Improved McMaster algorithm for intersection congestion discrimination

Abstract

The invention relates to the field of traffic safety travel, and discloses a method for detecting intersection congestion according to millimeter wave radar detection data and an improved McMaster algorithm. In recent years, with the increase of travel demands of people and the increase of automobile ownership, urban road congestion is more serious. At present, urban roads are seriously congested, and how to improve an intelligent traffic management system based on a travel plan and relieve traffic pressure is a target at the present stage. In order to solve the problem of urban congestion, the invention aims to analyze the correlation of traffic flow parameters after collecting the traffic flow parameters by using a millimeter-wave radar, proposes a McMaster algorithm based on an adaptive particle swarm, and further optimizes the correlation parameters, thereby assisting to judge the traffic congestion condition and improving the vehicle running state efficiency. And the information is timely fed back to a traffic management department for rapid processing of emergencies, and corresponding congestion evacuation means are adopted to relieve congestion conditions, so that congestion prediction and rapid congestion response based on travel plans are realized.

Description

Technical Field

[0001] This invention relates to the field of traffic safety and travel, and discloses a method for detecting intersection congestion based on millimeter-wave radar detection data and an improved McMaster algorithm. Background Technology

[0002] In recent years, with the increasing travel demands and rapid growth in car ownership, urban road congestion has become increasingly severe. Simultaneously, intelligent transportation is constantly developing, and autonomous driving technology is maturing. Achieving transparency of the road ahead, improving intelligent traffic management systems, and alleviating traffic pressure have become primary tasks for transportation development. To achieve road transparency, facilitate the integration of autonomous vehicles into people's lives, and provide opportunities for human-machine collaboration, the research team proposed a travel planning mechanism. This mechanism involves travelers publishing their travel time, departure point, and destination before setting off, aiming to build an intelligent traffic management system based on travel planning. Currently, urban road congestion is severe; therefore, improving the intelligent traffic management system based on travel planning and alleviating traffic pressure are current goals.

[0003] Millimeter-wave radar includes pulse-Doppler and frequency-modulated continuous wave (FMCW) technologies. Pulse-Doppler is a traditional radar detection method. It transmits short pulses over a short period and listens for the echo during the pulse repetition intervals. Its main advantage lies in its high peak power, enabling the radar to effectively detect small-amplitude moving targets against strong clutter backgrounds. However, a drawback of pulse-Doppler radar is the inherent contradiction between its range and velocity resolution; increasing range resolution leads to a decrease in velocity resolution, and vice versa. Frequency-modulated continuous wave (FMCW) is a continuous-wave radar detection method. In this technology, the frequency of the millimeter-wave signal transmitted by the radar changes linearly with time, forming a so-called "chirped" signal. When these signals encounter a target and are reflected back, there is a small time difference between the received echo and the transmitted signal. This time difference can be used to calculate the target's range and relative velocity. FMCW radar's advantages lie in its high range and velocity resolution, and it does not require the complex signal processing in time and frequency required by pulse-Doppler radar.

[0004] The McMaster algorithm is a widely used traditional single-section detection algorithm. It uses data from a single vehicle detector to construct a traffic occupancy template and detects traffic congestion by comparing actual data with the template. In the late 20th century, Deng Wei established a model for identifying frequent and occasional traffic congestion using catastrophe theory. This model relies on experience to build a decision tree model, calculates the sensitivity of actual traffic parameters to changes, and compares it with a decision threshold. If the threshold is exceeded, occasional traffic congestion is identified. In the early 21st century, Zhang Cheng adopted the McMaster algorithm, abandoning the method of parameters being determined by expert experience. He established an objective function for the McMaster algorithm parameters and data classification accuracy, used particle swarm optimization, and established the McMaster algorithm model based on the speed mutation type using cusp catastrophe theory. The McMaster algorithm can not only identify congestion types but also determine whether a road segment is congested. Its computational efficiency and accuracy are unparalleled compared to other algorithms. However, since the template construction parameters need to be set by relevant experts based on experience, they are greatly affected by subjectivity, resulting in a low detection rate of traffic congestion by the McMaster algorithm. Furthermore, different vehicle detectors often have different parameters, which need to be reset.

[0005] To address the aforementioned problems, the present invention aims to collect traffic flow parameters using millimeter-wave radar, analyze their correlations, propose an adaptive particle swarm optimization-based McMaster algorithm, further optimize relevant parameters, thereby assisting in the identification of traffic congestion and improving vehicle operating efficiency. This allows for timely feedback to traffic management departments for rapid handling of emergencies and the implementation of corresponding congestion mitigation measures, achieving congestion prediction and rapid congestion response based on travel plans. Summary of the Invention

[0006] In view of this, the purpose of this invention is to provide a three-dimensional McMaster method for detecting traffic congestion in highway cross-sections.

[0007] The objective of this invention is achieved through the following technical solution:

[0008] An adaptive particle swarm optimization McMaster algorithm for millimeter-wave radar detection of intersection congestion detection includes the following steps:

[0009] Step 1: Extract traffic parameters based on millimeter-wave radar data, including vehicle flow, speed, and time occupancy.

[0010] Step 2: Determine the relationship surface between occasional and frequent congestion data in the three-dimensional variable parameter space of traffic flow-occupancy-time occupancy;

[0011] Step 3: Analyze the relationship between occasional and frequent congestion data in the three-dimensional variable parameter space of traffic flow-occupancy-time occupancy;

[0012] Step 4: Establish and train the parameters of the McMaster algorithm model, including the minimum uncongested data boundary (LUD), critical occupancy rate (Ocrit), and critical flow rate (Vcrit). The specific method is as follows:

[0013] 1) Construct an adaptive particle swarm optimization (McMaster) algorithm model.

[0014] 2) Train the McMaster algorithm model using the relational dataset analyzed in step 3. The parameters LUD, Ocrit, and Vcrit divide the traffic flow-occupancy relationship graph into four regions: the road is in a smooth state, the downstream has frequent congestion, the downstream road is in a slow state, and the downstream has occasional traffic congestion. Based on the three parameters, propose the objective function that maximizes the congestion discrimination accuracy.

[0015] Step 5: Transform the coordinates of the relationship diagram in Step 2, and substitute the original traffic flow parameter data into the model.

[0016] Step 6: Detect traffic congestion at intersections based on the real-time vehicle movement status monitored by millimeter-wave radar using the trained model.

[0017] Furthermore, the specific process of step one is as follows:

[0018] 1) Based on the vehicle status data detected by the millimeter-wave radar detector, detect the traffic parameters of each lane of the selected experimental road;

[0019] 2) The calculation formula for traffic flow parameters within a cycle detected by the millimeter-wave radar detector is as follows:

[0020] Q = VK

[0021] K = N / L

[0022] O = (T / T total) × 100%

[0023] In the formula, speed V, flow rate Q, and density K, L is the number of vehicles in the road segment, Kj is the congestion density, V0 is the average driving speed when the speed is 0, and O is the time occupancy rate.

[0024] Furthermore, the relationship surface under the specific occasional and frequent congestion data in step two is as follows:

[0025] Based on the different causes of congestion, it is divided into frequent congestion and occasional congestion. It analyzes the spatial distribution characteristics and relationships of traffic flow (vehicle volume), average speed and time occupancy under different congestion conditions.

[0026] Furthermore, the specific content of step four is as follows:

[0027] If the data falls above LUD, the road segment is in a smooth state; otherwise, it is in a congested state.

[0028] If the data falls below LUD, it is compared with the parameter variables Qcrit and Vcrit to determine the congestion type.

[0029] If the data falls on the LUD line, then take a line parallel to the x-axis and y-axis. The Vcrit line parallel to the x-axis distinguishes between frequent traffic congestion and slow traffic flow, and the Ocrit line parallel to the y-axis distinguishes between occasional traffic congestion and slow traffic flow.

[0030] Furthermore, the specific process of step four, process 1), the adaptive particle swarm optimization McMaster algorithm model, is as follows: The expression is E(v, o, q) = av4 + bov 2 +cqv

[0031] In the formula, a, b, and c are parameters. Once these three parameters are confirmed, the model can be used to describe changes in traffic flow.

[0032] Furthermore, the calculation process of the objective function for maximizing accuracy in step four, process 2) is as follows:

[0033] The objective function for maximizing LUD congestion discrimination accuracy is expressed as:

[0034] argmaxN1=∑[k i o data / q data ]∩C data

[0035] The formula argmax is a function that evaluates the set of parameters of a function. argmaxN1 represents the value of parameter set k when N reaches its maximum value; N1 represents the maximum number of correct classifications in the McMaster algorithm for both congested and uncongested scenarios; N represents the total number of actual data points; ki represents the slope of the line connecting particle Pi in the particle swarm optimization algorithm to the origin of the flow-occupancy graph; Odata and qdata represent the actual occupancy and flow data; Cdata represents the actual state of traffic flow, i.e., congested or uncongested; [k i o data / q data It can be determined whether the traffic flow data is located above or below the LUD line [k] i o data / q data ]∩C dataThe intersection of the predicted and actual results is taken; if the results are the same, the value is incremented by 1; otherwise, it remains unchanged. The optimal LUD line is obtained when all particles in the particle swarm have traversed all the data and found the maximum number of correct classifications, N1.

[0036] The objective function for maximizing the congestion detection accuracy of Ocrit and Vcrit is expressed as:

[0037]

[0038] In the formula: N2—the sum of the maximum number of correct classifications by the McMaster algorithm for both occasional and frequent congestion; Vcrit i Represents particle p i Draw a line parallel to the x-axis; Ocrit i Represents particle p i Draw a line parallel to the y-axis; [Vcrit q] i data ] indicates that the true data is located in Vcrit i Above or below; [Ocrit o i data [] indicates that the true data is located in Ocrit i Left or right side; R data Represents the state of traffic congestion, namely, frequent congestion, occasional congestion, and slow traffic flow; [Vcrit q R] i data data [Ocrit o R] i data data The symbol ∩ represents the intersection of the predicted and actual results; if the results are the same, it increments by 1; otherwise, it remains unchanged. The optimal Ocrit and Vcrit are obtained when all particles in the particle swarm have traversed all the data and the maximum number of correct classifications N² is found.

[0039] Furthermore, the coordinate transformation calculation process in step five is as follows:

[0040] In the adaptive particle swarm optimization algorithm, the McMaster algorithm coordinates and the traditional McMaster 3D coordinates have an angle in the control plane, requiring a coordinate transformation. First, the coordinate system needs to be translated, moving the origin of the improved McMaster theoretical coordinate system to the origin of the standard coordinate system. The origin of the improved McMaster theoretical coordinate system coincides with the projection of the data bifurcation point onto the control plane, and the projection of the data bifurcation point onto the control plane coincides with the point (Vcrit, Ocrit) in the McMaster algorithm. Therefore, the coordinate position of the origin of the mutation theoretical coordinate system in the standard coordinate system is (Vcrit, Ocrit, 0). The transformation is expressed as:

[0041] In the formula: v, o, and q are the original traffic flow parameters of claim 2; v', o', and q' represent the translated data; Octit and Vcrit represent the optimal parameters obtained by the adaptive particle swarm optimization algorithm. Next, the coordinate axes are rotated. Rotating the conventional coordinate axes to coincide with the new coordinate axes allows the q-axis to become the dividing line between congested and uncongested states. The positive half of the q-axis represents congested states, and the negative half represents uncongested states. This transformation can be expressed as:

[0042] In the equations V, O, and Q, the final traffic flow parameters are obtained after coordinate transformation; θ represents the angle between the conventional coordinate axis and the cusp catastrophe theory coordinate axis; Qm represents the maximum flow rate in the data; Om represents the occupancy rate when the flow rate is maximum; and KLUD represents the optimal parameters obtained by the adaptive particle swarm optimization algorithm in the McMaster algorithm.

[0043] Furthermore, the model calculation process in step five is as follows:

[0044] By taking the second partial derivative function with respect to V in the McMaster adaptive particle swarm optimization model expression, and using the cubic equation discriminant formula, we can determine when a sudden change in traffic flow occurs. The graphical expression is as follows: Δ=27a 2 c 2 Q 2 +8b 3 O 3

[0045] Furthermore, the discrimination process in step six is ​​as follows:

[0046] When Δ < 0, the changes in the state variables are similar to those during sporadic traffic congestion; when Δ > 0, the changes in the state variables are similar to those during frequent traffic congestion. In this bifurcation set, we can determine whether a sudden change has occurred in the state variables by judging whether the state variables pass through the Δ < 0 portion. Similarly, in traffic flow systems, we can determine whether a sudden change has occurred in traffic flow parameters, i.e., whether sporadic traffic congestion has occurred, by judging whether the traffic flow data passes through the Δ < 0 portion.

[0047] Due to the adoption of the above technical solution, the present invention has the following beneficial effects:

[0048] The adaptive particle swarm optimization improved 3D McMaster method for detecting traffic congestion at intersections provided by this invention fully considers the impact of speed parameters on flow rate and occupancy, analyzes the correlation between flow rate, speed, and occupancy, and improves the discrimination efficiency and accuracy of the 3D McMaster algorithm after improving the coordinates, thereby providing better decision-making.

[0049] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0050] The accompanying drawings of this invention are described below.

[0051] Figure 1 This is a flowchart illustrating the workflow for congestion identification and type determination based on travel plans;

[0052] Figure 2 This is a schematic diagram illustrating the congestion path.

[0053] Figure 3 This is a diagram illustrating the division of different parameters;

[0054] Figure 4 This is a diagram illustrating traffic flow branching and congestion identification. Detailed Implementation

[0055] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0056] Example 1

[0057] An adaptive particle swarm optimization McMaster algorithm for millimeter-wave radar detection of intersection congestion detection is characterized by the following steps:

[0058] Step 1: Extract traffic parameters based on millimeter-wave radar data, including vehicle flow, speed, and time occupancy.

[0059] 1) Based on the vehicle status data detected by the millimeter-wave radar detector, detect the traffic parameters of each lane of the selected experimental road;

[0060] 2) The formula for calculating the total traffic flow parameters detected by the millimeter-wave radar detector within a given period is as follows:

[0061] Q = VK

[0062] K = N / L

[0063] Q = K j (VV 2 / V0)

[0064] O = (T / T total) × 100%

[0065] In the formula, speed V, flow rate Q, and density K, L is the number of vehicles in the road segment, Kj is the congestion density, V0 is the average driving speed when the speed is 0, and O is the time occupancy rate.

[0066] Step two is to divide congestion into frequent congestion and occasional congestion based on the different causes of congestion, and analyze the spatial distribution characteristics and relationships of traffic flow (vehicle volume), average speed and time occupancy under different congestion conditions.

[0067] Step 3 is to establish a three-dimensional McMaster algorithm model of the relationship between occasional and frequent congestion data in the three-dimensional variable parameter space of traffic flow-occupancy-time occupancy.

[0068] Step 4: Train the parameters of the McMaster algorithm model. The specific method is as follows:

[0069] 1) such as Figure 3 As shown, if the data falls above the LUD (Local Area Diagram), the road segment is in a smooth state; otherwise, it is in a congested state. If the data falls below the LUD, it is compared with the parameter variables Qcrit and Vcrit to determine the congestion type. If the data falls on the LUD line, parallel lines parallel to the x-axis and y-axis are taken. Vcrit, parallel to the x-axis, distinguishes between frequent traffic congestion and slow traffic flow, while Ocrit, parallel to the y-axis, distinguishes between occasional traffic congestion and slow traffic flow.

[0070] 2) The theoretical model equation for the calculation process is as follows:

[0071] E(v, o, q) = av 4 +bov 2 +cqv

[0072] Where a, b, and c are parameters, once these three parameters are confirmed, the model can be used to depict changes in traffic flow.

[0073] Construct the objective function that maximizes the accuracy of LUD congestion detection, expressed by the formula:

[0074] argmaxN1=∑[k i o data / q data ]∩C data

[0075] The formula argmax is a function that evaluates the set of parameters of a function. argmaxN1 represents the value of parameter set k when N reaches its maximum value; N1 represents the maximum number of correct classifications in the McMaster algorithm for both congested and uncongested scenarios; N represents the total number of actual data points; ki represents the slope of the line connecting particle Pi in the particle swarm optimization algorithm to the origin of the flow-occupancy graph; Odata and qdata represent the actual occupancy and flow data; Cdata represents the actual state of traffic flow, i.e., congested or uncongested; [k i o data / q data It can be determined whether the traffic flow data is located above or below the LUD line [k] i o data / q data ]∩C data The intersection of the predicted and actual results is taken; if the results are the same, the value is incremented by 1; otherwise, it remains unchanged. The optimal LUD line is obtained when all particles in the particle swarm have traversed all the data and found the maximum number of correct classifications, N1.

[0076] Construct an objective function that maximizes the congestion detection accuracy of Ocrit and Vcrit, expressed by the following formula:

[0077]

[0078] In the formula: N2—the sum of the maximum number of correct classifications by the McMaster algorithm for both occasional and frequent congestion; Vcrit i Represents particle p i Draw a line parallel to the x-axis; Ocrit i Represents particle p i Draw a line parallel to the y-axis; [Vcrit q] i data ] indicates that the true data is located in Vcrit i Above or below; [Ocrit o i data [] indicates that the true data is located in Ocrit i Left or right side; R data Represents the state of traffic congestion, namely, frequent congestion, occasional congestion, and slow traffic flow; [Vcrit q R] i data data [Ocrit o R] i data data The symbol ∩ represents the intersection of the predicted and actual results; if the results are the same, it increments by 1; otherwise, it remains unchanged. The optimal Ocrit and Vcrit are obtained when all particles in the particle swarm have traversed all the data and the maximum number of correct classifications N² is found.

[0079] 3) The McMaster algorithm coordinates under the adaptive particle swarm optimization algorithm and the traditional McMaster 3D coordinates have an angle in the control plane, requiring coordinate transformation. First, the coordinate system needs to be translated, moving the origin of the improved McMaster theoretical coordinate system to the origin of the standard coordinate system. The origin of the improved McMaster theoretical coordinate system coincides with the projection of the data bifurcation point onto the control plane, and the projection of the data bifurcation point onto the control plane coincides with the point (Vcrit, Ocrit) in the McMaster algorithm. Therefore, the coordinate position of the origin of the mutation theory coordinate system in the standard coordinate system is (Vcrit, Ocrit, 0). The transformation formula is:

[0080] In the formula: v, o, and q are the original traffic flow parameters of claim 2; v', o', and q' represent the translated data; Octit and Vcrit represent the optimal parameters obtained through the adaptive particle swarm optimization algorithm. Next, the coordinate axes are rotated. Rotating the conventional coordinate axes to coincide with the new coordinate axes allows the q-axis to become the dividing line between congested and uncongested states. The positive half of the q-axis represents congested states, and the negative half represents uncongested states. The transformation formula is as follows:

[0081] In the equations V, O, and Q, the final traffic flow parameters are obtained after coordinate transformation; θ represents the angle between the conventional coordinate axis and the cusp catastrophe theory coordinate axis; Qm represents the maximum flow rate in the data; Om represents the occupancy rate when the flow rate is maximum; and KLUD represents the optimal parameters obtained by the adaptive particle swarm optimization algorithm in the McMaster algorithm.

[0082] Step 5 involves calculating the second partial derivative of the adaptive particle swarm optimization (APSO) McMaster model expression with respect to V. The timing of sudden traffic flow changes is determined using a cubic equation discriminant formula, as shown in the graphical representation below. The trained APSO-improved McMaster algorithm is then used to detect intersection traffic congestion based on traffic data detected by millimeter-wave radar.

[0083]

[0084]

[0085] Δ=27a 2 c 2 Q 2 +8b 3 O 3

[0086] Step six involves the following process for determining the output result Δ in step five: When Δ < 0, the change in the state variable is similar to that of occasional traffic congestion; when Δ > 0, the change in the state variable is similar to that of frequent traffic congestion. In this bifurcation set, a sudden change in the state variable can be determined by whether it passes through the Δ < 0 portion. Similarly, in a traffic flow system, a sudden change in traffic flow parameters, i.e., whether occasional traffic congestion has occurred, can also be determined by whether the traffic flow data passes through the Δ < 0 portion.

Claims

1. An adaptive particle swarm optimization McMaster algorithm for millimeter-wave radar detection in intersection congestion detection, characterized in that, Includes the following steps: Step 1: Extract traffic parameters based on millimeter-wave radar data, including vehicle flow, speed, and time occupancy. Step 2: Determine the relationship surface between occasional and frequent congestion data in the three-dimensional variable parameter space of traffic flow-occupancy-time occupancy; Step 3: Establish a three-dimensional McMaster algorithm model of the relationship between occasional and frequent congestion data in the three-dimensional variable parameter space of traffic flow-occupancy-time occupancy; Step 4: Train the parameters of the McMaster algorithm model, including the minimum uncongested data threshold (LUD), critical occupancy rate (Ocrit), and critical flow rate (Vcrit). The specific method is as follows: 1) Construct an adaptive particle swarm optimization (McMaster) algorithm model. 2) Train the McMaster algorithm model using the relational dataset constructed in step 3. The parameters LUD, Ocrit, and Vcrit divide the traffic flow-occupancy relationship graph into four regions: the road is in a smooth state, the downstream has frequent congestion, the downstream road is in a slow state, and the downstream has occasional traffic congestion. Based on the three parameters, propose the objective function that maximizes the congestion discrimination accuracy. Step 5: Transform the coordinates of the relationship diagram in Step 2, and substitute the original traffic flow parameter data into the model. Step 6: Detect traffic congestion at intersections based on the real-time vehicle movement status monitored by millimeter-wave radar using the trained model.

2. The method according to claim 1, characterized in that, The specific process of step one described is as follows: 1) Based on the vehicle status data detected by the millimeter-wave radar detector, detect the traffic parameters of each lane of the selected experimental road; 2) The calculation formula for traffic flow parameters within a cycle detected by the millimeter-wave radar detector is as follows: Q = VK K = N / L Q=K j (VV 2 / V0) In the formula, speed V, flow rate Q, and density K, L is the number of vehicles in the road segment, Kj is the congestion density, and V0 is the average driving speed when the speed is 0.

3. The method according to claim 1, characterized in that, Step two involves dividing congestion into frequent and occasional congestion based on the different causes of congestion, and analyzing the spatial distribution characteristics and relationships of traffic flow (vehicle volume), average speed, and time occupancy under different congestion conditions.

4. The method according to claim 1, characterized in that, The specific process of step four described is as follows: 1) If the data falls above LUD, the road segment is in a smooth state; otherwise, it is in a congested state. 2) If the data falls below LUD, it is compared with the parameter variables Qcrit and Vcrit to determine the congestion type. 3) If the data falls on the LUD line, then take a line parallel to the x-axis and y-axis. The Vcrit line parallel to the x-axis distinguishes between frequent traffic congestion and slow traffic flow, and the Ocrit line parallel to the y-axis distinguishes between occasional traffic congestion and slow traffic flow.

5. The method according to claim 1, characterized in that, The calculation process for step four described is as follows: The theoretical model equation is E(v,o,q)=av4+bov 2 +cqv Where a, b, and c are parameters, once these three parameters are confirmed, the model can be used to depict changes in traffic flow.

6. The method according to claim 1, characterized in that, The calculation process for step four, process 2), is as follows: The objective function for maximizing LUD congestion discrimination accuracy is expressed as: argmaxN1=∑[k i o data / q data ]∩C data The formula argmax is a function that evaluates the set of parameters of a function. argmaxN1 represents the value of parameter set k when N reaches its maximum value; N1 represents the maximum number of correct classifications in the McMaster algorithm for both congested and uncongested scenarios; N represents the total number of actual data points; ki represents the slope of the line connecting particle Pi in the particle swarm optimization algorithm to the origin of the flow-occupancy graph; Odata and qdata represent the actual occupancy and flow data; Cdata represents the actual state of traffic flow, i.e., congested or uncongested; [k i o data / q data It can be determined whether the traffic flow data is located above or below the LUD line [k] i o data / q data ]∩C data The intersection of the predicted and actual results is taken; if the results are the same, the value is incremented by 1; otherwise, it remains unchanged. The optimal LUD line is obtained when all particles in the particle swarm have traversed all the data and found the maximum number of correct classifications, N1. The objective function for maximizing the congestion detection accuracy of Ocrit and Vcrit is expressed as: In the formula: N2—the sum of the maximum number of correct classifications by the McMaster algorithm for both occasional and frequent congestion; Vcrit i Represents particle p i Draw a line parallel to the x-axis; Ocrit i Represents particle p i Draw a line parallel to the y-axis; [Vcritq] i data ] indicates that the true data is located in Vcrit i Above or below; [Ocrit o i data [] indicates that the true data is located in Ocrit i Left or right side; R data Represents the state of traffic congestion, namely, frequent congestion, occasional congestion, and slow traffic flow; [Vcrit q R] i data data [Ocrit o R] i data data The symbol ∩ represents the intersection of the predicted and actual results; if the results are the same, it increments by 1; otherwise, it remains unchanged. The optimal Ocrit and Vcrit are obtained when all particles in the particle swarm have traversed all the data and the maximum number of correct classifications N² is found.

7. The method according to claim 1, characterized in that, The calculation process for step four, process 2), is as follows: In the adaptive particle swarm optimization algorithm, the McMaster algorithm coordinates and the traditional McMaster 3D coordinates have an angle in the control plane, requiring a coordinate transformation. First, the coordinate system needs to be translated, moving the origin of the improved McMaster theoretical coordinate system to the origin of the standard coordinate system. The origin of the improved McMaster theoretical coordinate system coincides with the projection of the data bifurcation point onto the control plane, and the projection of the data bifurcation point onto the control plane coincides with the point (Vcrit, Ocrit) in the McMaster algorithm. Therefore, the coordinate position of the origin of the mutation theory coordinate system in the standard coordinate system is (Vcrit, Ocrit, 0). The transformation formula is: In the formula: v, o, and q are the original traffic flow parameters of claim 2; v', o', and q' represent the translated data; Octit and Vcrit represent the optimal parameters obtained through the adaptive particle swarm optimization algorithm. Next, the coordinate axes are rotated. Rotating the conventional coordinate axes to coincide with the new coordinate axes allows the q-axis to become the dividing line between congested and uncongested states. The positive half of the q-axis represents congested states, and the negative half represents uncongested states. The transformation formula is as follows: Equations V, O, and Q represent the final traffic flow parameter data after coordinate transformation; θ represents the angle between the conventional coordinate axis and the cusp catastrophe theory coordinate axis; Qm represents the maximum flow rate in the data; Om represents the occupancy value when the flow rate is maximum; KLUD represents the optimal parameters obtained by the adaptive particle swarm optimization algorithm in the McMaster algorithm.

8. The method according to claim 1, characterized in that, The calculation process for step five described is as follows: By taking the second partial derivative function with respect to V in the expression of the adaptive particle swarm optimization McMaster model, and using the cubic equation discriminant formula, we can determine when a sudden change in traffic flow occurs. The graphical expression is as follows: △=27a 2 c 2 Q 2 +8b 3 O 3 。 9. The method according to claim 1, characterized in that, The discrimination process described in step six is ​​as follows: When Δ < 0, the changes in the state variables are similar to those during sporadic traffic congestion; when Δ > 0, the changes in the state variables are similar to those during frequent traffic congestion. In this bifurcation set, we can determine whether a sudden change has occurred in the state variables by judging whether the state variables pass through the Δ < 0 portion. Similarly, in traffic flow systems, we can determine whether a sudden change has occurred in traffic flow parameters, i.e., whether sporadic traffic congestion has occurred, by judging whether the traffic flow data passes through the Δ < 0 portion.

Images