proportion of the second risk assessment index of the sth road section to be evaluated after normalization, where s = 1, 2, …, m; p s3 It is the proportion of the third risk assessment index of the sth road section to be evaluated after normalization, where s = 1, 2, …, m; p s4 It is the proportion of the 4th risk assessment index of the sth section to be evaluated after normalization, where s = 1, 2, …, m; p s5 It is the proportion of the 5th risk assessment index of the sth road section to be evaluated after normalization, where s = 1, 2, …, m; p s6 It is the proportion of the 6th risk assessment index of the sth road section to be evaluated after normalization, where s = 1, 2, …, m; Step 422: Based on the proportion p of the t-th risk assessment indicator of the s-th road section to be evaluated after normalization st Calculate the entropy value e of the t-th risk assessment indicator of the s-th road section to be evaluated after normalization t , where t = 1, 2, 3, 4, 5, 6; It is expressed as: In the formula: e1 is the entropy value of the first risk evaluation index of the s-th road section to be evaluated after normalization, s = 1, 2,..., m; $e_2$ is the entropy value of the second risk assessment index of the $s$-th road section to be evaluated after normalization, where $s = 1, 2, \ldots, m$; $e_3$ is the entropy value of the third risk assessment index of the $s$-th road section to be evaluated after normalization, where $s = 1, 2, \ldots, m$; $e_4$ is the entropy value of the fourth risk assessment index of the $s$-th road section to be evaluated after normalization, where $s = 1, 2, \ldots, m$; $e_5$ is the entropy value of the fifth risk assessment index of the $s$-th road section to be evaluated after normalization, where $s = 1, 2, \ldots, m$; $e_6$ is the entropy value of the sixth risk assessment index of the $s$-th road section to be evaluated after normalization, where $s = 1, 2, \ldots, m$; $m$ is the number of road sections to be evaluated; Step Four Two Three: According to the entropy value e obtained in Step Four Two Two t , calculate the weight value w of the t-th risk evaluation index t , where t = 1, 2, 3, 4, 5, 6; expressed as: In the formula: $w_1$ is the weight value of the first risk assessment index; $w_2$ is the weight value of the second risk assessment index; $w_3$ is the weight value of the third risk assessment index; $w_4$ is the weight value of the fourth risk assessment index; $w_5$ is the weight value of the fifth risk assessment index; $w_6$ is the weight value of the sixth risk assessment index; $n$ is the number of risk assessment indices.

7. A traffic accident risk assessment method based on urban checkpoint license plate data according to claim 6, characterized in that: In the third sub-step of the fourth step, the road section grade is evaluated by using the rank sum ratio comprehensive evaluation method based on the risk assessment index values and evaluation index weight values after normalization; The specific process is as follows: Sub-step 431: Rank the initial risk analysis matrix by using the non-integer rank sum ratio method: In the formula: R s1 is the rank of the element in the first column of the s-th row, R s2 is the rank of the element in the second column of the s-th row, R s3 is the rank of the element in the third column of the s-th row, R s4 is the rank of the element in the fourth column of the s-th row, R s5 is the rank of the element in the fifth column of the s-th row, R s6 is the rank of the element in the sixth column of the s-th row, s = 1, 2, …, m; Step Four Three Two: Based on the weight value w of the t-th risk assessment indicator t and the rank R of the element in the s-th row and t-th column st , calculate the weighted rank sum ratio WRSR of the s-th row s ; expressed as: where: w t is the weight value of the t-th risk evaluation index; R st is the rank of the element in the s-th row and t-th column; Step Four Three Three: For the weighted rank sum ratio WRSR s values, sort them from smallest to largest to obtain the sorted WRSR1, WRSR2, …, WRSR n ; Compile the frequency $f$ of the weighted rank sum ratio of each row after sorting, and calculate the cumulative frequency $\sum f$ of the weighted rank sum ratio of each row respectively; Determine the rank R and average rank of the weighted rank sum ratio for each row Calculate the downward cumulative frequency of the weighted rank sum ratio for each row where $m$ is the number of road sections to be evaluated; Convert the downward cumulative frequency p of the weighted rank sum ratio of each row into the Probit value p of the probability unit r ; Step Four Three Four: Using the probability unit value p corresponding to the cumulative frequency r as the independent variable and the corresponding weighted rank sum ratio WRSR s value as the dependent variable to calculate the regression equation, obtaining the parameters to be estimated a and b: WRSR s = a + bP r , s = 1, 2, …, m In the formula: p r is the probit value corresponding to the cumulative frequency, and a and b are parameters to be estimated; Sub-step 435: The accident risks of the road sections to be evaluated are divided into three levels: high-risk road sections, medium-risk road sections, and low-risk road sections; The percentile critical value L of the low-risk section < 15.886, and the probit value p r < 4; The percentile critical value of the medium-risk section is 15.886 ≤ L < 84.134, and the probit value is 4 ≤ p r < 6; The percentile critical value L of the high-risk section ≥ 84.134, and the probit value p r ≥ 6; Substitute the probit value p r = 4 into the regression equation WRSR for determining the parameters a and b to be estimated s = a + bP r , and obtain the estimated value A of the weighted rank sum ratio WRSR s ; Substitute the probit value p r = 6 into the regression equation WRSR for determining the parameters a and b to be estimated s = a + bP r to obtain the estimated value B of the weighted rank sum ratio WRSR s ; The critical value $A'$ of the WRSR for low-risk road sections is $A' \lt A$; The critical value $A'$ of the WRSR for medium-risk road sections satisfies $A \leq A' \lt B$ The critical value $A'$ of the WRSR for high-risk road sections satisfies $B \leq A'$; Sub-step 436: Substitute the Probit value p obtained in Step 4344 r into the regression equation for determining the parameters to be estimated a and b to obtain n estimated values of WRSR; Judge each WRSR s to determine the WRSR critical value range to which its estimated value belongs, and determine each WRSR s as a low-risk section, a medium-risk section or a high-risk section.

8. A traffic accident risk assessment method based on urban checkpoint license plate data according to claim 7, characterized in that: In step 433, the weighted rank sum ratio WRSR s is sorted in ascending order of values to obtain the sorted WRSR1, WRSR2, …, WRSR n ; Compile the frequency $f$ of the weighted rank sum ratio of each row after sorting, and calculate the cumulative frequency $\sum f$ of the weighted rank sum ratio of each row respectively; Determine the rank R and average rank of the weighted rank sum ratio for each row Calculate the downward cumulative frequency of the weighted rank sum ratio for each row where $m$ is the number of road sections to be evaluated; Convert the downward cumulative frequency p of the weighted rank sum ratio per row into the Probit value p of the probability unit r ; The specific process is as follows: Step Four Three Three One: Sort the weighted rank sum ratio (WRSR) s values from smallest to largest to obtain the sorted WRSR1, WRSR2, …, WRSR n ; Compile the frequency $f$ of the weighted rank sum ratio of each row after sorting, and calculate the cumulative frequency $\sum f$ of the weighted rank sum ratio of each row respectively; The specific process is as follows: Sort the weighted rank sum ratio values from small to large. The same weighted rank sum ratio values are in the same row, and different weighted rank sum ratio values are in different rows, with a total of $n$ rows; $n \leq m$ The frequency $f_1$ of the weighted rank sum ratio of the first row after sorting is the number of times the weighted rank sum ratio appears in the first row; The frequency $f_2$ of the weighted rank sum ratio of the second row after sorting is the number of times the weighted rank sum ratio appears in the second row; Until The frequency f of the weighted rank sum ratio of the n-th row after sorting n is the number of times the weighted rank sum ratio appears in the n-th row; The cumulative frequency $\sum f_1$ of the weighted rank sum ratio of the first row is the number of times the weighted rank sum ratio appears in the first row; The cumulative frequency $\sum f_2$ of the weighted rank sum ratio of the second row is the sum of the number of times the weighted rank sum ratio appears in the first row to the number of times the weighted rank sum ratio appears in the second row; Until The cumulative frequency ∑f of the weighted rank sum ratio of the nth row n is the sum of the number of times the weighted rank sum ratio appears in the first row to the number of times the weighted rank sum ratio appears in the nth row; Step Four Three Three Two: Determine the rank R and the average rank of the weighted rank sum ratio for each row The specific process is as follows: The rank $R$ of the weighted rank sum ratio of the first row is all the integers from small to large in the cumulative frequency $\sum f_1$ of the weighted rank sum ratio of the first row; The rank $R$ of the weighted rank sum ratio of the second row is all the integers from small to large from $\sum f_1 + 1$ of the cumulative frequency of the weighted rank sum ratio of the first row to $\sum f_2$ of the cumulative frequency of the weighted rank sum ratio of the second row; Until The rank R of the weighted rank sum ratio of the nth row is the cumulative frequency ∑f of the weighted rank sum ratio of the (n - 1)th row n-1 + 1 to the cumulative frequency ∑f of the weighted rank sum ratio of the nth row n for all integers from smallest to largest; The average rank of the first-row weighted rank sum ratio is the sum of all integers from smallest to largest in the cumulative frequency ∑f1 of the first-row weighted rank sum ratio divided by the frequency f1; Average rank of the second-row weighted rank sum ratio The value is the sum of all integers from the cumulative frequency ∑f1 + 1 of the first-row weighted rank sum ratio to the cumulative frequency Σf2 of the second-row weighted rank sum ratio, arranged in ascending order, divided by the frequency f2; Until The average rank of the weighted rank sum ratio for the nth row takes the value of the cumulative frequency of the weighted rank sum ratio for the (n - 1)th row Σf n-1 The cumulative frequency ∑f of the weighted rank sum ratio from the 1st to the nth row n The sum of all integers from smallest to largest in divided by the frequency f n value; Step Four Three Three Three: Calculate the downward cumulative frequency of the weighted rank sum ratio for each row where $m$ is the number of road sections to be evaluated; Step Four Three Three Four: Convert the downward cumulative frequency p of the weighted rank sum ratio for each row into the Probit value p of the probability unit r .

9. The traffic accident risk assessment method based on urban checkpoint license plate data according to claim 8, wherein: Calculate the downward cumulative frequency of the weighted rank sum ratio for each row in step 4333 where $m$ is the number of road sections to be evaluated; The specific process is as follows: The downward cumulative frequency of the first - row weighted rank - sum ratio is The downward cumulative frequency of the second - row weighted rank - sum ratio is The downward cumulative frequency of the weighted rank sum ratio of the nth row is To make corrections; the process is as follows: The downward cumulative frequency of the weighted rank sum ratio of the nth row is 10. The traffic accident risk assessment method based on urban checkpoint license plate data according to claim 9, characterized in that: In step 4334, convert the downward cumulative frequency p of the weighted rank sum ratio of each row into the Probit value p of the probability unit r ; The specific process is as follows: Cumulative frequency of the first row of weighted rank sum ratio in the downward direction Retrieve in the percentage-probit conversion table to obtain the value p of the corresponding probit Probit r ; Cumulative frequency downwards of the second - row weighted rank - sum ratio Retrieve in the percentage - probability unit conversion table to obtain the value p of the corresponding probability unit Probit r ; The downward cumulative frequency of the weighted rank sum ratio of the nth row Retrieve in the percentage and probit conversion table to obtain the value p of the corresponding probit Probit r .

Images