Road network boundary intersection signal timing scheme design based on feedback control

Abstract

The invention relates to the technical field of traffic signal control, in particular to a road network boundary intersection signal timing scheme design based on feedback control. Constructing a road network undirected graph by taking each road section in the road network as a node; dividing nodes in the road network undirected graph to obtain a plurality of sub-regions; determining the number of the divided sub-regions according to the homogeneity and density of each sub-region; obtaining the traffic flow in each sub-region at the current moment, and determining the traffic state of each sub-region; establishing a corresponding feedback control measure according to the traffic state of each sub-region; utilizing a machine learning algorithm to predict the traffic flow in each sub-region at the next moment, and determining the traffic state corresponding to each sub-region at the next moment; and according to the feedback control measure corresponding to each traffic state, regulating and controlling a boundary intersection signal in the sub-region under the corresponding traffic state. The established feedback control mechanism can optimize traffic light signal timing of the road network boundary, so that the efficiency of traffic flow is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic signal control, and particularly relates to a signal timing scheme design for road network boundary intersections based on feedback control. Background Art

[0002] With the development of cities, the scale of the road network has expanded, the number of urban vehicles has been increasing, and regional congestion has occurred frequently, making the already complex traffic network even more difficult to manage. To address urban traffic problems, urban traffic coordination control strategies have been proposed and continuously improved. The spatial distribution of traffic flow determines that the coordination control strategy must have a reasonable target area. Traffic sub-zone division plays an important role in coordination control and directly affects the efficiency and performance of traffic coordination control strategies. The existing traffic light control systems mainly rely on static data, lack real-time monitoring and feedback, resulting in slow response speed, insufficient prediction ability, and poor flexibility and adaptability in the face of sudden traffic flow changes. These deficiencies make it difficult for traditional signal control to effectively cope with peak hours and traffic congestion problems.

[0003] In view of the above problems, the present invention proposes a method for adjusting traffic light signals based on feedback control. By real-time monitoring the number of vehicles in the road network, setting reasonable thresholds and allowing fluctuations, and predicting the vehicle flow in the next time interval, when it is predicted that the number of vehicles entering the road network in the next stage is large, the traffic light signal timing is automatically adjusted to optimize traffic flow and improve traffic efficiency. Summary of the Invention

[0004] In order to solve the problems existing in the prior art, the present invention provides a signal timing scheme design for road network boundary intersections based on feedback control, including: constructing an undirected road network graph with each road section in the road network as a node; dividing the nodes in the undirected road network graph to obtain multiple sub-regions; determining the number of sub-regions after division according to the homogeneity and density of each sub-region; obtaining the traffic flow inside each sub-region at the current moment to determine the traffic state of each sub-region; establishing corresponding feedback control measures according to the traffic state of each sub-region; using a machine learning algorithm to predict the traffic flow inside each sub-region at the next moment and determine the traffic state of each sub-region at the next moment; and regulating the boundary intersection signals in the sub-regions corresponding to each traffic state according to the feedback control measures corresponding to each traffic state. The feedback control mechanism established by the present invention can optimize the traffic light signal timing at the road network boundary, thereby improving the efficiency of traffic flow.

[0005] The present invention adopts the following technical solutions. A signal timing scheme design for road network boundary intersections based on feedback control includes: Constructing an undirected road network graph with each road section in the road network as a node; Using the NCut algorithm to divide the nodes in the undirected road network graph to obtain multiple sub-regions; Evaluate sub - regions according to the homogeneity and density of each sub - region to determine the number of sub - regions after division; Obtain the traffic flow within each sub - region at the current moment, and determine the traffic state of each sub - region according to the traffic flow within at the current moment; Establish feedback control measures corresponding to each traffic state according to the traffic state of each sub - region; Use a machine learning algorithm to predict the traffic flow within each sub - region at the next moment, and determine the traffic state corresponding to each sub - region at the next moment; Regulate the boundary intersection signals in the sub - regions in the corresponding traffic state according to the feedback control measures corresponding to each traffic state.

[0006] Furthermore, construct an undirected road network graph with each road segment in the road network as a node, specifically: Obtain the traffic operation similarity of adjacent road segments in the road network according to the density of each road segment in the road network; Establish the Laplacian matrix of the road network according to the traffic operation similarity of adjacent road segments in the road network; Obtain the undirected road network graph according to the Laplacian matrix.

[0007] Furthermore, use the NCut algorithm to divide the nodes in the undirected road network graph, specifically: Use the NCut algorithm to obtain the similarity between nodes in the undirected road network graph; Calculate the between - group similarity and between - group correlation of nodes in the undirected road network graph according to the similarity between nodes in the undirected road network graph; Establish a mathematical model for the node division target, and solve the mathematical model according to the between - group similarity and between - group correlation to obtain multiple divided sub - regions; Furthermore, after dividing the nodes in the undirected road network graph to obtain multiple sub - regions, it also includes: Obtain the homogeneity of each sub - region according to the number of road segments in each sub - region; Obtain the first density difference within each sub - region and the second density difference between each sub - region and its adjacent sub - regions according to the road segment density in each sub - region; Obtain the division evaluation index of each sub - region according to the homogeneity, the first density difference, and the second density difference; Divide the undirected road network graph multiple times, and determine the optimal number of sub - regions according to the division evaluation index of each sub - region after each division.

[0008] Furthermore, determining the traffic state of each sub - region according to the traffic flow within at the current moment includes: Set the critical cumulative vehicle number within each sub - region , at this time, the maximum traffic flow within the sub-region is ; When the current traffic flow within the sub-region , the traffic state of this sub-region is in free flow and stable flow states; When the current traffic flow within the sub-region , the traffic state of this sub-region is in unstable flow state; When the current traffic flow within the sub-region , the traffic state of this sub-region is in forced flow state.

[0009] Furthermore, feedback control measures corresponding to each traffic state are established according to the traffic state of each sub-region, including: When the sub-region is in free flow and stable flow states, no boundary control is required; When the sub-region is in unstable flow state, boundary control is performed on some nodes and road segments in the sub-region until the traffic state is transformed into free flow and stable flow states, and then the boundary control is stopped; When the sub-region is in forced flow state, boundary control is performed on each node and road segment in the sub-region until the traffic state is transformed into free flow and stable flow states, and then the boundary control is stopped.

[0010] Furthermore, the method for predicting the traffic flow within each sub-region at the next moment is: using the autoregressive integrated moving average model to predict the traffic flow within each sub-region at the next moment; using the mean square error and mean absolute error indicators to evaluate the prediction results of the model.

[0011] The beneficial effects of the present invention are: The present invention takes feedback control as the core, divides the road network into sub-regions according to the traffic homogeneity principle, ensures that each sub-region should show a consistent density level internally, and at the same time shows an obvious density difference from adjacent regions. The NCut algorithm can effectively isolate each road segment, support balanced partitioning while avoiding the emergence of isolated high-density regions. Furthermore, the Laplacian matrix is introduced for density similarity measurement, which helps to determine which regions may exhibit similar traffic behaviors. In terms of feedback control, a control algorithm is designed based on real-time data and prediction results. When the predicted number of incoming vehicles exceeds the set threshold, the signal timing of the traffic lights is automatically adjusted to extend the green light time or shorten the red light time, thereby controlling the number of vehicles entering the road network, effectively optimizing the traffic flow and improving the traffic efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0013] Figure 1 It is a schematic diagram of the design process of the signal timing scheme for the road network boundary intersection based on feedback control in the embodiments of the present invention; Figure 2 It is a schematic diagram of the control idea of the feedback control measure in the embodiments of the present invention; Figure 3 It is a schematic diagram of the principle of the signal regulation method for the boundary intersection in the embodiments of the present invention. Specific embodiments

[0014] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0015] A schematic diagram of the design process of the signal timing scheme for the road network boundary intersection based on feedback control in the embodiments of the present invention is as Figure 1 shown and includes: Construct an undirected graph of the road network with each road section in the road network as a node; In the embodiments of the present invention, based on graph theory, the road section is used as a node in the undirected graph of the road network and the division of the road section is realized by dividing the node. The traffic operation similarity between adjacent road sections i and j is expressed based on the road section density as follows: In the formula, where, is the traffic operation similarity between road sections i and j, is the density of road section i in the road network , is the density of road section j in the road network , is the exponential function.

[0016] Map the result in the above formula to the interval of 0-1. When , it indicates that the traffic operation similarity between road sections is the lowest. On the contrary, when When it is [a certain value], it indicates that the traffic operation similarity between road segments is the highest; since the traffic operation similarity between road segments is within the range of 0 - 1 and the larger the value, the higher the similarity, so the similarity can be used to represent the weight of the arc segments in the figure, and then by using the traffic operation similarity to construct the Laplacian matrix of the road network ; when there is a directed connection between road segments i and j in the road network , : when road segments i and j are not connected, This ensures the situation when some road segments are one-way; when is [a certain value], , the weighted adjacency matrix of the road network is , The elements in it are: In the formula, there are two value-taking situations for the elements in the weighted adjacency matrix W. When is [a certain value], , is an element in the adjacency matrix. At this time, takes the weight value . Let , and the matrix is a diagonal matrix, then there is: Among them, represents the diagonal matrix composed of elements . The above formula means that the main diagonal element of the diagonal matrix D is the sum of the weights of the edges connecting node i to the jth node different from the ith node, that is, the ith diagonal element of the diagonal matrix D is the sum of the weights of all the edges of node i Use the NCut algorithm to partition the nodes in the undirected graph of the road network to obtain multiple sub-regions; NCut, that is, the Normalized Cut algorithm, is a graph partitioning algorithm. The goal of graph partitioning is to divide the nodes in the graph into multiple non-overlapping subsets, making the connections between different subsets as sparse as possible, while the connections between nodes within the same subset are as tight as possible. The specific content of this algorithm can refer to the prior art.

[0017] In the embodiments of the present invention, the division of the road network sub - regions is extremely important, and its results will affect the effect of boundary control to a certain extent. When dividing the road network sub - regions, the principles that the sub - regions should follow are "homogeneity", a certain scale, a moderate number, and connectivity. Only a road network with "homogeneity" can ensure a stable and clear macroscopic fundamental diagram (MFD), which is the premise for using the MFD theory for boundary control. Therefore, for the divided sub - regions, it is necessary to ensure that the density of each road segment within a sub - region is similar, and the density difference from other sub - regions is large. Essentially, MFD is a regular change curve formed by aggregating traffic parameters of road segments in the road network. Therefore, the area of the divided sub - regions should be moderate, which will be more conducive to grasping the curve law of MFD. The number of divided sub - regions should be moderate. This is because in the process of implementing the boundary control strategy, if the number of sub - regions is too large, the operation efficiency will be greatly reduced; on the contrary, it is difficult to improve the overall operation efficiency of the road network. The interior of the divided sub - regions should be connected, and adjacent sub - regions should also be closely connected in space to facilitate the implementation of control measures.

[0018] Using the NCut algorithm is likely to divide out isolated individual nodes, so that isolated individual road segments are divided out when dividing the road network. To avoid this situation, it is necessary to ensure that the density similarity of road segments divided into the same sub - region is high, and the density similarity of road segments divided into different sub - regions is low. In the embodiments of the present invention, it is set that the node set is divided into two parts and , and their similarity is defined as . The optimal division result is . The inter - group similarity and the intra - group correlation are calculated as shown in the following formulas respectively: In the formulas, is the inter - group similarity, is the intra - group correlation, represents the similarity between partial node set A and partial node set B, represents the similarity between partial node set A and node set V, represents the similarity between partial node set B and node set V, , represent the similarity within partial node set A and the similarity within partial node set B.

[0019] It can be seen that the above two formulas can be mutually transformed into: .

[0020] Establish a mathematical model for dividing the target, expressed as: In the formula, means to find a vector x such that obtains the minimum value, where x is a column vector composed of 1 and -1, and i is the i-th node included in the undirected graph in the formula. Evaluate the sub-regions according to the homogeneity and density of each sub-region to determine the number of sub-regions after division; When the NCut algorithm is used for road network division, it can ensure the optimal result for each division, but it cannot determine the number of divided sub-regions, making it difficult to determine the final road network division scheme. Therefore, in order to determine the number of divided sub-regions and evaluate the sub-region division results under different division numbers, the embodiments of the present invention introduce the average value of the homogeneity in each sub-region as the evaluation index: In the formula, , are the number of road segments in sub-regions A and B, is the average value of the homogeneity of sub-region A and sub-region B, represents the value of the i-th element in sub-region A, represents the value of the j-th element in sub-region B.

[0021] The above formula only includes traffic flow data and can quantify the density differences between different sub-intervals, but it is difficult to quantify the density differences within the sub-regions. Therefore, the embodiments of the present invention further propose to evaluate the overall division effect of each region itself after division. The specific calculation formula is as follows: In the formula, is the sub-region adjacent to sub-region A, is the sub-region inside the density difference, is the density difference between sub-region A and its adjacent sub-region; according to the division principle, the difference within the same sub-region should be minimized, and the difference between different sub-regions should be maximized. The smaller the value, the better the division effect; finally, the division evaluation index of the overall road network division result is obtained by calculating the average value of each region: In the formula, is the number of sub - regions after division, is the set of sub - regions after division.

[0022] Obtain the traffic flow inside each sub - region at the current moment, and determine the traffic state of each sub - region according to the traffic flow inside at the current moment; For each sub - region, there is a critical cumulative number of vehicles inside it such that the sub - region has the maximum completed traffic flow When the traffic flow , the sub - region is mostly in the free - flow and stable - flow states and is in a non - congested state; when continues to increase until , the sub - region is mostly in the unstable - flow state and is in a saturated state; when , As increases, it will decrease, and the sub - region is mostly in the forced - flow state and is in a congested state.

[0023] Establish feedback control measures corresponding to each traffic state according to the traffic state of each sub - region; According to the different traffic states of the regional road network, the boundary control should also be different. In the embodiments of the present invention, it can be divided into the following three types: State 1: Free - flow and stable - flow stage; There is no need to perform boundary control on the controlled area because the road network is already in the best expected state at this time. Control measures cannot further improve the traffic operation state of the area and will also cause waste of management resources; State 2: In the second half of the stable - flow stage, there is still some room for improvement in the traffic operation state. The traffic volume of key nodes and sections in the area is gradually increasing. When the cumulative value of regional vehicles reaches near the critical value, the current control effect should be maintained until the road network state returns to the free - flow stage, then the control is turned off; It should be noted that when in the first half of the unstable - flow stage, in a certain period of time, some key nodes and sections in the sub - region are already in a congested state. At this time, boundary control is also required to change the traffic volume in the sub - region so that the cumulative number of vehicles in the sub - region satisfies the formula and avoid entering the forced - flow stage; State 3: The second half of the unstable - flow stage and forced - flow: Urgently need to take boundary control to change the traffic volume in the sub - region until the road network state returns to the free - flow stage, then the control is turned off.

[0024] In the embodiments of the present invention, according to the division result of the sub - regions, select the controlled areas prone to congestion, and regard the remaining sub - regions as peripheral areas. Use signal timing control to control the traffic flow entering the controlled areas through boundary control, so as to achieve the control goal, as Figure 2 shown, specifically: The total traffic flow transfer is divided into two parts. One part is the traffic flow entering the controlled area, and the other part is the traffic flow leaving the peripheral area Among them, the traffic flow entering the controlled area is is the traffic flow entering the controlled area from the signal-controlled intersection, which is controllable. Specifically, the traffic flow that should enter the controlled area is obtained through boundary control and then the actual inflow is obtained through a specific signal control scheme to achieve the boundary control goal; while the traffic flow leaving the outer area , part of it passes through boundary control to obtain the traffic flow entering the controlled area , and the other part bypasses the controlled area through traffic guidance and other means. At the same time, traffic information collection work is carried out in the two areas in each control cycle, and the boundary control is adjusted and changed through data analysis to form a feedback closed-loop control.

[0025] By controlling the boundary flow to adjust the internal flow of the area, the cumulative number of vehicles on the road network in the area can always be maintained near the critical traffic volume , which can ensure that the area has the maximum flow, achieve regional congestion alleviation, and keep the traffic running smoothly. The cumulative number of vehicles in the controlled area after traffic control should meet: It should satisfy: In the formula, is the critical cumulative vehicle number of area ; is the fluctuation range of traffic control. Usually, it takes 1% - 3% of

[0026] Use machine learning algorithms to predict the traffic flow in each sub-area at the next moment and determine the traffic state corresponding to each sub-area at the next moment; For the method of predicting the traffic flow in each sub-area at the next moment, in the embodiments of the present invention, an autoregressive moving average model or an autoregressive integrated moving average model can be selected. Taking the autoregressive integrated moving average model as an example: Collect past vehicle flow data, denoted as , where is the time index (for example, the number of vehicles per minute). The specific content of this model includes: Autoregressive model (AR): The model form is expressed as: Where: is the constant term, are the autoregressive coefficients, and p is the autoregressive order, indicating that the current value depends on the observations of the past p time points, is white noise, represents Vehicle flow data at a moment; Moving average model (MA): The model form is expressed as: Where: is the constant term, is the moving average coefficient, q is the moving average order, is white noise, represents white noise at the moment; Autoregressive integrated moving average model (ARIMA): When the data is not stationary, first-order differencing is required. The form of the ARIMA model is: Where: B is the lag operator, d is the number of differencing times (the number of times to make the data stationary), and are the autoregressive and moving average polynomials respectively; Through the trained model, the vehicle flow in the next time interval t + 1 can be predicted. At the same time, the accuracy of the prediction is evaluated by indicators such as MSE (mean square error) and MAE (mean absolute error).

[0027] According to the feedback control measures corresponding to each traffic state, the boundary intersection signals in the sub-region corresponding to the traffic state are regulated.

[0028] In the embodiment of the present invention, by adjusting the intersection signal timing at the boundary of the controlled sub-region, the entry and exit of vehicles at the boundary are controlled, so as to maintain the passenger value in the road network at the optimal critical value nearby, ensure the optimal road network efficiency inside the sub-region, relieve the congestion state, and the principle of signal timing adjustment is as Figure 3 shown, specifically: System input: The optimal critical cumulative passenger volume in the traffic control sub-region ; System output: The total number of passengers in the sub-region ; represents the relationship between the number of passengers in the road network and the interpolation of signal timing adjustment at the intersection; When the system output stabilizes at the optimal critical value , the number of passengers entering and leaving the control sub-region is equal, and at this time, the traffic state inside the sub-region remains in the optimal state.

[0029] In the embodiment of the present invention, boundary control adopts asymmetric traffic signal control. Asymmetric signal control mainly means that the traffic timing of traffic signal control lights in opposite driving directions is asymmetric. In daily life, most intersection signal lights are symmetrically designed. However, due to the uneven distribution of traffic flow in the city in terms of time and space, in many cases, the traffic flow in a certain direction cannot fully and effectively utilize all the green light time of this phase, which results in a waste of green light time. At this time, adopting asymmetric traffic signal control can effectively improve the utilization efficiency of green light time and the service level of intersections. The embodiment of the present invention adopts asymmetric traffic signal control to adjust the proportion of different vehicles entering and leaving at the boundary of the control sub-region, and keeps the number of passengers in the sub-region near the optimal critical value, so as to maintain a stable traffic state in the sub-region.

[0030] In the phase modification plan, the green light for entering the sub-region will be switched to red in advance to prohibit some vehicles from entering the sub-region through the boundary. In addition, it is also necessary to consider how long in advance the phase switch needs to be; it is known that the difference between the number of passengers in the road network and the optimal critical value in the th cycle : In the formula, is the optimal critical value. Allocate to the boundary intersections according to the proportion of the number of passengers entering each boundary intersection in the total number of passengers entering the boundary intersections in the th cycle. Taking the number of passengers allocated to boundary intersection 7 as an example: In the formula, represents the number of passengers entering boundary intersection 7 in the road network in the th cycle, represents the optimal critical value of the number of passengers entering the road network in the th cycle In the th cycle, the number of passengers entering the sub-region from boundary intersection 7 is: represents the signal timing of boundary intersection 7 in the road network in the th cycle; Therefore, in the th cycle, the adjustment value of the signal timing for boundary intersection 7 to enter the sub-region is: In the th cycle, the signal timing for boundary intersection 7 to enter the sub-region is: The embodiments of the present invention further impose the following restrictions on the timing plan, including: Minimum green light time limit: Its main purpose is to ensure that pedestrians can pass through the intersection during the minimum green light time and at the same time provide a certain reaction time for drivers. The expression is: Wherein, represents the green light signal time (s) when pedestrians pass through the intersection, represents the pedestrian reaction time, generally 2 - 3 s, represents the number of pedestrians queuing to cross the street, represents the length (m) of the crosswalk, represents the walking speed (m / s) of pedestrians, and the average walking speed of pedestrians is 1.2 m / s.

[0031] Maximum green light time limit: Since there is a certain time limit for the cycle of the intersection signal lights and at the same time to avoid too long waiting time for pedestrians or vehicles in the opposite direction, it is necessary to set a maximum green light time to limit the green light time. Specifically, it can be calculated by subtracting the minimum green light time of the oncoming traffic signal from the signal cycle.

[0032] Queue length limit outside the boundary: Since the boundary control strategy is adopted in the embodiments of the present invention, there will be some vehicles queuing and waiting on the road section outside the boundary of the control sub - area due to restricted entry. To avoid the queue length spreading to the upstream intersection and causing upstream traffic congestion or even large - area spread, therefore, a boundary queue length limit is added. When the boundary queue length exceeds the limit condition, the green light time in this direction is increased to release more vehicles into the control sub - area.

[0033] The calculation model of the queue length outside the boundary adopts a vehicle queue length prediction model, which is a linear prediction model derived based on traffic wave theory and is expressed as follows: Wherein, represents the vehicle queue length at time represents the vehicle queue length at time represents the speed of the stopping wave.

[0034] The stopping wave model is as follows: Wherein, represents the driving speed of the vehicle in the free - flow state, represents the traffic flow density on the road section at this time, It represents the congestion density, that is, the traffic flow density when the traffic is completely blocked and the vehicle flow is stationary. The negative sign indicates that the propagation direction of the stop wave is opposite to the vehicle driving direction.

[0035] When the vehicle queue length on the road section outside the control sub-region boundary is greater than the road section length, it indicates that traffic overflow has occurred on the current road section, and it is necessary to lift the green light restriction in the direction of the boundary intersection to speed up the release of the queued vehicles on this road section.

[0036] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A design of a signal timing scheme for road network boundary intersections based on feedback control, characterized in that: include: Use each road section in the road network as a node to construct an undirected graph of the road network; The NCut algorithm is used to divide the nodes in the road network undirected graph to obtain multiple sub-areas; The sub-regions are evaluated according to their homogeneity and density to determine the number of sub-regions after division; Obtain the current traffic volume in each sub-area, and determine the traffic status of each sub-area based on the current traffic volume; Establish feedback control measures corresponding to each traffic state according to the traffic state of each sub-area; Use machine learning algorithms to predict the traffic volume in each sub-area at the next moment and determine the traffic status of each sub-area at the next moment; According to the feedback control measures corresponding to each traffic state, the boundary intersection signals in the sub-area under the corresponding traffic state are regulated.

2. The design of a signal timing scheme for road network boundary intersections based on feedback control according to claim 1 is characterized in that: Each road segment in the road network is used as a node to construct an undirected graph of the road network, specifically: Obtain the traffic operation similarity of adjacent road sections in the road network according to the density of each road section in the road network; The Laplace matrix of the road network is established according to the traffic operation similarity of adjacent road sections in the road network; A road network undirected graph is obtained according to the Laplace matrix.

3. The design of a signal timing scheme for road network boundary intersections based on feedback control according to claim 1 is characterized in that: The NCut algorithm is used to divide the nodes in the undirected graph of the road network, specifically: Use NCut algorithm to obtain the similarity between nodes in the undirected graph of the road network; According to the similarity between nodes in the road network undirected graph, the inter-group similarity and inter-group correlation of nodes in the road network undirected graph are calculated; A mathematical model of the node division target is established, and the mathematical model is solved according to the inter-group similarity and inter-group correlation to obtain a plurality of sub-areas after division.

4. The design of a signal timing scheme for road network boundary intersections based on feedback control according to claim 1 is characterized in that: After dividing the nodes in the road network undirected graph into multiple sub-areas, it also includes: Obtain the homogeneity of each sub-area according to the number of road sections in each sub-area; According to the road segment density in each sub-region, a first density difference within each sub-region and a second density difference between each sub-region and its adjacent sub-region are obtained; Acquire a division evaluation index for each sub-region according to the homogeneity, the first density difference, and the second density difference; The road network undirected graph is divided multiple times, and the optimal number of sub-areas is determined according to the division evaluation index of each sub-area after each division.

5. The design of a signal timing scheme for road network boundary intersections based on feedback control according to claim 1 is characterized in that: The traffic status of each sub-area is determined based on the current internal traffic flow, including: Set the critical cumulative number of vehicles in each sub-area , the maximum flow in the sub-area is ; When the current traffic volume in the sub-area When , the traffic state of the sub-area is free flow and stable flow; When the current traffic volume in the sub-area When , the traffic state of the sub-area is an unstable flow state; When the current traffic volume in the sub-area When , the traffic state of the sub-area is forced flow state.

6. The design of a signal timing scheme for road network boundary intersections based on feedback control according to claim 5 is characterized in that: According to the traffic status of each sub-area, feedback control measures corresponding to each traffic status are established, including: When the sub-area is in free flow and steady flow state, no boundary control is required; When a sub-area is in an unstable flow state, boundary control is performed on some nodes and road sections in the sub-area until the traffic state is transformed into a free flow and a stable flow state, and then the boundary control is stopped; When a sub-area is in a forced flow state, boundary control is performed on each node and road section in the sub-area until the traffic state is converted into a free flow and a stable flow state, and then the boundary control is stopped.

7. The design of a signal timing scheme for road network boundary intersections based on feedback control according to claim 1 is characterized in that: The method for predicting the traffic flow in each sub-area at the next moment is: using the autoregressive integrated moving average model to predict the traffic flow in each sub-area at the next moment; using the mean square error and mean absolute error indicators to evaluate the prediction results of the model.

Images