A method for macroscopic path planning of vehicles in urban road network for recurrent congestion

Abstract

The application discloses a kind of urban road network vehicle macroscopic path planning method for common congestion, belong to traffic path planning field;The method uses double-layer planning model, including macroscopic path planning and local path planning;In macroscopic path planning, the effective management of large-scale urban road network is realized by using existing traffic subarea division method and macroscopic traffic flow distribution strategy;Local path planning is then based on the traffic subarea passing order given by macroscopic planning, and within each traffic subarea, A* algorithm is used to plan a time-consuming shortest path for vehicles from the boundary of the previous subarea to the boundary of the next subarea. In order to deal with the common traffic congestion of urban road network, the application uses the flow balance theory to distribute traffic flow from the macroscopic level under the global perspective. According to the macroscopic path planning algorithm, the flow between the current traffic subareas is dispatched, so that the urban traffic flow reaches the global optimum.

Description

Technical Field

[0001] This invention relates to the field of traffic route planning, and in particular to a macro-route planning method for vehicles in urban road networks with frequent congestion. Background Technology

[0002] Traditional route planning often performs well when dealing with small-scale road networks, but its resilience decreases significantly when faced with increasingly complex traffic data, while the planning time increases exponentially. Against this backdrop, transforming conventional route planning into a route planning algorithm that combines macro and micro perspectives—emphasizing the use of existing local route planning algorithms for small-scale road networks at the micro level, and then selecting the local traffic sequence based on the macro-path planning—effectively reduces the complexity of the road network and improves the resilience of route planning algorithms for large-scale road networks.

[0003] Taking the road network within Beijing's Fifth Ring Road as an example, by dividing this large-scale road network into traffic sub-zones ranging from one to three square kilometers based on the historical travel order of ride-hailing vehicles, the overall road network is transformed into hundreds of traffic sub-zones with spatial relationships and containing historical travel data, effectively reducing the complexity of the road network and significantly improving the resilience of the macro-path planning algorithm.

[0004] Data shows that in 2013, Beijing experienced an average of 1 hour and 55 minutes of traffic congestion per weekday, with an average vehicle speed of 25.8 km / h during peak hours. In 2015, data from Gaode Maps showed that Beijing's average peak-hour speed was 22.61 km / h, with a peak-hour delay index of 2.06, meaning that commuting by road during peak hours took twice as long as under smooth traffic conditions. Alleviating urban traffic congestion has become a central task in all stages of urban traffic planning, management, control, and policy formulation. Taking Beijing as an example, as of December 2016, Beijing had implemented congestion mitigation projects for 13 consecutive years. In 2017, the city planned to carry out 40 tasks across 7 areas to alleviate traffic congestion.

[0005] The introduction of this macro-path planning concept offers a new approach to recurring traffic congestion. It transforms recurring traffic congestion into a large influx of new traffic flows within one or more traffic sub-zones. By monitoring the average speed within each sub-zone over a given time period and classifying its congestion level, vehicles on the road network can avoid recurring congestion from a macro perspective, preventing its further spread. Simultaneously, it allows recurring congestion to shift to lower-density sub-zones, achieving overall traffic flow balance across the road network.

[0006] This method integrates traffic sub-zone division, dynamic path planning based on traffic flow data, and improved path planning algorithms to provide travelers with optimal travel suggestions, alleviate urban traffic congestion, and improve urban road capacity. Furthermore, this method can be widely applied in intelligent traffic management systems, cooperative control of autonomous vehicles, and other fields, providing strong support for smart city construction. Summary of the Invention

[0007] To address the above problems, this invention proposes a macro-path planning method for vehicles in urban road networks prone to frequent congestion.

[0008] To achieve the objectives of this invention, the technical solution adopted is as follows:

[0009] A macro-path planning method for vehicles in urban road networks prone to frequent congestion includes the following steps:

[0010] S10, according to the existing method of dividing traffic sub-zones, such as based on the historical travel data of ride-hailing services, divide the urban road network into several traffic sub-zones;

[0011] S20. Establish a two-layer planning model: The first layer is the macro-path planning layer, which selects the shortest travel sequence connecting each traffic sub-zone based on the sub-zone where the starting point is located; the second layer is the local path planning layer, which uses the A* path planning algorithm to plan within each traffic sub-zone with the shortest travel time as the optimization objective.

[0012] S30, addressing recurring congestion: To resolve recurring congestion in the current transportation system at the traffic sub-region level, Markov chains are applied in the research as a strategy to alleviate recurring congestion.

[0013] The steps to establish a bilevel programming model include:

[0014] S21 integrates OD demands originating from the same traffic sub-region into a macro OD source, and integrates all internal paths into multiple heterogeneous macro paths based on traffic sub-region nodes.

[0015] S22. A Markov chain model based on macroscopic traffic balance is established to predict the occurrence of recurring congestion in the network equilibrium state, and to transfer vehicles at the macroscopic level in congested sub-regions, thereby alleviating traffic congestion. By providing an effective transfer probability matrix to reflect the real-time traffic sub-region carrying capacity, macroscopic path planning between traffic sub-regions is realized.

[0016] S23. Within the traffic sub-region, the shortest path planning algorithm is used. Through the A* path planning algorithm, the local path planning model within the sub-region is established, starting from the boundary of the previous traffic sub-region and ending at the boundary of the next sub-region.

[0017] The macro-path planning strategy determines the traffic congestion status based on the average driving speed within a sub-region over a fixed time period, then determines the probability transition matrix, and uses Markov chains to transfer traffic flow within the traffic sub-region.

[0018] A Markov chain is a stochastic process in which, given the current state, the probability distribution of the future states depends only on the current state and is independent of past states. This property is known as the "Markov property" or "memoryless property".

[0019] Among them, a special class of Markov chains is called "equilibrium Markov chains" or "ergodic Markov chains." In an equilibrium Markov chain, there exists a stable probability distribution. As time approaches infinity, the probability distribution of the system's states converges to this stable probability distribution, regardless of the initial state. This stable probability distribution is called the equilibrium distribution or steady-state distribution.

[0020] A key property of balanced Markov chains is ergodicity, meaning that starting from any initial state, the chain will eventually traverse all possible states and remain in a balanced distribution in the long run. Since the model primarily utilizes the ergodicity property of Markov chains, the research problem is limited to homogeneous Markov chains.

[0021] A state transition function (or probability transition matrix) is a rule that defines the transitions between states in an equilibrium Markov chain. It is usually represented by a matrix called the state transition matrix, where each element represents the probability of transitioning from one state to another.

[0022] The probabilistic state transition function must satisfy the following properties:

[0023] Nonnegativity: Transition probabilities must be nonnegative. This is because they represent the probability of an event occurring, and the probability value must be between 0 and 1.

[0024] Normalization: For each state, the sum of the probabilities of transitioning to all possible states (including itself) must equal 1. This is because a state must transition to some state in the next time step (including possible self-transition).

[0025] Markov property: Transition probabilities depend only on the current state and are independent of past or future states. This is the core property of Markov chains.

[0026] Ergodicity: In an ergodic Markov chain (or an equilibrium Markov chain), any state can be reached in a finite number of steps. Ergodicity guarantees that a Markov chain converges to a steady-state distribution in the long run.

[0027] The handling of recurring traffic congestion includes the following steps:

[0028] S31, using the following matrix:

[0029]

[0030] This represents the traffic flow transition probability matrix for all roads in the current urban traffic network (recalculated every 10 minutes after collecting network information). Within each counting period, the matrix increments by 1 for each vehicle that moves from one traffic sub-zone to another. After collecting data on all passing vehicles within 10 minutes, the transition probability matrix is ​​normalized by summing the values ​​of each row.

[0031] S32, the equilibrium state formula is as follows:

[0032] π T ·P=π T

[0033] Where P is the probability transition matrix obtained from S31, and π is the macroscopic traffic distribution of the current urban road network. A state is called an equilibrium distribution if and only if π satisfies the above equation at a certain moment. This means that the distribution will not change as traffic flow transfers between traffic sub-areas continue. At this point, the row eigenvector of P is:

[0034]

[0035] S33: After obtaining the road network equilibrium state through S32, a relevant speed-flow diagram is drawn based on the average driving speed and flow rate relationship within each traffic sub-zone. This speed-flow diagram is then fitted to obtain the corresponding speed-flow rate relationship. Since the road network within a traffic sub-zone is constant, the average speed within that sub-zone can be obtained given the number of vehicles in that sub-zone. The road network congestion situation is then assessed based on the average speed. When the average driving speed within a region reaches a congestion level, relevant diversion strategies are employed to alleviate the congestion.

[0036] S34, within every ten-minute time interval, based on the speed-flow relationship in S33, the flow transfer probability matrix P in S31 is adjusted. flow Transformed into velocity transition matrix P speed ,Right now:

[0037]

[0038] Where α ij and β ij The set of parameters related to speed and flow in the road network is A = {α} ij ...} and B = {β} ij ...}

[0039] When frequent congestion occurs in the road network, the diversion strategy involves rationally transferring vehicles from high-traffic sub-zones to adjacent low-traffic sub-zones to alleviate congestion. Due to differences in road network capacity and road attributes within sub-zones, the specific values ​​of the traffic transfer matrix cannot be accurately described. Therefore, an indirect method is adopted: the difference in average speed between sub-zones is used to represent the elements of the transfer matrix. The matrix is ​​then processed row-by-row, and the elements are projected to (-1,1) using the tanh function for normalization when calculating probabilities using Markov chains. A new transfer matrix is ​​then obtained row-by-row using the softmax function, enabling traffic flow to shift from high-traffic congested sub-zones to low-traffic sub-zones according to a specific relationship. The formula for the new transfer matrix is ​​as follows:

[0040]

[0041] After calculating the new transition probability matrix, it is summed with the original probability transition function according to certain weights to obtain the final probability transition function. The new road network equilibrium state is then calculated. The formula for the probability transition function is as follows:

[0042]

[0043] Where ρ is the weighting parameter, the value of which is adjusted according to the experimental congestion evacuation effect.

[0044] The beneficial effects of this invention are:

[0045] This invention achieves macro-level vehicle path planning for frequently congested urban road networks through a bi-level programming model and a balanced Markov chain. Simultaneously, by determining the transfer priority between sub-zones under congestion status division, vehicles can avoid congested areas, while traffic flow within congested sub-zones can be diverted with higher priority as quickly as possible. This provides a new approach to alleviating urban congestion. Attached Figure Description

[0046] Figure 1 A flowchart of a macro-path planning method for vehicles in urban road networks prone to frequent congestion;

[0047] Figure 2 The results of dividing the road network within Beijing's Fifth Ring Road into traffic zones based on historical ride-hailing travel data;

[0048] Figure 3 This is a schematic diagram showing the connection relationship between traffic sub-zones within the Second Ring Road. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0050] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0051] refer to Figure 1 As shown, Figure 1 The flowchart presents a macro-path planning method for vehicles in urban road networks prone to frequent congestion, comprising the following steps:

[0052] S10, according to the existing method of dividing traffic sub-zones, such as based on the historical travel data of ride-hailing services, divide the urban road network into several traffic sub-zones;

[0053] refer to Figure 2 As shown, Figure 2 This is a method for dividing the road network within Beijing's Fifth Ring Road into traffic zones based on historical ride-hailing travel data. The Fifth Ring Road is divided into 402 traffic zones, with each traffic zone having an average area of ​​2.15 square kilometers.

[0054] S20. Establish a two-layer planning model: The first layer is the macro-path planning layer, which selects the shortest travel sequence connecting each traffic sub-zone based on the sub-zone where the starting point is located; the second layer is the local path planning layer, which uses the A* path planning algorithm to plan within each traffic sub-zone with the shortest travel time as the optimization objective.

[0055] Specifically, the steps to establish a two-level programming model include:

[0056] S21 integrates OD demands originating from the same traffic sub-region into a macro OD source, and integrates all internal paths into multiple heterogeneous macro paths based on traffic sub-region nodes.

[0057] S22. Establish a Markov chain model based on macro-flow equilibrium to predict the occurrence of frequent congestion in the network equilibrium state, and transfer vehicles at the macro level in congested sub-regions to alleviate traffic congestion. By providing an effective transfer probability matrix to reflect the real-time traffic sub-region carrying capacity, macro-path planning between traffic sub-regions is realized.

[0058] S23. Within the traffic sub-region, the shortest path planning algorithm is used. Through the A* path planning algorithm, the local path planning model within the sub-region is established, starting from the boundary of the previous traffic sub-region and ending at the boundary of the next sub-region.

[0059] S30, addressing recurring congestion: To resolve recurring congestion in the current transportation system at the traffic sub-region level, Markov chains are applied in the research as a strategy to alleviate recurring congestion.

[0060] Specifically, the handling of recurring congestion includes the following steps:

[0061] S31, using the following matrix:

[0062]

[0063] This represents the traffic flow transition probability matrix for all roads in the current urban traffic network (recalculated every 10 minutes after collecting network information). Within each counting period, the matrix increments by 1 for each vehicle that moves from one traffic sub-zone to another. After collecting data on all passing vehicles within 10 minutes, the transition probability matrix is ​​normalized by summing the values ​​of each row.

[0064] S32, the equilibrium state formula is as follows:

[0065] π T ·P=π T

[0066] Where P is the probability transition matrix obtained from S31, and π is the macroscopic traffic distribution of the current urban road network. A state is called an equilibrium distribution if and only if π satisfies the above equation at a certain moment. This means that the distribution will not change as traffic flow transfers between traffic sub-areas continue. At this point, the row eigenvector of P is:

[0067]

[0068] S33: After obtaining the road network equilibrium state through S32, a relevant speed-flow diagram is drawn based on the average driving speed and flow rate relationship within each traffic sub-zone. This speed-flow diagram is then fitted to obtain the corresponding speed-flow rate relationship. Since the road network within a traffic sub-zone is constant, the average speed within that sub-zone can be obtained given the number of vehicles in that sub-zone. The road network congestion situation is then assessed based on the average speed. When the average driving speed within a region reaches a congestion level, relevant diversion strategies are employed to alleviate the congestion.

[0069] S34, within every ten-minute time interval, based on the speed-flow relationship in S33, the flow transfer probability matrix P in S31 is adjusted.flow Transformed into velocity transition matrix P speed ,Right now:

[0070]

[0071] Where α ij and β ij The set of parameters related to speed and flow in the road network is A = {α} ij ...} and B = {β} ij ...}

[0072] When frequent congestion occurs in the road network, the diversion strategy involves rationally transferring vehicles from high-traffic sub-zones to adjacent low-traffic sub-zones to alleviate congestion. Due to differences in road network capacity and road attributes within sub-zones, the specific values ​​of the traffic transfer matrix cannot be accurately described. Therefore, an indirect method is adopted: the difference in average speed between sub-zones is used to represent the elements of the transfer matrix. The matrix is ​​then processed row-by-row, and the elements are projected to (-1,1) using the tanh function for normalization when calculating probabilities using Markov chains. A new transfer matrix is ​​then obtained row-by-row using the softmax function, enabling traffic flow to shift from high-traffic congested sub-zones to low-traffic sub-zones according to a specific relationship. The formula for the new transfer matrix is ​​as follows:

[0073]

[0074] After calculating the new transition probability matrix, it is summed with the original probability transition function according to certain weights to obtain the final probability transition function. The new road network equilibrium state is then calculated. The formula for the probability transition function is as follows:

[0075]

[0076] Where ρ is the weighting parameter, the value of which is adjusted according to the experimental congestion evacuation effect.

[0077] Based on congestion levels, weighting parameters are determined for different states. For a traffic sub-zone, firstly, a baseline speed needs to be established, which is typically the average speed of that area when there is no congestion. Then, based on this baseline speed, the congestion level can be divided into several levels as follows:

[0078] Smooth: Average speed is greater than or equal to 90% of the baseline speed. This indicates that the traffic conditions in the area are very smooth and there is no congestion.

[0079] Mild congestion: Average speed is between 70% and 90% of the baseline speed. This indicates that traffic in the area is starting to get congested, but relatively high speeds can still be maintained.

[0080] Moderate congestion: Average speed is between 50% and 70% of the baseline speed. This indicates that traffic conditions in the area are relatively congested, and driving speeds have decreased.

[0081] Severe congestion: Average speed is between 30% and 50% of the baseline speed. This indicates that traffic conditions in the area are very congested, and driving speeds have dropped significantly.

[0082] Extremely congested: Average speed is less than 30% of the baseline speed. This indicates that traffic conditions in the area are extremely congested and travel speeds are very slow.

[0083] The above classification method is based on relative speed changes, not absolute speed. This means that even if the baseline speeds of two areas are different, the same method can be used to classify congestion levels. Weighting parameters for the five congestion states are determined experimentally to address different congestion states within each sub-area, thereby determining different priorities for congestion management within each sub-area. Specifically, the higher the congestion level within a given time period, the higher the flow rate within that sub-area is prioritized for transfer. Through this macro-level congestion management method with different priorities, rapid management of recurring congestion can be achieved.

[0084] Through the above embodiments, the present invention realizes a macro-path planning method for vehicles in urban road networks with frequent congestion, which effectively addresses urban traffic congestion problems and improves the accuracy and real-time performance of path planning.

[0085] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0086] It should be noted that the terms "first," "second," and "third" used in the embodiments of this application are merely to distinguish similar objects and do not represent a specific order of objects. It is understood that "first," "second," and "third" can be interchanged in a specific order or sequence where permitted. It should be understood that the objects distinguished by "first," "second," and "third" can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in an order other than those illustrated or described herein.

[0087] The terms "comprising" and "having," and any variations thereof, in this application are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or device that includes a series of steps or modules is not limited to the steps or modules listed, but may optionally include steps or modules not listed, or may optionally include other steps or modules inherent to such processes, methods, products, or devices.

[0088] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A macro-path planning method for vehicles in urban road networks prone to frequent congestion, characterized in that, Includes the following steps: S10, based on the existing method of dividing traffic sub-zones and the historical travel data of ride-hailing services, divides the urban road network into several traffic sub-zones. S20. Establish a two-layer planning model: The first layer is the macro-path planning layer, which selects the shortest travel sequence connecting each traffic sub-zone based on the sub-zone where the starting point is located; the second layer is the local path planning layer, which uses the A* path planning algorithm to plan within each traffic sub-zone with the shortest travel time as the optimization objective. S30, addressing recurring congestion: In order to solve the recurring congestion in the current traffic system at the traffic sub-region level, Markov chains are applied to the research as a strategy to alleviate recurring congestion. The specific steps for addressing recurring congestion include: S31, using the following matrix: ， This represents the traffic flow transition probability matrix for all roads in the current urban traffic network. The matrix is ​​recalculated every 10 minutes by collecting new network information. Within each counting time period, each time a motor vehicle moves from a traffic sub-zone... Transfer to the traffic sub-zone Correspondingly Then it will be incremented by 1; after collecting data on all passing vehicles within 10 minutes, the transition probability matrix is ​​normalized by summing each row; S32, the equilibrium state formula is as follows: ， Where 𝑃 is the probability transition matrix obtained from S31, and 𝜋 is the macroscopic traffic distribution of the current urban road network; and 𝜋 is said to satisfy the above formula if and only if 𝜋 satisfies the above formula at a certain moment. This is a balanced distribution; that is, as traffic flow transfers between traffic sub-regions proceed, This state will not change further; at this point, That is, the row feature vector of 𝑃, and we have: ； S33: After obtaining the road network equilibrium state through S32, draw a relevant speed-flow diagram based on the average driving speed and flow relationship within each traffic sub-zone; fit the speed-flow relationship diagram to obtain the corresponding speed-flow relationship; since the road network within the traffic sub-zone is constant, the average speed within the sub-zone is obtained given the number of vehicles in the sub-zone; the road network congestion is judged based on the average speed; when the average driving speed in the area reaches the congestion level, relevant diversion strategies are adopted to alleviate the congestion. S34, within every 10-minute time interval, based on the speed-flow relationship described in S33, adjust the flow transfer probability matrix in S31. Transformed into a velocity transition matrix ,Right now: ， in and This belongs to the set of parameters relating speed and flow in the road network. and ; When the road network experiences frequent congestion, the diversion strategy involves rationally transferring vehicles from high-traffic sub-zones to adjacent low-traffic sub-zones to alleviate congestion. Due to differences in road network capacity and road attributes within sub-zones, the specific values ​​of the traffic transfer matrix cannot be accurately described. Therefore, an indirect method is adopted: the difference in average speed between sub-zones is used to represent the elements of the transfer matrix. The matrix is ​​then processed row-by-row, and the tanh function is used to project the elements to (-1,1) for normalization when calculating probabilities using Markov chains. Finally, the softmax function is used row-by-row to obtain a new transfer matrix, which realizes the transfer of traffic from high-traffic congestion sub-zones to low-traffic sub-zones according to a certain relationship. The formula for the new transfer matrix is ​​as follows: ， The new transition probability matrix is ​​calculated. Then, the result is summed with the original probability transfer function according to certain weights to obtain the final probability transfer function. And calculate the new road network equilibrium state, probability transition function. The formula is as follows: ， in This is a weighting parameter, the value of which is adjusted based on the experimental congestion evacuation effect.

2. The method for macro-path planning of vehicles in urban road networks with frequent congestion as described in claim 1, characterized in that, The steps to establish a bilevel programming model include: S21 integrates OD demands with starting points located within the same traffic sub-region into a macro OD source, and integrates all internal paths into multiple heterogeneous macro paths based on traffic sub-region nodes; S22. Establish a Markov chain model based on macro-flow equilibrium to predict the occurrence of frequent congestion in the network equilibrium state, and transfer vehicles at the macro level in congested sub-regions to alleviate traffic congestion. By providing an effective transfer probability matrix to reflect the real-time traffic sub-region carrying capacity, macro-path planning between traffic sub-regions is realized. S23. Within the traffic sub-region, the shortest path planning algorithm is used. Through the A* path planning algorithm, the local path planning model within the sub-region is established, starting from the boundary of the previous traffic sub-region and ending at the boundary of the next sub-region.

3. The method for macro-path planning of vehicles in urban road networks with frequent congestion as described in claim 2, characterized in that, The shortest path planning algorithm is used within the traffic sub-region, and the planning is carried out through the following process: First, based on the arrival of vehicles in the previous traffic sub-zone at the boundary point of this traffic sub-zone as the starting point and the boundary of the next traffic sub-zone as the endpoint set, a heuristic function is defined using real-time traffic status. Each intersection is treated as a node, and a node set is created to store the nodes to be explored. Initially, it only contains the starting point, and the distance of the starting point is set to 0. An optimal node is selected from the node set, that is, the node with the minimum estimated time, based on the total estimated time of the nodes. Then expand the selected node, that is, consider its adjacent reachable nodes, calculate the total estimated time for expanding the node, and update the node's distance and parent node information. Then, based on the new node distance and the total estimated time, update the information of the corresponding nodes in the node set; If a point in the destination set is found, the shortest path from the starting point to the destination can be obtained by tracing back the parent node information of each node. Finally, the shortest path is output as the planning result within this traffic sub-zone.

4. The method for macro-path planning of vehicles in urban road networks with frequent congestion as described in claim 1, characterized in that, The probability transfer function The determination of the weighting parameters is based on the congestion status, defining the weighting parameters for different statuses. For a traffic sub-zone, firstly, a baseline speed needs to be determined, which is the average speed of that area when there is no congestion. Then, based on this baseline speed, the congestion level is divided into the following levels: Smooth: The average speed is greater than or equal to 90% of the baseline speed, which indicates that the traffic conditions in the area are very smooth and there is no congestion. Mild congestion: Average speed is between 70% and 90% of the baseline speed, indicating that traffic in the area is starting to get congested, but a relatively high driving speed can still be maintained; Moderate congestion: Average speed is between 50% and 70% of the base speed, indicating that traffic conditions in the area are relatively congested and driving speeds have decreased; Severe congestion: Average speed is between 30% and 50% of the base speed, indicating that traffic conditions in the area are very congested and driving speeds have dropped significantly; Extreme congestion: Average speed is less than 30% of the baseline speed, which indicates that the traffic conditions in the area are extremely congested and the speed of travel is very slow. The above classification method is based on relative speed changes, not absolute speed; this means that even if the baseline speeds of two regions are different, the same method can be used to classify congestion levels. The weighting parameters for the five congestion states were determined experimentally. The system addresses different congestion levels within each sub-zone by determining different priorities for congestion mitigation within that sub-zone. Specifically, the higher the level of congestion within a given time period, the higher the flow rate within that sub-zone is prioritized for transfer. Through macro-level mitigation methods with different priorities, the system aims to achieve rapid mitigation of recurring congestion.

Images