CN116383678B - A method for identifying sections where abnormal speed change behavior of operating buses frequently occurs - Google Patents

Abstract

The invention discloses a method for identifying frequent road sections of abnormal speed change behaviors of a passenger car, and relates to a method for identifying frequent road sections of abnormal speed change behaviors of a passenger car. The invention aims to solve the problems that the existing traffic safety hidden danger road section investigation method belongs to statistical analysis after accident occurrence, hidden danger road sections are not recognized and early-warned before accident occurrence, and the economic cost is high and the precision is low. The method for identifying the frequent road sections of the abnormal speed change behavior of the operating passenger car comprises the following specific processes: step 1, basic data acquisition; step 2, correcting the coordinates; step 3, calculating basic data; step 4, identifying abnormal speed change behaviors; and 5, identifying the frequently abnormal speed change road sections. The invention belongs to the technical field of public transportation operation management.

Description

A method for identifying sections where abnormal speed change behavior of operating buses frequently occurs

Technical Field

The invention relates to a method for identifying a road section where abnormal speed change behavior of an operating passenger bus often occurs. The invention belongs to the technical field of public transportation operation management.

Background technique

As an important mode of public transportation, operating buses have the characteristics of large passenger capacity, high transportation efficiency, and comfortable riding environment, and are more suitable for medium and short distance travel. During driving, abnormal speed change behaviors such as sudden braking or sudden accelerator by operating bus drivers may cause discomfort or injury to passengers, and may also cause rear-end collisions, rollovers and other accidents, causing more serious personal injury and property losses. On actual roads, the occurrence of abnormal speed change behaviors often has a certain spatial distribution regularity. The sections where abnormal speed change behaviors often occur have higher safety hazards. Identifying sections where abnormal speed change behaviors of operating buses often occur can, on the one hand, timely warn drivers and avoid traffic accidents; on the other hand, it can also provide information on frequent sections to road management departments to transform sections with hidden dangers to improve road driving safety.

Currently, the identification of abnormal vehicle behavior mostly uses video surveillance or multi-sensor data fusion methods, which require high data scale and quality, and the economic cost of acquisition equipment and storage equipment is high. However, there are relatively few studies on the identification of road sections where abnormal vehicle behavior often occurs. Most of the related studies are based on the existing traffic accident data to identify accident risk sections, which are statistical analyses after the accident occurs. It has not yet identified and warned the potential danger sections before the accident occurs.

Operating buses have the characteristics of fixed routes, dense frequency, and high similarity between operations. In addition, all current operating buses are equipped with the Global Positioning System (GPS), and bus companies can obtain a large amount of data such as the position, speed, acceleration, etc. of operating buses when they are running on fixed routes. The abnormal movement behavior of the vehicle will be reflected in the GPS data. If the sections with frequent abnormal speed changes can be identified based on the above GPS data, there will be no need to purchase additional equipment for data collection, which can not only save economic costs, but also facilitate the use of the accumulated large amount of data to check the sections with hidden dangers.

Summary of the invention

The purpose of the present invention is to solve the problems that the existing methods for checking sections with traffic safety hazards are statistical analysis after an accident occurs, and the hazardous sections have not been identified and warned before the accident occurs, and the economic cost is high and the accuracy is low. A method for identifying sections with frequent abnormal speed changes of operating passenger buses is proposed to improve the accuracy and reliability of the frequent section identification technology.

A method for identifying road sections where abnormal speed change behavior of operating passenger vehicles often occurs. The specific process is as follows:

Step 1. Basic data collection;

Step 2. Perform coordinate correction based on basic data;

Step 3. Calculate the speed and acceleration based on the basic data and coordinate correction;

Step 4. Identify abnormal speed change behavior based on speed and acceleration;

Step 5. Identify frequently occurring abnormal speed change sections based on abnormal speed change behavior.

The beneficial effects of the present invention are:

The method proposed in the present invention is a method for identifying sections where abnormal speed change behaviors of operating buses frequently occur. It relies on easily accessible on-board GPS data and completes coordinate correction through map matching, thereby reducing the negative impact caused by hardware errors of acquisition equipment and improving the accuracy of basic operating parameters of operating buses. It determines the acceleration threshold corresponding to each speed by fitting the acceleration threshold curve to identify abnormal speed change behaviors of operating buses. It then uses the density-based noise spatial clustering method (DBSCAN) to identify sections where abnormal speed change behaviors frequently occur. This technology can provide road safety warnings to operating bus drivers to avoid traffic accidents on sections where abnormal speed change behaviors frequently occur. It also provides the location of sections with potential safety hazards to road management departments to help with road reconstruction.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 is a flow chart of the present invention.

Detailed ways

Specific implementation method 1: In this implementation method, a method for identifying road sections where abnormal speed change behavior of operating passenger vehicles often occurs is as follows:

Step 1. Basic data collection;

Step 2. Perform coordinate correction based on basic data;

Step 3. Calculate the speed and acceleration based on the basic data and coordinate correction;

Step 4. Identify abnormal speed change behavior based on speed and acceleration;

Step 5. Identify frequently occurring abnormal speed change sections based on abnormal speed change behavior.

Specific implementation method 2: This implementation method is different from the specific implementation method 1 in that the basic data is collected in step 1; the specific process is:

Step 1.1. Extract the GPS data of all the running trips of an operating bus line in the past N consecutive days, including the collection time, sampling interval, longitude and latitude of the GPS data, vehicle ID, and trip direction information. It is recommended that the value of N be 180;

Step 1.2. The spatial distribution of the road sections where the abnormal speed change behavior of the operating buses in the up and down directions often occurs has obvious differences. The roads where the operating buses run in the up direction are selected as the research objects. The present invention is also applicable to the roads where the operating buses run in the down direction.

Step 1.3. Define the set of operating shifts in the upward direction of the line as I = {i:i = 1, 2, 3, ..., I}; shift i has a total of K GPS data collection points, and the coordinates of the kth data collection point of shift i are expressed as Where lng and lat represent longitude and latitude respectively, in degrees; GPS sampling interval is T, in seconds; k = 1, 2, 3, ..., K;

Step 1.4. Extract the GIS road network of the area through which the operating bus routes pass, including road names, road numbers, and road grade information.

The other steps and parameters are the same as those in the first embodiment.

Specific implementation method three: This implementation method is different from specific implementation methods one or two in that in step 2, coordinate correction is performed based on basic data (because the collection of GPS coordinates is biased, it cannot be guaranteed that all coordinate points fall accurately on the road network. In order to be more accurate, it is believed that the coordinates after map matching are more accurate. Here, the conversion to plane coordinates after map matching is for the convenience of subsequent calculation); the specific process is:

Step 2.1. Use a map matching algorithm based on a hidden Markov model to complete the map matching of the GPS data collection point and the GIS road network. The matched road network is the actual road network where the GPS data collection point is located.

Step 2.2. The coordinates of the GPS data collection point after map matching is the actual location of the GPS data collection point on the road network, expressed as

Step 2.3. Coordinates after map matching Uniform conversion to plane coordinates/> (For example, the coordinates after map matching can be uniformly converted into plane coordinates using a projection method, and the projection method can be a UTM 3-degree zone projection method). x and y are the projection distances of the GPS data collection point from the central meridian and the equator in a coordinate system with the intersection of the equator and the central meridian as the origin of the coordinates, in meters.

The other steps and parameters are the same as those in the first or second embodiment.

Specific implementation method 4: This implementation method is different from any one of the specific implementation methods 1 to 3 in that the speed and acceleration are calculated based on the basic data and coordinate correction in step 3; the specific process is:

Step 3.1. Express the speed of the kth data collection point after coordinate correction in shift i as The unit is m/s; calculate the speed/> As shown in the following formula;

Where: They are the x-axis coordinates of the plane coordinates of the k+1th and kth data collection points of shift i after the coordinates are corrected, in meters;

They are the y-axis coordinates of the plane coordinates of the k+1th and kth data collection points of shift i after correction, in meters;

Step 3.2. Express the acceleration of the kth data collection point after coordinate correction in shift i as The unit is m/s ² ; calculate the acceleration/> As shown in the following formula;

Where: is the speed of the k-1th data collection point after coordinate correction in shift i.

The other steps and parameters are the same as those in Specific Embodiments 1 to 3.

Specific implementation method 5: This implementation method is different from any one of the specific implementation methods 1 to 4 in that in step 4, abnormal speed change behavior is identified based on speed and acceleration; the specific process is as follows:

Step 4.1. Put all the acceleration data collected in I shifts that are greater than or equal to 0 into the set C ⁺ ,

Put all the acceleration data less than 0 in the data collected during I shifts into the set C ^- ,

The maximum value of the shift running speed is denoted as v _max ;

Where: _vn is the velocity of the nth data point in the set C ⁺ , is the acceleration of the nth data point in the set C ⁺ v _m is the velocity of the mth data point in the set C ^- , /> is the acceleration of the mth data point in the set C ^- ;

Step 4.2. Divide the speed interval [0, v _max ] into a set Q with an interval of 1 m/s.

No. The interval is denoted as Q _σ , the first/> The midpoint of the interval is />

in, is the mathematical symbol for rounding up;

Among the data collection points included in the interval Q _σ , there are data collection points belong to the set C ⁺ , /> The data collection points belong to the set C ^- ;

Step 4.3. Set the interval The data collection points belonging to the sets C ⁺ and C ^- are arranged from small to large according to the acceleration values and put into the sets /> and/>

Step 4.4. Set the The 95th percentile of the medium acceleration value is represented by/> Collection/> The 95th percentile of the medium acceleration value is represented by/>

As the distinguishing value of abnormal speed change behavior in each speed range (here, the 95th percentile of the acceleration value is used as the threshold, that is, the acceleration value exceeding this value is considered to be abnormal and extreme speed change behavior, and the acceleration threshold is given by polynomial fitting later to determine whether it is an abnormal speed change behavior), only the behavior corresponding to 5% of the acceleration is considered to be extreme speed change behavior;

Step 4.5. Express the positive acceleration and negative acceleration distinction value point set as P ⁺ , P ^- , as shown in the following formula;

The acceleration threshold curves are drawn using the data points in the point sets P ⁺ and P ^-, respectively, with the horizontal axis representing the velocity and the vertical axis representing the acceleration;

Step 4.6. Use the least squares method to fit the polynomial curve to the point sets P ⁺ and P ^- , respectively, with velocity v as the independent variable and acceleration a as the dependent variable. Let the polynomial curve functions be a ⁺ (v) and a ^- (v) respectively;

Step 4.7. Fit the polynomials a ⁺ (v) and a ^- (v) with orders 2 to 5 (the highest orders M and H of the polynomials a ⁺ (v) and a ^- (v) are 2.3 and 4.5 respectively), and determine the fitting polynomial coefficients α ₀ ,α ₁ ,...,α _M ,β ₀ ,β ₁ ,...,β _H and the highest orders M and H with the goal of minimizing the sum of squares of deviations L(α) and L(β);

Step 4.8. The fitted polynomials a ⁺ (v) and a ^- (v) are used as the positive and negative acceleration threshold curves, respectively, to determine whether the operating bus has abnormal speed change behavior.

The other steps and parameters are the same as those in Specific Embodiments 1 to 4.

Specific implementation method 6: This implementation method is different from one of the specific implementation methods 1 to 5 in that the method for determining the 95th percentile in step 4.4 is as shown in the following formula:

Where: Represents a collection /> Middle/> The acceleration value of the bit; Represents a collection /> Middle/> The acceleration value of the bit.

The other steps and parameters are the same as those in Specific Implementations 1 to 5.

Specific implementation method 7: This implementation method is different from any one of specific implementation methods 1 to 6 in that, in step 4.6, the least square method is used to fit the polynomial curve to the point sets P ⁺ and P ^- , respectively, with the velocity v as the independent variable and the acceleration a as the dependent variable, and the polynomial curve functions are set to be a ⁺ (v) and a ^- (v), respectively, as shown in the following formula;

Wherein: M, H represent the highest order of polynomial curve functions a ⁺ (v) and a ^- (v) respectively; α ₀ , α ₁ , ..., α _M represent fitting polynomial coefficients, β ₀ , β ₁ , ..., β _H represent fitting polynomial coefficients; v ^M represents the Mth order of velocity v, v ^H represents the Hth order of velocity v, v ^j represents the jth order of velocity v, and v ^l represents the lth order of velocity v.

The other steps and parameters are the same as those in Specific Embodiments 1 to 6.

Specific implementation eight: This implementation differs from any one of specific implementations one to seven in that the calculation of L(α) and L(β) in step 4.7 is as follows:

The other steps and parameters are the same as those in Specific Embodiments 1 to 7.

Specific implementation method 9: This implementation method is different from any one of specific implementation methods 1 to 8 in that the polynomials a ⁺ (v) and a ^- (v) fitted in step 4.8 are used as positive and negative acceleration threshold curves, respectively, to determine whether the operating passenger bus has abnormal speed change behavior; the specific process is:

Step 4.8.1. k = 1;

Step 4.8.2. Determine the acceleration of the kth data collection point after coordinate correction in shift i Corresponding speed/> The acceleration threshold With/> (That is, after a ⁺ (v) and a ^- (v) are fitted, each speed value in the acceleration threshold curve can correspond to an acceleration threshold).

Step 4.8.3. Judgment If yes, go to step 4.8.4; otherwise, go to step 4.8.5;

Step 4.8.4 Judgment If yes, it is determined that the operating bus has an abnormal acceleration behavior; otherwise, it is determined that the operating bus has no abnormal acceleration behavior;

Step 4.8.5. Judgment If yes, it is determined that the operating bus has an abnormal deceleration behavior; otherwise, it is determined that the operating bus has no abnormal deceleration behavior;

Step 4.8.6. k = k + 1, determine whether k ≤ K. If yes, proceed to step 4.8.2; otherwise, proceed to step 4.8.7;

Step 4.8.7. After traversing all accelerations, filter out all abnormal speed change behavior points.

The other steps and parameters are the same as those in Specific Embodiments 1 to 8.

Specific implementation method 10: This implementation method is different from any one of specific implementation methods 1 to 9 in that in step 5, the frequently occurring abnormal speed change section is identified based on the abnormal speed change behavior; the specific process is as follows:

The spatial distribution of abnormal speed change behavior is studied, the location where the abnormal speed change behavior occurs is recorded as an abnormal speed change behavior data, and the abnormal speed change behavior data is spatially clustered using the improved DBSCAN clustering method;

Step 5.1. Parameter definition; the specific process is:

Step 5.1.1. The coordinates of the location where a single abnormal speed change behavior occurs are r_p(x,y) and denoted as p ^ψ (x,y). The abnormal speed change behavior data form a set D with a total number of z, denoted as

Step 5.1.2. Calculate abnormal speed change behavior sample points using Euclidean distance With/> The distance between As shown in the following formula;

Where: γ,η∈[1,z];

They are sample points/> The x-axis coordinate of the plane coordinate, m;

They are sample points/> The y-axis coordinate of the plane coordinate, m;

Step 5.1.3. The neighborhood of the abnormal speed change behavior point p ^ψ is expressed as N _pψ , and the neighborhood radius is ε, as shown in the following formula;

Step 5.1.4. Neighborhood of point ^pψ The minimum number of sample points in the range is expressed as λ, and the conditions that point p ^ψ needs to meet as a core sample point are shown in the following formula;

Where: represents the number of sample points in the neighborhood with p ^ψ as the core point;

Step 5.1.5. Use the silhouette coefficient to evaluate the clustering effect of cluster samples and abnormal speed change behavior sample points The silhouette coefficient is calculated as follows:

Where: For sample points/> The average distance to other points in the cluster;

For sample points/> The average distance to points in other clusters;

Step 5.1.6. Use the average value of the overall silhouette coefficient As an evaluation index of the overall clustering effect of the sample, the specific calculation is shown in the following formula;

Step 5.2. Specific clustering steps; the specific process is:

Step 5.2.1. Initialize the neighborhood radius ε. According to the abnormal speed change behavior characteristics, it is not recommended to choose a larger ε, and ε = 100m;

Step 5.2.2. Determine the minimum number of sample points in a cluster according to the road grade. The minimum value is λ ₁ and the maximum value is λ ₂ (it can be flexibly adjusted according to the road grade, such as expressway, first-class highway, second-class highway, etc., and different minimum sample point value ranges can be defined), λ is an integer, and the value range is [λ ₁ ,λ ₂ ];

Step 5.2.3. Initialize, set parameter h = 0;

Step 5.2.4.λ＝λ ₁ +h, input ε, λ;

5.2.5. Use the DBSCAN clustering method to complete spatial clustering of all abnormal speed change behavior points p ^ψ (x, y) in set D;

Step 5.2.6. Calculate the contour coefficient of each abnormal speed change behavior sample point The average value of the overall silhouette coefficient

Step 5.2.7. Determine whether λ ₁ +h ≤ λ ₂ . If yes, h = h + 1, and return to step 5.2.4; otherwise, proceed to step 5.2.8;

Step 5.2.8. Calculate the maximum value of the overall silhouette coefficient As shown in the following formula;

Each λ as input corresponds to the calculation of an average value of the overall silhouette coefficient, and step 5.2.8 is to take the maximum value of these overall silhouette coefficients;

Step 5.2.9. Determine The corresponding parameter λ is λ ₀ , that is, the final key parameter λ＝λ ₀ ;

Step 5.3. Determine the clustering parameters of DBSCAN ε = 100m, λ = λ ₀ as the optimal clustering;

Step 5.4. After optimal clustering of all abnormal speed change behavior points p ^ψ (x, y) in set D, output the number of clusters and the number of sample points in each cluster. The number of clusters is the number of sections with frequent abnormal speed changes, and the number of sample points in each cluster is the number of abnormal speed change behaviors occurring in the section.

Step 5.5. The sample points are distributed on the road network, so the cluster shape after clustering (since all the sample points are on the road network, the cluster shape after clustering is the road section) fits the road network. The obtained cluster is the road section with frequent abnormal speed changes, and the edge points of the cluster along the road network direction are the starting and ending points of the road section with frequent abnormal speed changes.

The other steps and parameters are the same as those in Specific Embodiments 1 to 9.

The present invention may also have many other embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art may make various corresponding changes and modifications based on the present invention, but these corresponding changes and modifications should all fall within the scope of protection of the claims attached to the present invention.

Claims (9)

1. A method for identifying road sections where abnormal speed change behavior of operating passenger vehicles often occurs, characterized in that the specific process of the method is as follows: Step 1. Basic data collection; Step 2. Perform coordinate correction based on basic data; Step 3. Calculate the speed and acceleration based on the basic data and coordinate correction; Step 4. Identify abnormal speed change behavior based on speed and acceleration; Step 5. Identify the frequently abnormal speed change sections based on the abnormal speed change behavior; the specific process is as follows: Step 5.1. Parameter definition; The specific process is: Step 5.1.1. The coordinates of the location where a single abnormal speed change behavior occurs are r_p(x,y) and denoted as p ^ψ (x,y). The abnormal speed change behavior data form a set D with a total number of z, denoted as Step 5.1.2. Calculate abnormal speed change behavior sample points using Euclidean distance With/> The distance between As shown in the following formula (7); Where: γ,η∈[1,z]; They are sample points/> The x-axis coordinate of the plane coordinate, m; They are sample points/> The y-axis coordinate of the plane coordinate, m; Step 5.1.3. The neighborhood of the abnormal speed change behavior point p ^ψ is expressed as The neighborhood radius is ε, as shown in the following formula (8); Step 5.1.4. Neighborhood of point ^pψ The minimum number of sample points in the range is denoted as λ, and the conditions that point p ^ψ needs to satisfy as a core sample point are shown in the following formula (9); Where: represents the number of sample points in the neighborhood with p ^ψ as the core point; Step 5.1.5. Use the silhouette coefficient to evaluate the clustering effect of clustering samples and abnormal speed change behavior sample points The silhouette coefficient is calculated as shown in formula (10): Where: For sample points/> The average distance to other points in the cluster; For sample points/> The average distance to points in other clusters; Step 5.1.6. Use the average value of the overall silhouette coefficient As the evaluation index of the overall clustering effect of the sample, the specific calculation is shown in the following formula (11); Step 5.2. Specific clustering steps; The specific process is: Step 5.2.1. Initialize the neighborhood radius ε, set ε = 100m; Step 5.2.2. Determine the minimum number of sample points in a cluster according to the road grade. The minimum value is λ ₁ and the maximum value is λ ₂ . λ is an integer and its value range is [λ ₁ ,λ ₂ ]; Step 5.2.3. Initialize, set parameter h = 0; Step 5.2.4.λ＝λ ₁ +h, input ε, λ; Step 5.2.5. Use the DBSCAN clustering method to complete spatial clustering of all abnormal speed change behavior points p ^ψ (x, y) in set D; Step 5.2.6. Calculate the contour coefficient of each abnormal speed change behavior sample point and the average value of the overall silhouette coefficient/> Step 5.2.7. Determine whether λ ₁ +h ≤ λ ₂ . If yes, h = h + 1, and return to step 5.2.4; otherwise, proceed to step 5.2.8; Step 5.2.8. Calculate the maximum value of the overall silhouette coefficient As shown in the following formula (12); Step 5.2.9. Determine The corresponding parameter λ is λ ₀ , that is, the final key parameter λ＝λ ₀ ; Step 5.3. Determine the clustering parameters of DBSCAN ε = 100m, λ = λ ₀ as the optimal clustering; Step 5.4. After optimal clustering of all abnormal speed change behavior points p ^ψ (x, y) in set D, output the number of clusters and the number of sample points in each cluster. The number of clusters is the number of sections with frequent abnormal speed changes, and the number of sample points in each cluster is the number of abnormal speed change behaviors occurring in the section. Step 5.5. The sample points are distributed on the road network, so the shape of the cluster after clustering fits the road network. The obtained cluster is the section with frequent abnormal speed changes, and the edge points of the cluster along the road network direction are the starting and ending points of the section with frequent abnormal speed changes. 2. A method for identifying road sections where abnormal speed change behavior of operating passenger vehicles often occurs according to claim 1, characterized in that: the basic data collection in step 1; the specific process is: Step 1.1. Extract the GPS data of all running shifts of an operating bus line in the past N consecutive days, including the collection time, sampling interval, longitude and latitude of the GPS data, vehicle ID, and shift direction information; Step 1.2. The spatial distribution of the sections where the abnormal speed change behavior of the operating buses in the up and down directions often occurs is obviously different. The roads where the operating buses run in the up direction are selected as the research objects, and the same applies to the roads where the operating buses run in the down direction. Step 1.3. Define the set of operating shifts in the upward direction of the line as I = {i:i = 1, 2, 3, ..., I}; shift i has a total of K GPS data collection points, and the coordinates of the kth data collection point of shift i are expressed as Where lng and lat represent longitude and latitude respectively, in degrees; GPS sampling interval is T, in seconds; k = 1, 2, 3, ..., K; Step 1.4. Extract the GIS road network of the area through which the operating bus routes pass, including road names, road numbers, and road grade information. 3. A method for identifying road sections where abnormal speed change behavior of operating passenger vehicles often occurs according to claim 2, characterized in that: coordinate correction is performed based on basic data in step 2; the specific process is: Step 2.1. Use a map matching algorithm based on a hidden Markov model to complete the map matching of the GPS data collection point and the GIS road network. The matched road network is the actual road network where the GPS data collection point is located. Step 2.2. The coordinates of the GPS data collection point after map matching is the actual location of the GPS data collection point on the road network, expressed as Step 2.3. Coordinates after map matching Uniform conversion to plane coordinates/> x, y are the projected distances of the GPS data collection point from the central meridian and the equator in a coordinate system with the intersection of the equator and the central meridian as the origin, in meters. 4. A method for identifying road sections where abnormal speed change behavior of operating passenger vehicles often occurs according to claim 3, characterized in that: in step 3, the speed and acceleration are calculated based on basic data and coordinate correction; the specific process is: Step 3.1. Express the speed of the kth data collection point after coordinate correction in shift i as The unit is m/s; calculate the speed/> As shown in the following formula; Where: They are the x-axis coordinates of the plane coordinates of the k+1th and kth data collection points of shift i after the coordinates are corrected, in meters; They are the y-axis coordinates of the plane coordinates of the k+1th and kth data collection points of shift i after correction, in meters; Step 3.2. Express the acceleration of the kth data collection point after coordinate correction in shift i as The unit is m/s ² ; calculate the acceleration/> As shown in the following formula; Where: is the speed of the k-1th data collection point in shift i after coordinate correction. 5. A method for identifying road sections where abnormal speed change behavior of operating passenger vehicles frequently occurs according to claim 4, characterized in that: in step 4, abnormal speed change behavior is identified based on speed and acceleration; the specific process is: Step 4.1. Put all the acceleration data collected in I shifts that are greater than or equal to 0 into the set C ⁺ , Put all the acceleration data less than 0 in the data collected during I shifts into the set C ^- , The maximum value of the shift running speed is denoted as v _max ; Where: _vn is the velocity of the nth data point in the set C ⁺ , is the acceleration of the nth data point in the set C ⁺ v _m is the velocity of the mth data point in the set C ^- , /> is the acceleration of the mth data point in the set C ^- ; Step 4.2. Divide the speed interval [0, v _max ] into a set Q with an interval of 1 m/s. No. The interval is denoted as Q _σ , the first/> The midpoint of the interval is /> in, is the mathematical symbol for rounding up; Among the data collection points included in the interval Q _σ , there are data collection points belong to the set C ⁺ , /> The data collection points belong to the set C ^- ; Step 4.3. Set the interval The data collection points belonging to the sets C ⁺ and C ^- are arranged from small to large according to the acceleration values and put into the sets /> and/> Step 4.4. Set the The 95th percentile of the medium acceleration value is represented by/> Collection/> The 95th percentile of the medium acceleration value is represented by/> As the distinguishing value of abnormal speed change behavior in each speed range, only the behavior corresponding to 5% acceleration is regarded as extreme speed change behavior; Step 4.5. Express the positive acceleration and negative acceleration distinction value point set as P ⁺ , P ^- , as shown in the following formula; The acceleration threshold curves are drawn using the data points in the point sets P ⁺ and P ^-, respectively, with the horizontal axis representing the velocity and the vertical axis representing the acceleration; Step 4.6. Use the least squares method to fit the polynomial curve to the point sets P ⁺ and P ^- , respectively, with velocity v as the independent variable and acceleration a as the dependent variable. Let the polynomial curve functions be a ⁺ (v) and a ^- (v) respectively; Step 4.7. The polynomials a ⁺ (v) and a ^- (v) are fitted with 2 to 5 orders respectively, with the goal of minimizing the sum of squares of deviations L(α) and L(β), and the fitting polynomial coefficients α ₀ ,α ₁ ,...,α _M , β ₀ ,β ₁ ,...,β _H and the highest orders M and H are determined; Step 4.8. The fitted polynomials a ⁺ (v) and a ^- (v) are used as the positive and negative acceleration threshold curves, respectively, to determine whether the operating bus has abnormal speed change behavior. 6. A method for identifying road sections where abnormal speed change behavior of operating passenger vehicles often occurs according to claim 5, characterized in that: the method for determining the 95th percentile in step 4.4 is as shown in the following formula; Where: Represents a collection /> Middle/> The acceleration value of the bit; Represents a collection /> Middle/> The acceleration value of the bit. 7. A method for identifying road sections where abnormal speed change behavior of operating passenger vehicles often occurs according to claim 6, characterized in that: in said step 4.6, the least squares method is used to fit the polynomial curve to the point sets P ⁺ and P ^- , with speed v as the independent variable and acceleration a as the dependent variable, and the polynomial curve functions are set to be a ⁺ (v) and a ^- (v), respectively, as shown in the following formula; Wherein: M, H represent the highest order of polynomial curve functions a ⁺ (v) and a ^- (v) respectively; α ₀ , α ₁ , ..., α _M represent fitting polynomial coefficients, β ₀ , β ₁ , ..., β _H represent fitting polynomial coefficients; v ^M represents the Mth order of velocity v, v ^H represents the Hth order of velocity v, v ^j represents the jth order of velocity v, and v ^l represents the lth order of velocity v. 8. A method for identifying road sections where abnormal speed change behavior of operating passenger vehicles often occurs according to claim 7, characterized in that: L(α) and L(β) in step 4.7 are calculated as shown in the following formula; 9. A method for identifying road sections where abnormal speed change behavior of an operating passenger bus often occurs according to claim 8, characterized in that: the polynomials a ⁺ (v) and a ^- (v) fitted in step 4.8 are used as positive and negative acceleration threshold curves, respectively, to determine whether the operating passenger bus has abnormal speed change behavior; the specific process is: Step 4.8.1. k = 1; Step 4.8.2. Determine the acceleration of the kth data collection point after coordinate correction in shift i Corresponding speed/> The acceleration threshold With/> Step 4.8.3. Judgment If yes, go to step 4.8.4; otherwise, go to step 4.8.5; Step 4.8.4 Judgment If yes, it is determined that the operating bus has an abnormal acceleration behavior; otherwise, it is determined that the operating bus has no abnormal acceleration behavior; Step 4.8.5. Judgment If yes, it is determined that the operating bus has an abnormal deceleration behavior; otherwise, it is determined that the operating bus has no abnormal deceleration behavior; Step 4.8.6. k = k + 1, determine whether k ≤ K. If yes, proceed to step 4.8.2; otherwise, proceed to step 4.8.7; Step 4.8.7. After traversing all accelerations, filter out all abnormal speed change behavior points.