Urban road frequently-congested section identification method and device

Abstract

The invention discloses a method and a device for identifying frequently-congested sections of urban roads. The device comprises an RFID acquisition module, a counting module, a road network information acquisition module, a road network layer generation module, a control module and a display module. The RFID acquisition module is used for acquiring vehicle information on a road in real time; the counting module is used for counting the number q1 of the new energy automobiles, the number q2 of the trucks and the total passing number value q3 according to the vehicle information and cumulative summation; the road network information acquisition module is used for acquiring GIS road network signals; the road network layer generation module is used for generating a GIS road network layer according to the GIS road network signal; the control module is used for outputting a classification result of frequently-congested road sections according to the vehicle information and the GIS road network layer; and the display module is used for displaying the GIS road network map layer and the classification result of the frequently-congested road sections.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic management, and particularly to a method and device for identifying frequently congested sections of urban roads. Background Art

[0002] With the rapid development of social economy in recent years, the problem of urban road traffic congestion has become increasingly serious. The rapid development of the Intelligent Transport System (ITS) provides new ideas for urban congestion governance.

[0003] Among them, traffic big data mining is the core of high-quality ITS applications. For example, the high-precision prediction and identification of frequently congested sections are of great significance to urban traffic management departments and are the premise for traffic congestion governance, optimization plan formulation, and congestion prevention. Therefore, how to quickly identify frequently congested road sections has become a research hotspot in the industry in recent years.

[0004] The existing methods for identifying frequently congested road sections mainly focus on post-event monitoring. For example, indicators such as road running speed and traffic volume are used to evaluate whether a section is a frequently congested section. However, the post-event monitoring of frequently congested sections obviously cannot meet the needs of pre-event congestion governance, nor can it meet the requirements for analyzing the congestion mechanism of frequently congested sections. Therefore, it is of great significance to design an intelligent identification algorithm that can accurately predict frequently congested sections in advance.

[0005] At the same time, the existing influencing factors for identifying frequently congested sections mainly focus on the built environment, such as floor area ratio, road network density, etc. However, the number and spacing of parallel diversion channels and the proportion of different vehicle types are lacking, and the index range should also be combined with the current urban operation status. In addition, current ensemble learning models such as CatBoost need to set multiple hyperparameters, otherwise it will affect the classification accuracy of the model. Common particle swarm optimization algorithms (PSO) and others have problems such as insufficient convergence speed and being easily trapped in local optima, and need to be integrated and innovated to improve the prediction performance of the classification model. Summary of the Invention

[0006] In view of the technical problem of low accuracy in identifying frequently congested sections in the prior art, the present invention proposes a method and device for identifying frequently congested sections of urban roads.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] An apparatus for identifying frequently congested sections of urban roads includes an RFID acquisition module, a counting module, a road network information acquisition module, a road network layer generation module, a control module, and a display module;

[0009] The output end of the RFID acquisition module is connected to the input end of the counting module, the output end of the road network information acquisition module is connected to the input end of the road network layer generation module, the output ends of the counting module and the road network layer generation module are respectively connected to the input end of the control module, and the output end of the control module is connected to the input end of the display module;

[0010] The RFID acquisition module is used to collect vehicle information on the road in real time;

[0011] The counting module is used to accumulate and sum up the number of new energy vehicles q1, the number of trucks q2, and the total passing quantity value q3 through vehicle information;

[0012] The road network information acquisition module is used to collect GIS road network signals;

[0013] The road network layer generation module is used to generate a GIS road network layer according to the GIS road network signal;

[0014] The control module is used to output the classification result of frequently congested road sections according to vehicle information and the GIS road network layer;

[0015] The display module is used to display the GIS road network layer and the classification result of frequently congested road sections.

[0016] Preferably, the RFID acquisition module consists of an RFID receiver and an RFID transmitter. The RFID receiver is installed at the top of the road gantry, and the RFID transmitter is installed on the vehicle.

[0017] Preferably, the GIS road network layer contains road ID, longitude and latitude, and road name information.

[0018] Preferably, the GIS road network layer is a road network formed by curves. The curves are light gray, pink, blue, and red for branch roads, secondary arterial roads, arterial roads, and high-speed and expressways respectively.

[0019] Preferably, it further includes an alarm module connected to the control module, which is used to scroll and broadcast the road name field information of frequently congested road sections on the road.

[0020] The present invention also provides a method for identifying frequently congested road sections on urban roads, which specifically includes the following steps:

[0021] S1: Collect vehicle information and GIS road network signals on urban roads in real time to obtain sample data to be detected;

[0022] S2: Input the sample data to be detected into the trained CatBoost classifier and output the classification result;

[0023] S3: Flash the road according to the classification result.

[0024] Preferably, the S2 includes:

[0025] S2-1: Collect the historical vehicle information and historical GIS road network signals of urban roads from the server, and establish a training data set U1;

[0026] S2-2: Optimize the CatBoost classifier based on the polynomial delay particle swarm algorithm;

[0027] S2-3: Input the training data set U1 into the optimized CatBoost classifier for training;

[0028] S2-4: Input the sample data to be detected into the trained CatBoost classifier and output the classification result.

[0029] Preferably, in the S2-1, the training data set U1 includes road ID, road nature, new energy vehicle ratio, truck ratio, number of intersections, interchange ratio, area road network density, number of parallel channels, minimum distance between parallel channels, average plot ratio of the area, and frequently congested sections.

[0030] Preferably, in the S2-2, the optimization method of the CatBoost classifier is specifically as follows:

[0031] S2-2-1: Initialize the particle swarm: Randomly initialize the position and velocity of the particles. The particle dimension is 3, representing three hyperparameters: quantity, learning rate, and maximum depth of the tree;

[0032] S2-2-2: Update the velocity and position of the particles by using the current position, velocity, and historical optimal solution of the particles:

[0033] First, perform particle position update:

[0034]

[0035] In formula (1), v i,t+1 represents the velocity of the i-th particle in the (t + 1)-th iteration; v i,t represents the velocity of the i-th particle in the t-th iteration; ω represents the inertia factor; c1, c2, c3, c4 represent delay factors, c1 = c3, c2 = c4; r1, r2, r3, r4 represent random numbers between 0 and 1; p i,t represents the individual optimal solution of the i-th particle in the t-th iteration; p g,t represents the global optimal solution of the i-th particle in the t-th iteration; x i,t represents the position of the i-th particle in the t-th iteration; m l (ξ), m g (ξ) represents the intensity change factor in the hysteresis term, and ξ is based on the distance d between this particle and other particlesi Calibrate; a (τ) is a random 0 or 1; N is the upper limit of the delay times; τ represents the delay times; p i,t-τ represents the individual optimal solution of the i-th particle at the (t - τ)-th iteration; p g,t-τ represents the global optimal solution of the i-th particle at the (t - τ)-th iteration; x i,t+1 represents the position of the i-th particle at the (t + 1)-th iteration;

[0036] S2-2-3: The maximum update speed of the particle changes from the traditional fixed value to a non-linear decay curve:

[0037]

[0038] In formula (2), V iter is the maximum speed of the particle at the iter-th iteration; V max represents the specified maximum speed V min represents the specified minimum speed; iter max represents the maximum number of iterations;

[0039] S2-2-4: When the maximum number of iterations reaches 1000 times or the difference in the fitness of the particle before and after is less than 0.0005, terminate the iteration and return the optimal solutions of the three hyperparameters, thus completing the optimization of the CatBoost classifier.

[0040] Preferably, when the classification result is a frequently congested section, the road blinks; when the classification result is an infrequently congested section, the road does not blink.

[0041] In summary, due to the adoption of the above technical solutions, compared with the prior art, the present invention has at least the following beneficial effects:

[0042] The present invention provides an integrated learning recognition algorithm and device for frequently congested sections of urban roads. When the system is working, the RFID acquisition device collects vehicle information passing through the road in real time and sends it to the basic database. By cumulative summation, the number of new energy vehicles, the number of trucks, and the total number of passing vehicles q1, q2, and q3 are statistically obtained. At the same time, the unmanned aerial vehicle mapping device collects high-altitude monitoring image information, identifies the number of road intersections and the number of interchange intersections, and automatically generates a two-dimensional plane GIS road network signal, which is sent to the basic database. Finally, the basic database statistically obtains 11 field information including road ID, road nature, new energy vehicle ratio, truck ratio, number of intersections, interchange ratio, regional road network density, number of parallel channels, minimum distance between parallel channels, average plot ratio of the region, and frequently congested sections, and quantifies and scores each feature according to the rules. The 10 feature data of road nature to frequently congested sections in the basic database are used to establish a training dataset U1 and imported into the CatBoost classifier. The optimal values of the three hyperparameters of the CatBoost classifier, namely N_estimators, Learning_rate, and Max_Depth, are solved using the polynomial delay particle swarm optimization algorithm (PDPSO), and the CatBoost classifier model is trained. Finally, the sample data xi to be detected is imported into the trained CatBoost classifier, the classification result yi is output, and the numerical value ("1" or "0") of the classification result is transmitted to the basic database, and then dynamically displayed by the display device, realizing high-precision recognition of frequently congested sections of urban roads, providing a basis for subsequent analysis of the causes of congestion, and effectively guiding the optimization of subsequent engineering measures. Brief Description of the Drawings

[0043] Figure 1 Schematic diagram of an urban road frequently congested section recognition device according to an exemplary embodiment of the present invention.

[0044] Figure 2 Schematic diagram of an urban road frequently congested section recognition method according to an exemplary embodiment of the present invention. Detailed Embodiments

[0045] The present invention will be further described in detail below in conjunction with embodiments and specific implementation manners. However, this should not be construed as limiting the scope of the above-mentioned subject matter of the present invention to the following embodiments. Any technology implemented based on the content of the present invention belongs to the scope of the present invention.

[0046] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "longitudinal", "lateral", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on the present invention.

[0047] As Figure 1 shown, the present invention provides an identification device for frequently congested sections of urban roads, which includes an RFID acquisition module, a counting module, a road network information acquisition module, a road network layer generation module, a control module, and a display module.

[0048] The output end of the RFID acquisition module is connected to the input end of the counting module, the output end of the road network information acquisition module is connected to the input end of the road network layer generation module, the output ends of the counting module and the road network layer generation module are respectively connected to the input end of the control module, and the output end of the control module is connected to the input end of the display module.

[0049] In this embodiment, the RFID acquisition module is used to collect vehicle information on the road in real time; the RFID acquisition module is composed of an RFID receiver and an RFID transmitter. The RFID receiver is installed at the top of the road gantry, and the RFID transmitter is installed on the vehicle. When the vehicle equipped with the RFID transmitter passes through the road gantry, the RFID receiver receives the signal of the RFID transmitter, thereby completing the acquisition of vehicle information.

[0050] In this embodiment, the counting module is used to accumulate and sum up the number of new energy vehicles q1, the number of trucks q2, and the total number of passing vehicles q3 (including new energy vehicles, trucks, fuel vehicles, etc.) through vehicle information.

[0051] In this embodiment, the road network information acquisition module can use an unmanned aerial vehicle to collect GIS road network signals, including the number of intersections, the number of interchange intersections, etc.

[0052] In this embodiment, the road network layer generation module is a GIS layer, which is used to generate a GIS road network layer according to the GIS road network signals, including road ID, longitude and latitude, and road name information. The road ID is a unique coding ID.

[0053] In this embodiment, the control module is used to output the classification result of frequently congested sections according to vehicle information and the GIS road network layer.

[0054] In this embodiment, the display module is used to display the GIS road network layer and the classification result of frequently congested sections.

[0055] Among them, the GIS road network layer is a road network formed by curves. The curves are light gray, pink, blue, and red for branch roads, secondary arterial roads, arterial roads, and expressways / arterial roads, respectively. When it is a frequently congested section, the road blinks; when it is an infrequently congested section, the road does not blink. Clicking on the road automatically displays the road ID, longitude and latitude, and road name information, and you can choose whether to display the information fields of the basic database associated with the road ID.

[0056] In this embodiment, an alarm module connected to the control module is further included. The alarm module can use a high - pitched horn to scroll - broadcast the road name field information of frequently congested sections on the road.

[0057] Based on the above device, as Figure 2 shown, the present invention proposes a method for identifying frequently congested sections of urban roads, which specifically includes the following steps:

[0058] S1: Real - time collect the vehicle information and GIS road network signals of urban roads to obtain the sample data to be detected.

[0059] S2: Input the sample data to be detected into the trained CatBoost classifier and output the classification result.

[0060] S2 - 1: Collect the historical vehicle information and historical GIS road network signals of urban roads from the server to establish a training data set U1.

[0061] In this embodiment, characteristic parameters such as road ID, road nature, proportion of new - energy vehicles, proportion of trucks, number of intersections, proportion of interchange overpasses, road network density of the area, number of parallel channels, minimum distance between parallel channels, average plot ratio of the area, and frequently congested sections can be obtained from the historical vehicle information and historical GIS road network signals. The road nature, proportion of new - energy vehicles, proportion of trucks, number of intersections, proportion of interchange overpasses, road network density of the area, number of parallel channels, minimum distance between parallel channels, and average plot ratio of the area are used as input features, and the frequently congested sections are used as output features.

[0062] In this embodiment, the road nature includes 4 categories: branch roads, secondary arterial roads, arterial roads, and expressways / arterial roads, and the quantitative scores are 1, 2, 3, and 4 respectively;

[0063] The proportion of new - energy vehicles \(a = q1 / q3\), \(a\in(0,5\%]\), \(a\in(5\%,10\%]\), \(a\in(10\%,15\%]\), \(a > 15\%\), and the quantitative scores are 1, 2, 3, and 4 respectively;

[0064] The proportion of trucks \(b = q2 / q3\), \(b\in(0,5\%]\), \(b\in(5\%,10\%]\), \(b\in(10\%,15\%]\), \(b > 15\%\), and the quantitative scores are 1, 2, 3, and 4 respectively;

[0065] The number of intersections c is the sum of the number of signal-controlled intersections and non-signal-controlled intersections on the road. c ∈ (0, 2], c ∈ (2, 4], c ∈ (4%, 6%], c > 6%, and the quantitative scores are 1, 2, 3, and 4 respectively;

[0066] The interchange ratio d is the ratio of the number of interchanges on the road to the number of intersections c. d ∈ (0, 15%], b ∈ (15%, 30%], b ∈ (30%, 45%], b > 45%, and the quantitative scores are 1, 2, 3, and 4 respectively;

[0067] The road network density α4 of the area = n r / (πr 2 ), where n r is the total road mileage within a range with the center point of the road section as the center and a radius of r. r = 1 km, α4 ∈ (0, 4], α4 ∈ (4, 6], α4 ∈ (6, 8], α4 > 8, and the quantitative scores are 1, 2, 3, and 4 respectively;

[0068] The number of parallel channels a5 is the total number of other road sections in the same direction as the road section within a statistical radius of r = 1 km with the center point of the road section as the center, but the road grade should be the secondary grade or above of the road section. a5 ∈ (0, 1], a5 ∈ (1, 2], a5 ∈ (2, 3], a5 > 3, and the quantitative scores are 1, 2, 3, and 4 respectively;

[0069] The minimum spacing a6 of the parallel channels is the closest interval distance among all the parallel channels of the road section. a6 ∈ (0, 0.3], a6 ∈ (0.3, 0.5], a6 ∈ (0.5, 0.8], a6 > 0.8, and the quantitative scores are 1, 2, 3, and 4 respectively;

[0070] The average plot ratio α7 of the area = s d / (πr 2 ), s d is the total building area within a statistical radius of r with the center point of the road section as the center; when a7 ∈ (0, 0.8], a7 ∈ (0.8, 1.6], a7 ∈ (1.6, 2.4], a7 > 2.4, the quantitative scores are 1, 2, 3, and 4 respectively;

[0071] The frequently congested road sections are manually calibrated according to the frequently congested road sections in the urban traffic operation annual report, where 1 represents non-frequently congested road sections and 0 represents frequently congested road sections.

[0072] S2-2: Optimize the CatBoost classifier based on the polynomial delay particle swarm optimization algorithm (PDPSO).

[0073] S2-2-1: Initialize the particle swarm: Randomly initialize the positions and velocities of the particles. In the present invention, the particle dimension is 3, which respectively represent three hyperparameters: the number of estimators (N_estimators), the learning rate (Learning_rate), and the maximum depth of the tree (Max_Depth). Among them, the position represents the solution vector, and the velocity represents the direction and speed.

[0074] S2-2-2: In this embodiment, the velocities and positions of the particles are updated by using the current positions, velocities, and historical optimal solutions of the particles. The update formulas are as follows:

[0075]

[0076] In formula (1), v i,t+1 represents the velocity of the i-th particle at the (t + 1)-th iteration; v i,t represents the velocity of the i-th particle at the t-th iteration; ω represents the inertia factor, reflecting the influence degree of the previous velocity on the current velocity; c1, c2, c3, c4 represent the delay factors, where c1 = c3 and c2 = c4; r1, r2, r3, r4 represent random numbers between 0 and 1; p i,t represents the personal best solution of the i-th particle at the t-th iteration; p g,t represents the global best solution of the i-th particle at the t-th iteration; x i,t represents the position of the i-th particle at the t-th iteration; m l (ξ), m g (ξ) represent the intensity change factors in the hysteresis term, and ξ is calibrated according to the distance d i between this particle and other particles; a (τ) is a random 0 or 1; N is the upper limit of the delay times; τ represents the delay times; p i,t-τ represents the personal best solution of the i-th particle at the (t - τ)-th iteration; p g,t-τ represents the global best solution of the i-th particle at the (t - τ)-th iteration; x i,t+1 represents the position of the i-th particle at the (t + 1)-th iteration;

[0077] Among them, ξ is calibrated according to the distance d i between this particle and other particles,

[0078]

[0079] In formula (2), d i represents the distance between the i-th particle and other particles; M is the size of the particle swarm, D is the dimension of the particle, x j,t represents the position of the j-th particle at the t-th iteration; d g is the global best particle in d i ; dmax is the maximum value among d i ; d min is the minimum value among d i ; I f represents the evolution factor;

[0080]

[0081] When 0 ≤ I f ≤ 0.25, the particle is in the convergence state at this time, and m l (ξ), m g (ξ) are both equal to 0. When 0.25 ≤ I f ≤ 0.75, the particle is in the search state at this time, and m l (ξ) is equal to m g (ξ) is equal to When 0.75 ≤ I f ≤ 1, the particle is in the jump state at this time, and m l (ξ), m g (ξ) are both equal to 0.01;

[0082] In this embodiment, the fitness of each particle can be compared with its historical optimal solution and updated. The individual optimal solution is the optimal solution found by the particle itself, and the global optimal solution is the optimal solution among all particles.

[0083] S2-2-3: At the same time, the maximum update speed of the particle changes from the traditional fixed value to a non-linear decay curve:

[0084]

[0085] In formula (3), V iter is the maximum speed of the particle at the iter-th iteration; V max represents the specified initial maximum speed; Vx in represents the specified initial minimum speed; iter max represents the maximum number of iterations.

[0086] S2-2-4: Terminate the iteration when the maximum number of iterations reaches 1000 times or the difference between the particle fitness before and after is less than 0.0005, and return the optimal solution: Return the global optimal solution or the individual optimal solution as the final result, which is the optimal solution of the hyperparameters N_estimators, Learning_rate, and Max_Depth.

[0087] In this embodiment, the optimal values of three hyperparameters, namely the number of decision trees (N_estimators), learning rate (Learning_rate), and maximum depth of the tree (Max_Depth) in the CatBoost classifier, are solved based on the polynomial delay particle swarm optimization (PDPSO) algorithm, thereby completing the optimization of the CatBoost classifier.

[0088] S2-3: Input the training data set U1 into the optimized CatBoost classifier for training.

[0089] S2-4: Input the sample data to be detected into the trained CatBoost classifier and output the classification result.

[0090] In this embodiment, the classification results include highly congested roads (1) and frequently congested roads (0).

[0091] S3: Flash the road according to the classification result to remind people to pay attention.

[0092] When it is a frequently congested road, the road flashes; when it is a highly congested road, the road does not flash; when the road does not flash; click on the road to automatically display the road ID, longitude and latitude, and road name information, and you can choose whether to display the information fields of the basic database associated with the road ID.

[0093] Those of ordinary skill in the art can understand that the above embodiments are specific embodiments for implementing the present invention, and in practical applications, various changes can be made in form and details without departing from the spirit and scope of the present invention.

Claims

1. A device for identifying frequently congested sections of urban roads, characterized in that: It includes RFID acquisition module, counting module, road network information acquisition module, road network layer generation module, control module and display module; The output end of the RFID acquisition module is connected to the input end of the counting module, the output end of the road network information acquisition module is connected to the input end of the road network layer generation module, the output end of the counting module and the output end of the road network layer generation module are respectively connected to the input end of the control module, and the output end of the control module is connected to the input end of the display module; RFID collection module, used to collect vehicle information on the road in real time; The counting module is used to calculate the number of new energy vehicles q1, the number of trucks q2, and the total number of vehicles passing q3 through the vehicle information; Road network information collection module, used to collect GIS road network signals; A road network layer generation module is used to generate a GIS road network layer according to GIS road network signals; A control module is used to output classification results of frequently congested road sections based on vehicle information and GIS road network layers; The display module is used to display the GIS road network layer and the classification results of frequently congested road sections.

2. The device for identifying frequently congested sections of urban roads according to claim 1, characterized in that: The RFID acquisition module consists of an RFID receiver and an RFID transmitter. The RFID receiver is installed on the top of the road gantry, and the RFID transmitter is installed on the vehicle.

3. The device for identifying frequently congested sections of urban roads according to claim 1, characterized in that: The GIS road network layer contains road ID, longitude and latitude, and road name information.

4. The device for identifying frequently congested sections of urban roads according to claim 1, characterized in that: The GIS road network layer is a road network formed by curves, and the curves are light gray, pink, blue, and red according to branch roads, secondary roads, main roads, and expressways.

5. The device for identifying frequently congested sections of urban roads according to claim 1, characterized in that: It also includes an alarm module connected to the control module, which is used to scroll and broadcast the road name field information of the frequently congested sections on the road.

6. A method for identifying frequently congested sections of urban roads based on the device according to any one of claims 1 to 5, characterized in that: The specific steps include: S1: Collect vehicle information and GIS road network signals on urban roads in real time to obtain sample data to be tested; S2: Input the sample data to be tested into the trained CatBoost classifier and output the classification result; S3: Perform road flashing according to the classification results.

7. The method for identifying frequently congested sections of urban roads according to claim 6, characterized in that: The S2 includes: S2-1: Collect historical vehicle information and historical GIS road network signals of urban roads from the server to establish a training data set U1; S2-2: Optimize CatBoost classifier based on polynomial delayed particle swarm algorithm; S2-3: Input the training data set U1 into the optimized CatBoost classifier for training; S2-4: Input the sample data to be tested into the trained CatBoost classifier and output the classification result.

8. The method for identifying frequently congested sections of urban roads as claimed in claim 7, characterized in that: In S2-1, the training data set U1 includes road ID, road nature, proportion of new energy vehicles, proportion of trucks, number of intersections, proportion of interchanges, area road network density, number of parallel channels, minimum spacing between parallel channels, average area floor area ratio, and frequently congested sections.

9. The method for identifying frequently congested sections of urban roads according to claim 7, characterized in that: In S2-2, the optimization method of the CatBoost classifier is specifically as follows: S2-2-1: Initialize the particle swarm: Randomly initialize the position and speed of the particles. The particle dimension is 3, which represent the three hyperparameters of quantity, learning rate and maximum depth of the tree respectively. S2-2-2: Update the particle's speed and position by using the particle's current position, speed and historical optimal solution: First, update the particle position: In formula (1), v i,t+1 represents the velocity of the i-th particle in the t+1 iteration; v i,t represents the speed of the ith particle in t iterations; ω represents the inertia factor; c1, c2, c3, c4 represent delay factors, c1=c3, c2=c4; r1, r2, r3, r4 represent random numbers between 0 and 1; p i,t represents the individual optimal solution of the i-th particle in t iterations; p g,t represents the global optimal solution of the ith particle in t iterations; x i,t represents the position of the i-th particle in t iterations; m l (ξ), m g (ξ) represents the intensity variation factor in the hysteresis term, which is determined by the distance d between the particle and other particles. i Calibrate; a(τ) is a random 0 or 1; N is the upper limit of delay times; τ represents the delay times; p i,t-τ represents the individual optimal solution of the i-th particle in the t-τ iteration number; p g,t-τ represents the global optimal solution of the i-th particle in the t-τ iteration number; x i,t+1 represents the position of the i-th particle in the t+1 iteration; S2-2-3: The maximum particle update speed changes from the traditional fixed value to a nonlinear attenuation curve: In formula (2), V iter is the maximum velocity of the particle at the iterth iteration; V max Indicates the specified maximum speed V min Indicates the specified minimum speed; iter max Indicates the maximum number of iterations; S2-2-4: When the maximum number of iterations reaches 1000 or the difference between the particle fitness before and after is less than 0.0005, the iteration is terminated and the optimal solution of the three hyperparameters is returned, thus completing the optimization of the CatBoost classifier.

10. The method for identifying frequently congested sections of urban roads according to claim 6, characterized in that: In S3, when the classification result is a frequently congested road section, the road flashes; when the classification result is an extremely congested road section, the road does not flash.

Images