Expressway hard shoulder dynamic opening and upstream shunting collaborative management and control decision method

Abstract

The application belongs to the technical field of intelligent transportation, and discloses a highway hard shoulder dynamic opening and upstream shunting collaborative management and control decision method, which comprises the following steps: S1. A multi-level highway traffic flow operation state evaluation system is constructed based on a fuzzy c-means clustering algorithm FCM; S2. A hard shoulder dynamic opening decision model is established; S3. An upstream induction shunting management and control decision model considering road network performance is established; and S4. The type of a highway abnormal event is judged. According to the road network traffic flow operation state index and the type of an accident and other factors, the method can dynamically select the most suitable management and control combination, effectively reduce the queuing length of the main line lane of the highway, promote the rapid recovery of the average speed, significantly reduce the average delay of vehicles, and improve the operation efficiency of the main line of the highway.

Description

Technical Field

[0001] This invention belongs to the field of intelligent transportation technology, specifically relating to a decision-making method for dynamic opening of hard shoulders and upstream diversion coordination on highways. Background Technology

[0002] The hard shoulder of a highway refers to a road surface structure parallel and adjacent to the driving lanes, narrower than a standard driving lane, and possessing sufficient load-bearing capacity. Its main functions include providing temporary parking space for disabled vehicles and providing passage for rescue vehicles in emergencies. Dynamic opening of the hard shoulder, as an Advanced Traffic Management (ATM) strategy, while offering significant advantages in increasing road capacity, has certain limitations in its application. For increased traffic demand during peak hours without accidents, opening the hard shoulder at appropriate sections and times can balance traffic safety and efficiency, fully utilizing the highway's capacity. However, in abnormal events such as major traffic accidents and severe weather, opening the hard shoulder as a driving lane is not suitable, as the highway's mainline capacity is limited. If traffic demand continues to increase, upstream diversion, as another ATM strategy, can alleviate traffic pressure on the highway's mainline and improve overall network utilization efficiency by guiding some vehicles to detour through the affected sections. By dynamically opening hard shoulders in conjunction with upstream traffic diversion, the capacity of the main expressway can be increased while traffic flow is rationally allocated, thereby optimizing the allocation of road network resources and improving the overall operational efficiency of the expressway network.

[0003] Current research on highway traffic flow management shows a trend towards the synergy of multiple proactive traffic control strategies. Recent studies indicate that systematically integrating two or three proactive traffic management strategies can significantly improve traffic flow management effectiveness. Variable speed limits, dynamic shoulder opening, upstream induced diversion, and ramp control, as core methods of Active Traffic Management (ATM), have become research focuses. Current research primarily aims to combine these methods according to a specific logical framework based on highway traffic conditions and parameters, constructing a multi-level collaborative traffic control decision-making system. At the level of selecting specific proactive traffic management strategies, most research focuses on variable speed limits and ramp management strategies for the highway mainline, i.e., controlling traffic flow from the perspectives of the highway mainline and surrounding roads respectively. However, research on the synergistic control strategy of dynamic shoulder opening and upstream induced diversion is scarce. Dynamic shoulder opening has significant advantages in temporarily increasing road capacity and alleviating congestion bottlenecks, while upstream induced diversion can effectively balance the regional road network load. The synergistic application of these two strategies will provide a new solution for highway traffic flow management.

[0004] Patent document CN118942237A only considers the timing and duration of opening the hard shoulder, but this method is not suitable for opening the hard shoulder as a driving lane under abnormal events such as major traffic accidents and severe weather. The main line of the highway has limited traffic capacity, and the applicability of this method will decrease if traffic demand continues to increase. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a collaborative management and decision-making method for dynamic opening of hard shoulders and upstream diversion on highways. Dynamic opening of hard shoulders has certain limitations. To consider network balance in traffic flow regulation under both accident and non-accident conditions, this invention proposes a collaborative management and decision-making method for hard shoulder opening and upstream diversion under non-accident high-volume and abnormal event conditions. This method organically combines dynamic opening of hard shoulders with upstream diversion to maximize traffic management benefits.

[0006] The objective of this invention is achieved through the following technical solution:

[0007] This invention provides a method for coordinated management and decision-making of dynamic opening of hard shoulders and upstream diversion on highways, comprising the following steps:

[0008] S1. Construct a multi-level highway traffic flow operation status evaluation system based on the fuzzy c-means clustering algorithm (FCM);

[0009] S2. Establish a dynamic decision-making model for opening hard shoulders;

[0010] S3. Establish an upstream induced diversion control decision model that considers road network performance;

[0011] S4. Determine the type of abnormal event on the highway;

[0012] If the congestion is under non-accident conditions, the nodes for implementing corresponding control strategies are determined based on the traffic flow operation status at the road segment, road and road network levels, taking into account the TPI. Dynamic opening of hard shoulders and upstream guidance and diversion are coordinated control decisions.

[0013] In the case of congestion caused by an accident, a combination strategy of opening the hard shoulder and guiding upstream traffic diversion should be determined by comprehensively considering the severity of the accident, the lane congestion status, and the traffic flow operation parameters of the road network.

[0014] Furthermore, step S1 includes the following sub-steps:

[0015] S1.1 Acquires ETC gantry data, preprocesses it, and calculates speed, flow rate, and density at five-minute intervals. FCM is used for clustering. Based on the clustering results, the traffic flow status is divided into six parts: extremely smooth, smooth, relatively smooth, relatively congested, congested, and severely congested. Based on the clustering results, the flow rate-density threshold for each traffic flow status is determined, and the traffic flow density index of the road segment is defined based on the relationship between the three parameters of flow rate, density, and speed, using the speed and flow rate thresholds.

[0016] S1.2 At the road level, the Traffic Flow Status Index (TPI) for highways is introduced. R ;

[0017]

[0018] In the formula, VHT Li q is the vehicle travel time of the i-th road segment within the discrimination time; i It is the traffic flow of the i-th segment at the time of judgment; L i v is the length of the i-th road segment; i It is the average vehicle speed of the i-th road segment at the discrimination time; n i It is the number of lanes in the i-th road segment within the discrimination time; TPI Li is the traffic operation status index of the i-th road segment within the discrimination time; m is the total number of road segments divided by the target expressway;

[0019] S1.3 At the road network level, the Traffic Flow Performance Index (TPI) for highways is introduced. N ;

[0020]

[0021] In the formula, VHT Ri R is the travel time of vehicles on the i-th road; i This is the weight value of the i-th road, with national highways set to 3, provincial highways to 2, and others to 1; TPI Ri is the traffic operation status index of the i-th road; n is the number of roads in the target expressway network.

[0022] Furthermore, step S2 includes the following sub-steps:

[0023] S2.1 divides the spatial segments of the highway;

[0024] The merging sections of the highway entrance ramps and the diverging sections of the highway exit ramps are regarded as nodes. The area between the nodes is a basic section of the highway. The basic section of the highway is divided into spatial segments of about 500m as the smallest decision-making unit.

[0025] For the nth spatial segment between two gantry frames, the flow rate of the nth spatial segment is estimated using linear interpolation. The calculation formula is as follows:

[0026]

[0027] In the formula, Q n Q represents the flow rate of the nth spatial segment; A and Q B It is the flow rate at adjacent gantry A and B; L AB d is the distance between adjacent gantry A and B; d is the distance of the nth spatial segment from ETC gantry A;

[0028] S2.2 Identify and predict traffic flow breakdowns;

[0029] When the speed drop difference between consecutive time intervals is greater than 16 km / h, the space occupancy rate increases by 5%, and the average speed does not recover to the original speed within the following 10 minutes, a traffic flow collapse is determined to have occurred.

[0030] S2.3 Determines the duration of hard shoulder opening;

[0031] After identifying the crash phenomenon, traffic flow is predicted over the next N time intervals based on the GCN-LSTM model. If there are n consecutive time intervals that satisfy The basic opening duration for hard shoulders can be expressed as:

[0032]

[0033] In the formula, S i (t) represents the hard shoulder switch HSR in the i-th spatial segment and the t-th time interval. ctrl Status; HSR ctrl 1 represents an open hard shoulder used as a driving lane, HSR ctrl A value of 0 indicates that the hard shoulder is not permitted; T represents the time interval.

[0034] After the predicted traffic drops below the threshold, the hard shoulder should not be closed immediately; instead, it should remain open for 1 to 2 time intervals.

[0035] S2.4 In the optimized control strategy, open spatiotemporal constraints considering safety factors are introduced;

[0036] I. Introduce time constraints;

[0037] The time constraint for dynamic opening control decision of hard shoulder is expressed as:

[0038]

[0039] In the formula, S i (t) represents the hard shoulder switch HSR in the i-th spatial segment and the t-th time segment. ctrl State; S i (t+1) represents the hard shoulder switch HSR in the i-th spatial segment and the (t+1)-th time period. ctrl State; n HSR This indicates the total number of hard shoulder switch state transitions across all spatial segments; n is the total number of state transitions of all hard shoulder switches in all spatial segments. HSR The threshold;

[0040] II. Introduce spatial constraints;

[0041] Within the same time interval T, a spatial segment of continuously open hard shoulder lanes is counted as a lane component, denoted by C. ci Indicates; where C ci =1 indicates that the corresponding area forms a lane component; otherwise, C ci =0;

[0042] Define the number of lane components: For a highway segment of length L, within the same time interval T, the sum of the lane components of the hard shoulder lane is the number of lane components for that time interval, denoted by N. lc Indicates; N lc The calculation expression is:

[0043]

[0044] The spatial constraints for dynamic opening control decisions of hard shoulders are expressed as follows:

[0045]

[0046] In the formula, The number of lane components N lc The threshold.

[0047] Furthermore, step S3 includes the following sub-steps:

[0048] S3.1 Determine the branching node;

[0049] Among them, vehicles traveling on the expressway are considered internal vehicles, and the diversion nodes for internal vehicles are distributed at key nodes on the expressway main road and detour routes; vehicles that have not yet entered the expressway are considered external vehicles, and the diversion nodes for external vehicles are distributed at key nodes on the expressway peripheral road system.

[0050] S3.2 Selection of routing paths based on redundancy;

[0051] I. Determine path 1, that is, from P odThe internal selection process identifies a route that significantly reduces travel time while minimizing additional mileage; among which, P od Let be the detour path set, representing the n paths between the diversion node and the destination; then compare path 1 with the remaining n-1 paths, and calculate the overlap R between the two paths, which is the ratio of the common length of the two paths to the length of the alternative paths. The calculation formula is:

[0052]

[0053] In the formula, i is the determined path i∈I; j is the alternative path, j∈J; R ij It is the degree of repetition between path i and path j; l ij It is the length of the road segment shared by paths i and j; l j It is the total length of path j;

[0054] II. Iteratively select the branching path, where paths with a repetition degree greater than 0.5 are not considered branching paths;

[0055] S3.3 Calculates the flow distribution based on the Logit model;

[0056] I. Calculate the traveler's perceived path impedance;

[0057] II. Calculate the traveler's perceived path impedance;

[0058] III. Assign paths to the distributed traffic;

[0059] The upstream induced diversion control decision model based on the overall performance of the road network is expressed as follows:

[0060]

[0061] In the formula, m is the total number of branching paths; Q i It is the flow distribution of the i-th path; P i C is the probability of choosing the i-th path; N0 is the traffic flow on the main highway; C a This refers to the remaining traffic capacity of the main highway; C i N represents the remaining capacity of the i-th path; i R is the flow of the i-th path; ij It is the degree of repetition between the i-th path and other paths.

[0062] Beneficial effects:

[0063] This invention proposes a collaborative management and decision-making method for dynamic opening of hard shoulders and upstream diversion on highways. This method considers both accident and non-accident conditions while fully taking into account network balance, thus avoiding "congestion drift." The upstream induced diversion in this invention is an ATM (Automated Traffic Management) strategy, which guides some vehicles to detour through the affected sections, thereby alleviating traffic pressure on the main highway and improving overall network utilization efficiency. By coordinating dynamic opening of hard shoulders and upstream diversion, this invention can increase the capacity of the main highway while achieving a rational allocation of traffic flow, thereby optimizing network resource allocation and improving the overall operational efficiency of the highway network.

[0064] This invention constructs a multi-level highway traffic flow operation status evaluation system. Using the FCM clustering method, it classifies the traffic status of road segments into six categories: extremely smooth, smooth, relatively smooth, relatively congested, congested, and severely congested, extending this classification to the road and road network levels. Based on this, from the perspective of overall road network performance, it proposes control decision-making methods for dynamic hard shoulder opening and upstream diversion coordination in two typical scenarios: non-accident high-volume traffic and abnormal events. This method can dynamically select the most suitable control combination based on factors such as the road network traffic flow operation status index and accident type, effectively reducing queue length on highway mainline lanes, promoting rapid recovery of average speed, significantly reducing average vehicle delays, and improving the operational efficiency of highway mainlines.

[0065] 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

[0066] Figure 1 This is a flowchart of a collaborative management and decision-making method for dynamic opening of hard shoulders and upstream diversion on highways, as described in this invention.

[0067] Figure 2 A flowchart for collaborative management and control decision-making under non-accident conditions;

[0068] Figure 3 This is a flowchart for collaborative control decision-making under accident conditions. Detailed Implementation

[0069] To make the technical solutions, advantages, and objectives of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention without creative effort are within the protection scope of this application.

[0070] like Figure 1 As shown, this invention provides a method for coordinated management and decision-making of dynamic opening of hard shoulders and upstream diversion on highways, including the following steps:

[0071] Step 1: Construct a multi-level highway traffic flow operation status evaluation system based on FCM to lay a quantitative foundation for subsequent control decisions. The highway traffic flow operation status evaluation system is constructed using flow rate and speed as core evaluation indicators. The fuzzy c-means algorithm (FCM) is used to classify traffic states into six categories: extremely smooth, smooth, relatively smooth, relatively congested, congested, and severely congested. At the road segment level, TPI (Traffic Flow Index) is used as the basis for further evaluation. L To characterize the traffic flow operation status of a road segment, vehicle hourly rate (VHT) and road grade weight parameters are introduced to adjust the TPI. L Extending to the road and road network levels, they are represented as TPI respectively. R and TPI N .

[0072] Step 1.1: At the road segment level, the traffic operation state classification results obtained from FCM are introduced as the Traffic Performance Index of Link (TPI) at the highway road segment level. L Classification criteria. Based on the flow-density thresholds for each traffic flow operating state determined by FCM, and according to the relationship between the three parameters of flow, density, and speed, the Traffic Flow Density Index (TPI) for road segments is defined using speed and flow thresholds. L .

[0073] Step 1: Obtain ETC gantry data, preprocess it, calculate speed, flow rate and density at five-minute intervals, and use FCM for clustering.

[0074] The objective function of FCM is expressed as follows:

[0075]

[0076] in, It is the membership degree of the j-th data sample to the i-th cluster center, and u ij∈[0,1], for data point i, we have d ij It is the Euclidean distance between the sample data and the cluster center; m is the fuzzy weighting coefficient, which is generally m=2.

[0077] During initialization, the number of clusters *c*, the maximum number of iterations *T*, the convergence threshold *ε*, and the initial membership matrix *U0* need to be determined. During iterative optimization, the cluster centers are first calculated based on the current membership matrix *U*, expressed by the following formula:

[0078]

[0079] Secondly, based on the newly calculated cluster centers, update the membership degree of each data point to each cluster:

[0080]

[0081] Finally, iteration stops when the maximum number of iterations T is reached or the termination condition is met early. The termination condition expression is:

[0082] ||U (t) -U (t-1) ||≤ε (1.4)

[0083] The initial cluster centers significantly impact the FCM results, and setting the initial cluster size based on experience is highly subjective. To enhance the objectivity and scientific rigor of the clustering results, the silhouette coefficient method (Sil(i)) is introduced to determine the initial cluster size c. This method evaluates the clustering quality by assessing the similarity of samples within clusters and the dissimilarity between clusters. Its value ranges from [-1, 1], with values ​​closer to 1 indicating more ideal clustering results. The formula is as follows:

[0084]

[0085] Among them, a i b is the cohesion of data point i, representing the degree of closeness between the data point and other points of the same type; i is the separation degree of data point i, representing the degree to which the data point is separated from other categories of points. i and b i Satisfy the following formula:

[0086]

[0087] Among them, C i It is the cluster to which data point i belongs; C k It is the cluster k to which data point i does not belong; |C i |and|C k | represents the number of samples in the corresponding cluster; d(i,j) is the distance between data points i and j. The average of the silhouette coefficients for all samples is:

[0088]

[0089] The optimal number of clusters, c, is chosen as the number of clusters with the largest average silhouette coefficient, i.e., when S N When the value is maximized, the optimal number of clusters is N.

[0090] Step 2: Based on the clustering results, the traffic flow status is divided into six parts. Combining the macroscopic traffic flow basic map and the three-phase traffic flow theory, the traffic status of these six parts can be described as: extremely smooth, smooth, relatively smooth, relatively congested, congested, and severely congested.

[0091] Step 3: Based on the clustering results, determine the flow-density thresholds for each traffic flow operation state, and define the traffic flow density index of the road segment using speed and flow thresholds according to the relationship between the three parameters of flow, density and speed.

[0092] Step 1.2: At the road level, introduce the Traffic Performance Index of Road (TPI) for highways. R The weighted average of the conditions of each road segment that makes up the road is determined by the weighted average of the conditions of the segments that make up the road, and the driver’s evaluation of congestion is introduced as a weight.

[0093] VHT is selected as the weight value to characterize traffic congestion characteristics, and the calculation method is as follows:

[0094]

[0095] Among them, VHT Li q represents the travel time of the vehicle on the i-th road segment within the discrimination time. i It is the traffic flow of that road segment at that time, L i It is the length of the road segment, v i It is the average speed of vehicles on that road segment at that time; n i It is the number of lanes in the i-th road segment within the discrimination time; TPI Li is the traffic operation status index of the i-th road segment within the discrimination time; m is the total number of sub-segments of the expressway.

[0096] Due to TPI L It is not a continuous variable, but TPI R ' is a continuous variable within the interval [1, 6]. And when the road is in a state of perfect unobstructed flow, TPI... R A TPI ≥ 1 may incorrectly classify extremely smooth traffic as smooth traffic, failing to accurately reflect the actual traffic flow status, i.e., the calculated TPI. R 'Too large. Therefore, TPI needs to be adjusted.' RTo more accurately reflect the actual road traffic flow conditions, a correction was made using rounding. The transformed road traffic flow operation state index is:

[0097]

[0098] Step 1.3: At the road network level, introduce the Traffic Performance Index of network (TPI) at the highway road network level. N The weighted average of the operating status of each road within the road network is determined by the weighted average of the operating status of all roads. Given the differences in the functions and importance of roads of different grades within the road network, the road grade R is introduced as a weighted value in addition to the VHT (Variable Hierarchy Testing). The calculation formula is as follows:

[0099]

[0100] Among them, VHT Ri R is the travel time of vehicles on the i-th road; i This is the weight value for that road: 3 for national highways, 2 for provincial highways, and 1 for others; TPI Ri is the traffic performance index of the i-th road; n is the number of roads in the highway network. Similarly, for TPI... N The range of values ​​of ' is mapped and transformed:

[0101]

[0102] Step 2: Establish a dynamic decision-making model for opening hard shoulders. Under the condition of meeting the conditions for opening hard shoulders, determine the timing of opening hard shoulders based on traffic flow prediction and breakdown phenomenon identification. Considering traffic demand and safety factors, spatiotemporal constraints are introduced to determine the opening duration.

[0103] Step 2.1: Based on spatial nodes and decision-making needs, divide the highway spatial segments, further identify the Breakdown phenomenon based on predicted traffic flow, and determine the timing of hard shoulder opening.

[0104] Step 1: Divide the highway spatial segments. Treat the merging sections of highway entrance ramps and the diverging sections of highway exit ramps as nodes. The areas between these nodes constitute a basic highway segment, allowing for the implementation of active traffic management strategies on a given segment. Further, divide the basic highway segment into smaller segments of approximately 500m each as the smallest decision-making unit for subsequent research. A basic highway segment of length L is divided into m spatial segments, denoted by i, with each segment having a length of l. i In the decision-making process regarding the opening and closing of hard shoulders, HSR is used. ctrlHSR is the control variable for the hard shoulder switch. ctrl 1 represents an open hard shoulder used as a driving lane, HSR ctrl A value of 0 indicates that the hard shoulder is not allowed to be used. Therefore, the control decision for opening and closing the hard shoulder is the variable HSR. ctrl Combinations of 0 and 1 in time sequence.

[0105] Since the smallest decision unit is approximately 500m, there are several spatial segments between each pair of ETC gantries. For the nth spatial segment between two gantries, the flow rate is estimated using linear interpolation. The calculation formula is as follows:

[0106]

[0107] Among them, Q n Q represents the flow rate of that spatial segment; A and Q B It is the flow rate at adjacent gantry A and B; L AB d is the distance between adjacent gantry A and B; d is the distance of this spatial segment from ETC gantry A.

[0108] Step 2: Identify the Breakdown phenomenon in predicted traffic flow as the core of the opening trigger mechanism, and intervene in advance before traffic flow collapses. The dynamic opening strategy for hard shoulders is specifically described as a 0-1 decision problem in discrete spatiotemporal segments. A traffic flow collapse occurs when the speed decrease difference between consecutive time intervals exceeds 16 km / h, the occupancy rate increases by 5%, and the average speed does not recover to its original speed within the following 10 minutes. T represents the time interval, and t represents the current time step. If T = 5 minutes, then the Breakdown condition is:

[0109]

[0110] The aim is to increase road capacity in the short term by opening hard shoulders before traffic flow collapse occurs, thus preventing traffic flow collapse, or to restore traffic flow to pre-collapse levels by opening hard shoulders as driving lanes after traffic flow collapse has occurred.

[0111] Step 2.2: Determine the opening duration of the hard shoulder based on the traffic flow prediction model. Determine the basic opening duration by analyzing the predicted traffic flow parameters and introduce spatiotemporal constraints to ensure driving safety.

[0112] Step 1: Determine the opening duration. For a road segment i, if a breakdown can be identified, the traffic flow in the cycle preceding the breakdown is used as the traffic demand threshold Q to trigger the opening of the hard shoulder. th Qth It is determined based on traffic conditions and can be dynamically adjusted according to special events, weather, and other factors, rather than relying solely on the road's designed capacity.

[0113] After identifying the breakdown phenomenon, traffic flow is predicted over N future time intervals based on the GCN-LSTM model. If there are n consecutive time intervals that satisfy The basic opening duration for opening hard shoulders can be expressed as:

[0114]

[0115] Among them, S i (t) represents the hard shoulder switch HSR in the i-th spatial segment and the t-th time interval. ctrl State. To ensure that traffic flow fully recovers to a stable state, the hard shoulder should not be closed immediately after the predicted flow decreases below a threshold. Instead, the hard shoulder should remain open for 1 to 2 time intervals, defined as the post-buffer period, denoted by T. buf This is indicated by [the following text is missing here, likely a formatting error]. During this period, drivers need to be informed that the hard shoulder lane is about to close so they can adapt to the upcoming lane change and allow vehicles traveling on the hard shoulder lane to switch to the main lane in a timely manner.

[0116] Step 2: In the optimization control strategy, open spatiotemporal constraints that take into account safety factors are introduced.

[0117] A time constraint is introduced. To avoid drastic changes in the opening or closing of the hard shoulder along the entire highway within the study area, which could lead to safety accidents, it is necessary to constrain the number of hard shoulder openings and closings. The sum of the number of hard shoulder opening and closing states in each spatial segment between two adjacent decisions is defined as the number of time window transitions. S i (t) represents the hard shoulder switch HSR in the i-th spatial segment and the t-th time segment. ctrl State; S i (t+1) represents the hard shoulder switch HSR in the i-th spatial segment and the (t+1)-th time period. ctrl State, variable S i A change from 0 to 1 or from 1 to 0 is counted as one time window transition. The total number of hard shoulder switch state transitions across all spatial segments is denoted by n. HSR Therefore, the time constraint for the dynamic opening control decision of the hard shoulder can be expressed as:

[0118]

[0119] in, n is the total number of state transitions of all hard shoulder switches in all spatial segments. HSR The threshold.

[0120] Introduce spatial constraints. For a highway segment of length L, divide it into m spatial segments, denoted by l. i S indicates that i This indicates the hard shoulder lane switch HSR in this spatial segment. ctrl State. Based on the concept of connected components in graph theory, a spatial segment of continuously open hard shoulder lanes within the same time interval T is counted as a lane component. Let C... ci It means that C ci =1 indicates that the area constitutes a lane component; otherwise, C ci =0.

[0121] To describe the discreteness of hard shoulder lane switching, based on the concept of lane components, the number of lane components is defined as follows: For a highway segment of length L, within the same time interval T, the sum of the lane components of the hard shoulder lane is the number of lane components for that time interval, denoted by N. lc The formula is expressed as follows:

[0122]

[0123] Therefore, the spatial constraints for dynamic opening control decisions of hard shoulders can be expressed as:

[0124]

[0125] in, The number of lane components N lc The threshold.

[0126] Step 3: Establish an upstream diversion control decision model that considers road network performance. When it is determined that an upstream diversion decision needs to be initiated, use the model to determine the diversion node, diversion path, and diversion volume calculation.

[0127] Step 3.1: Identify Diversion Nodes. Driver traffic guidance needs vary depending on vehicle location. Broadly speaking, they can be categorized into two types: vehicles within the expressway and vehicles not yet on the expressway. Diversion nodes for vehicles within the expressway are primarily located at key nodes on the main expressway arteries and detour routes; for vehicles outside the expressway, the focus is on key nodes within the outer road system.

[0128] Step 1: When vehicles are traveling on the highway and a queue appears ahead, the queue length and the distance to the upstream ramp exit must be assessed. If the tail of the queue extends to the ramp entrance, one or more exit ramps located upstream of the queue tail should be designated as diversion points. Simultaneously, the real-time traffic flow, capacity, and connecting path conditions of the ramps should be assessed to prevent congestion from spreading. In addition to determining the primary diversion nodes, secondary diversion nodes also need to be planned, the location of which depends on the characteristics of the diversion path.

[0129] Step 2: For vehicles that have not yet entered the highway, ramp control strategies should be prioritized, and drivers should be provided with information on alternative routes to avoid congestion on connecting roads. Diversion points for these vehicles are mainly located on roads parallel to the highway.

[0130] Step 3.2: Selecting Diversion Paths Based on Repetition. Once the diversion node is determined, there are usually multiple possible paths between the diversion node and the destination. However, some of these paths are unsuitable as detour options due to their excessive length. Excessive detour distances increase travel costs and reduce drivers' willingness to adopt guidance suggestions.

[0131] The n paths between the diversion node and the destination are taken as the detour path set P. od We need to filter out feasible paths from this set.

[0132] Step 1: First determine path 1: from P od First, select a route that significantly reduces travel time and minimizes additional mileage. Then, compare route 1 with the remaining n-1 routes and calculate the overlap R between the two routes, which is the ratio of the common length of the two routes to the length of the candidate route. The calculation formula is as follows:

[0133]

[0134] Where i is the determined path i∈I; j is the alternative path, j∈J; R ij It is the degree of repetition between path i and path j; l ij It is the length of the road segment shared by paths i and j; l j It is the total length of path j.

[0135] Step 2: Iteratively select diversion paths. As the number of determined diversion paths increases, the overlap of subsequent candidate paths will gradually increase. If the overlapping sections between different diversion paths are long, it will be detrimental to traffic distribution and may even cause new traffic congestion. Therefore, paths with an overlap greater than 0.5 will not be used as diversion paths.

[0136] Step 3.3: Calculate traffic diversion based on the Logit model. After determining the diversion points and paths, the traffic volume on the main line and diversion paths, as well as the remaining capacity, should be considered to determine the diversion traffic volume. When the highway network as a whole has the capacity to divert traffic, diversion measures should be implemented.

[0137] Step 1: Calculate the traveler's perceived path impedance. Assume R... od Let be the set of branching paths between the starting point r and the ending point s, where the path with the minimum impedance is k. Then, the expression for the path impedance perceived by the traveler is:

[0138]

[0139] in, It is the path impedance subjectively perceived by travelers; It is the objective, actual impedance of the path; This is a random error term, representing the difference between subjective perception and the actual objective path impedance. When When the path follows a Gumbel distribution, a multinomial Logit model can be obtained. In this case, the traveler's subjectively perceived path impedance can be expressed using the path flow and the path's objective, actual impedance, as shown in the following equation:

[0140]

[0141] Here, δ is a discrete parameter that reflects the degree of convergence of travelers' perception of path impedance, and is taken as δ = 1.

[0142] Step 2: Calculate the probability that the traveler chooses the k-th route. According to Wardrop's route selection principle, the probability that the traveler chooses the k-th route is:

[0143]

[0144] Here, β is a sensitivity parameter that reflects how sensitive travelers are to changes in costs, and β = 1 is taken.

[0145] Step 3: Finally, route allocation is performed on the diverted traffic. When the total number of diversion routes is m, and the traffic flow to be diverted on the main line is Q, the traffic flow allocated to each diversion route can be determined based on the traveler's probability of choosing each route. The calculation formula is as follows:

[0146]

[0147] Among them, Q k P is the flow rate of the k-th branch path; k It represents the probability of choosing the k-th path.

[0148] The upstream induced diversion control decision model based on the overall performance of the road network can be expressed as:

[0149]

[0150] Where m is the total number of branch paths; Q i It is the flow distribution of the i-th path; P i C is the probability of choosing the i-th path; N0 is the traffic flow on the main highway; C a This refers to the remaining traffic capacity of the main highway; C i N represents the remaining capacity of the i-th path; iR is the flow of the i-th path; ij It is the degree of repetition between the i-th path and other paths.

[0151] Step 4: Determine the type of abnormal event on the highway. If it is congestion under non-accident conditions, determine the nodes for implementing corresponding control strategies based on the traffic flow operation status at the segment, road, and road network levels, considering the TPI (Traffic Performance Index). This involves dynamic opening of hard shoulders and upstream traffic diversion coordination control decisions. In this embodiment, the overall framework diagram of the coordination control decision-making process under non-accident conditions can be found by referring to... Figure 2 .

[0152] Step 4.1: Determine the type of abnormal event on the highway. If it is determined to be congestion under non-accident conditions, obtain the traffic flow status of the road network. If TPI N A TPI > 3 indicates that the road network traffic flow is relatively fragile, and the first consideration should be to implement dynamic opening of hard shoulders to improve the mainline capacity; if TPI N If the value is ≤3, further judgment will be made based on the traffic flow status of the road network.

[0153] Step 4.2: If TPI N =3. At this point, the road network traffic flow status allows for the implementation of induced diversion strategies, but the road operating status also needs to be considered. Diversion nodes and paths are gradually determined, and the road traffic flow operating status of the diversion paths needs to be obtained. If the TPI of the diversion path... R If the value is ≤2, it indicates that vehicles are concentrated on the main line of the expressway and the remaining capacity of the alternative routes is sufficient, so the diversion flow can be further calculated and diversion measures can be implemented. Conversely, it indicates that the traffic operation status of the main line of the expressway and the diversion routes is similar, so no diversion measures should be implemented. However, the decision on whether to implement dynamic opening of the hard shoulder needs to further consider the traffic flow operation status of the main line of the expressway.

[0154] Step 4.3: If the main line of the expressway has a TPI R If the value is ≤2, it indicates that the main line of the expressway is unobstructed and there is no need to implement dynamic opening of the hard shoulder; otherwise, further consideration should be given to the traffic flow status of each road segment to determine whether dynamic opening of the hard shoulder is necessary.

[0155] Step 4.4: If the main line of the expressway has a segment traffic flow status (TPI) L For road sections with a value greater than 3, a dynamic decision to open the hard shoulder should be considered; otherwise, it indicates that no active traffic control strategy is needed.

[0156] Step 4.5: Generate a dynamic decision on opening the hard shoulder using the hard shoulder dynamic opening decision model. If the hard shoulder is to be opened, the algorithm will be used again to make a decision at the time interval before the hard shoulder is closed; otherwise, no active traffic control measures will be implemented.

[0157] Step 5: Determine the type of abnormal event on the highway. If it is congestion under accident conditions, comprehensively consider the severity of the accident, lane congestion status, and road network traffic flow parameters to determine a targeted combination strategy of opening the hard shoulder and upstream traffic diversion. In this embodiment, the overall framework diagram of the collaborative control decision-making process under accident conditions can be found by referring to... Figure 3 .

[0158] Step 5.1: First, obtain the accident type and determine whether the hard shoulder can be opened. If the hard shoulder can be opened, generate a dynamic hard shoulder opening decision through the hard shoulder dynamic opening decision method and implement it; otherwise, directly obtain the road network traffic flow operation status to determine whether to implement the traffic diversion strategy.

[0159] Step 5.2: Obtain the traffic flow operation status of the road network, if TPI N If the value is greater than 3, the diversion strategy will not be implemented; otherwise, the diversion nodes and alternative paths will be further determined.

[0160] Step 5.3: Determine whether the road traffic flow status of the alternative routes allows for the implementation of traffic diversion measures. If TPI R If the value is ≤2, it indicates that the remaining capacity of the alternative paths is sufficient, and the traffic flow will be further calculated and the induced diversion strategy will be implemented; otherwise, the remaining capacity of the alternative paths is insufficient, and the induced diversion strategy will not be implemented.

[0161] Step 5.4: Determine if the time limit has expired. Since the coordinated control strategy at this point is implemented to alleviate congestion caused by the accident, the hard shoulder will be closed and traffic diversion will cease once the event ends and traffic flow resumes. This control measure will then be in effect.

[0162] It is hereby declared that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A decision-making method for dynamic opening of hard shoulders and coordinated management of upstream traffic diversion on highways, characterized in that, Includes the following steps: S1. Construct a multi-level highway traffic flow operation status evaluation system based on the fuzzy c-means clustering algorithm (FCM); Step S1 includes the following sub-steps: S1.1 Acquires ETC gantry data, preprocesses it, and calculates speed, flow rate, and density at five-minute intervals. FCM is used for clustering. Based on the clustering results, the traffic flow status is divided into six parts: extremely smooth, smooth, relatively smooth, relatively congested, congested, and severely congested. Based on the clustering results, the flow-density thresholds for each traffic flow operation state are determined, and the traffic flow density index of the road segment is defined by the relationship between the three parameters of flow, density and speed, using speed and flow thresholds. S1.2 introduces a traffic flow operation status index at the highway level at the road level. ; , , In the formula, It is the first The vehicle travel time for each road segment within the specified time period; It is the first Traffic flow at different times for each road segment; It is the first The length of each road segment; It is the first The average vehicle speed on each road segment at the time of judgment; It is the first The number of lanes in each road segment within the judgment time; It is the first Traffic operation status index of each road segment within the judgment time period; It is the total number of road segments divided by the target expressway; S1.3 introduces a traffic flow operation status index at the highway network level at the road network level. ; , In the formula, It is the first The travel time of vehicles on this road; It is the first The weight values ​​for each road are set as follows: national highways are set to 3, provincial highways to 2, and others to 1. It is the first Traffic operation status index of the road; It is the number of roads within the target highway network; S2. Establish a dynamic decision-making model for opening hard shoulders; S3. Establish an upstream induced diversion control decision model that considers road network performance; S4. Determine the type of abnormal event on the highway; If the congestion is under non-accident conditions, based on The nodes for implementing corresponding control strategies are determined by comprehensively considering the traffic flow operation status at the road segment, road, and road network levels, and dynamic opening of hard shoulders and upstream guidance and diversion are coordinated control decisions. Step 1: Determine the type of abnormal event on the highway. If it is determined to be congestion under non-accident conditions, obtain the traffic flow status of the road network. This indicates that the traffic flow in the road network is currently quite fragile, and the first consideration should be to implement a dynamic opening of hard shoulders to improve the mainline's capacity; if Then, further judgment is made based on the traffic flow status of the road network; Step II: If At this point, the road network traffic flow status allows for the implementation of diversion strategies, but the road operating status also needs to be considered; diversion nodes and paths are gradually determined, and the road traffic flow status of the diversion paths needs to be obtained. If the diversion paths... If the traffic flow is positive, it indicates that vehicles are concentrated on the main highway and the remaining capacity of alternative routes is sufficient. Further calculation of traffic flow should be performed to implement traffic diversion measures. Conversely, if the traffic flow is negative, it indicates that the traffic conditions on the main highway and the diversion routes are similar. No traffic diversion measures should be implemented. The decision on whether to implement dynamic opening of the hard shoulder needs to further consider the traffic flow conditions on the main highway. Step III: If the main line of the expressway If the traffic flow status of each road segment is clear, it indicates that the main line of the expressway is unobstructed and there is no need to implement dynamic opening of the hard shoulder; otherwise, further consideration should be given to the traffic flow status of each road segment to determine whether dynamic opening of the hard shoulder is necessary. Step IV: If the main line of the highway has a segment traffic flow density index For road sections where the hard shoulder is open, a dynamic decision-making process for opening the hard shoulder should be considered; otherwise, it indicates that no active traffic control strategy is needed. Step V: Generate a dynamic hard shoulder opening decision using the hard shoulder dynamic opening decision model; if the hard shoulder is to be opened, the algorithm process is repeated again at the time interval before the hard shoulder is to be closed; otherwise, no active traffic control strategy is implemented at this time. If the congestion is caused by an accident, the severity of the accident, the lane congestion status, and the traffic flow operation parameters of the road network should be comprehensively considered to determine a targeted combination strategy of opening the hard shoulder and guiding upstream traffic diversion. Step 1: First, obtain the accident type and determine whether the hard shoulder can be opened. If the hard shoulder can be opened, generate a dynamic hard shoulder opening decision using the hard shoulder dynamic opening decision method and implement it; otherwise, directly obtain the road network traffic flow operation status to determine whether to implement the traffic diversion strategy. Step II: Obtain the traffic flow operation status of the road network. If the traffic diversion strategy is not implemented, then the diversion nodes and alternative paths will be further determined. Step III: Determine whether the traffic flow conditions of the alternative routes allow for the implementation of traffic diversion measures. This indicates that the remaining capacity of the alternative paths is sufficient, and the traffic diversion and induced diversion strategies will be further calculated; otherwise, the remaining capacity of the alternative paths is insufficient, and induced diversion strategies will not be implemented. Step IV: Determine if the event has ended; Since the coordinated control strategy at this time is implemented to alleviate congestion under the accident, the hard shoulder will be closed and the diversion will be stopped after the event ends and traffic flow resumes. This control measure ends.

2. The method for coordinated management and decision-making of dynamic opening of hard shoulders and upstream diversion on highways according to claim 1, characterized in that, Step S2 includes the following sub-steps: S2.1 divides the spatial segments of the highway; The merging sections of the highway entrance ramps and the diverging sections of the highway exit ramps are regarded as nodes. The area between the nodes is a basic section of the highway. The basic section of the highway is divided into spatial segments of about 500m as the smallest decision-making unit. For the first between the two gantry The first spatial segment is estimated using linear interpolation. The flow rate of a spatial segment is calculated using the following formula: In the formula, Representing the The flow of a spatial segment; and It represents the flow rate at adjacent gantry A and B; It is the distance between adjacent gantry A and B; It is the first The distance of each spatial segment from ETC gantry A; S2.2 Identify and predict traffic flow breakdowns; When the speed drop difference between consecutive time intervals is greater than 16 km / h, the space occupancy rate increases by 5%, and the average speed does not recover to the original speed within the following 10 minutes, a traffic flow collapse is determined to have occurred. S2.3 Determines the duration of hard shoulder opening; After identifying the collapse phenomenon, the future is predicted based on the GCN-LSTM model. Traffic flow within a time interval If there is a continuous The time interval satisfies Then the hard shoulder is opened, and the basic opening time of the hard shoulder is expressed as: In the formula, It is the first The spatial segment, the first Hard shoulder switch for a certain period of time state; A value of 1 indicates that the hard shoulder is open as a driving lane. A value of 0 indicates that the hard shoulder is not permitted. Indicates a time interval; When the predicted traffic decreases to the threshold After this, the hard shoulder should not be closed immediately; it should remain open for 1 to 2 time intervals. S2.4 In the optimized control strategy, open spatiotemporal constraints considering safety factors are introduced; I. Introduce time constraints; The time constraint for dynamic opening control decision of hard shoulder is expressed as: In the formula, Indicates the first The spatial segment, the first Hard shoulder switch for a certain period of time state; Indicates the first The spatial segment, the first Hard shoulder switch for a certain period of time state; This indicates the total number of hard shoulder switch state transitions across all spatial segments; It is the total number of state transitions of all hard shoulder switches in all spatial segments. The threshold; II. Introduce spatial constraints; Will be in the same time interval Within the space, a spatial segment of continuously open hard shoulder lanes is counted as one lane component. Indicates; among which, This indicates that the corresponding area constitutes a lane component; otherwise... ; Define the number of lane components: length is A section of highway, at the same time interval Within the hard shoulder lane, the sum of the lane components is the number of lane components for the corresponding time interval, expressed as... express; The calculation expression is: The spatial constraints for dynamic opening control decisions of hard shoulders are expressed as follows: In the formula, Lane component number The threshold.

3. The method for coordinated management and decision-making of dynamic opening of hard shoulders and upstream diversion on highways according to claim 1, characterized in that, Step S3 includes the following sub-steps: S3.1 Determine the branching node; Among them, vehicles traveling on the expressway are considered internal vehicles, and the diversion nodes for internal vehicles are distributed at key nodes on the main expressway and detour routes; vehicles that have not yet entered the expressway are considered external vehicles, and the diversion nodes for external vehicles are distributed at key nodes on the outer road system of the expressway. S3.2 Selection of routing paths based on redundancy; I. Determine path 1, that is, from The internal selection process identifies a route that significantly reduces travel time while minimizing additional mileage; among which, The detour path set represents the route between the branching node and the destination. 1 path; then connect path 1 with the remaining Compare the two paths and calculate the degree of overlap between them. This is the ratio of the common length of the two paths to the length of the alternative path, calculated using the following formula: In the formula, The path has already been determined. ; It is an alternative path. ; It is a path and path The degree of repetition between them; It is a path and path The length of the shared road section; It is a path The total length; II. Iteratively select the branching path, where paths with a repetition degree greater than 0.5 are not considered branching paths; S3.3 Calculates the flow distribution based on the Logit model; I. Calculate the traveler's perceived path impedance; II. Calculating Traveler Choices The probability of a path; III. Assign paths to the distributed traffic; The upstream induced diversion control decision model based on the overall performance of the road network is expressed as follows: In the formula, m is the total number of branch paths; It is the flow distribution of the i-th path; It represents the probability of choosing the i-th path; This refers to the traffic flow on the main highway. It refers to the remaining traffic capacity of the main highway. It represents the remaining capacity of the i-th path; It represents the traffic flow of the i-th path; It is the degree of repetition between the i-th path and other paths.

Images