Trunk traffic coordination control method and system

Abstract

The invention provides an arterial traffic coordination control method and system, and belongs to the field of road traffic control, and the method comprises the steps: introducing a graph neural network to carry out the fusion of global information of traffic, and carrying out the coding of a gating time sequence convolution network; global traffic prediction data is obtained according to the coding state and the fusion sum features, and traffic state grades are mapped according to the prediction data; obtaining traffic flow and vehicle speed of local traffic, calculating a fuzzy membership degree, calculating a weighted relation of traffic flow among nodes by using a dynamic traffic flow weighted model, and adjusting local green light timing; establishing a global-local joint objective function, inputting the traffic state grade and the local green light timing into the objective function, and outputting a global-local objective function value; and selecting a control strategy with the minimum global-local objective function value, and outputting the green light control duration. A globally coordinated signal timing scheme is established, the traffic efficiency of the whole arterial traffic is improved, and traffic flow changes and emergencies can be quickly responded.

Description

Technical Field

[0001] The present invention belongs to the field of road traffic control, and in particular relates to a traffic artery coordinated control method and system. Background Art

[0002] With the acceleration of urbanization and the continuous growth of the number of motor vehicles, the congestion problem of urban traffic arteries is becoming increasingly serious. Traditional traffic signal control methods mainly include fixed-period control and inductive control. These methods played an important role in early urban traffic management, but with the acceleration of urbanization and the surge in traffic flow, their limitations have become increasingly prominent. First, its adaptability is insufficient. Fixed-period control is based on a preset signal timing scheme and cannot respond to the dynamic changes of traffic flow in real time. In the case of large fluctuations in traffic demand, fixed timing may cause some sections to be oversaturated, while other sections waste resources and cannot meet the different needs during peak and non-peak hours. Secondly, it lacks global coordination. Although inductive control can adjust signals according to local traffic conditions, it is mainly aimed at a single intersection or local area, and lacks global optimization of the entire traffic network. This may lead to improvements in local areas at the expense of global efficiency, resulting in a "one gain while the other loses" situation.

[0003] For the above traffic conditions, experts and scholars have proposed a variety of traffic signal coordination control methods. Among them, the global coordination control method maximizes the overall traffic efficiency by macro-optimizing the entire traffic network. However, this method usually relies on accurate traffic models and a large amount of real-time data to predict traffic flow and adjust signal timing. However, the global method has high computational complexity, high implementation cost, and high requirements for data accuracy. The local coordination control method focuses on the optimization of a single intersection or local area, ignoring the impact on the entire traffic network, resulting in poor traffic efficiency on a global scale. Summary of the invention

[0004] In order to solve the limitation problem of global coordination and local coordination in traffic control, the present invention provides a traffic artery coordinated control method and system.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] A traffic artery coordinated control method comprises the following steps:

[0007] Acquire global information of traffic arteries within the target range; the global information specifically includes vehicle data and road condition data; construct a multi-source data analysis model based on a graph neural network and a gated temporal convolutional network, input the global information of the traffic into the multi-source data analysis model, and output a global traffic flow prediction value, density prediction value, and speed prediction value; map the traffic flow prediction value, density prediction value, and speed prediction value to a traffic state level through fuzzy logic;

[0008] Obtain the traffic volume and speed of the local traffic artery within the target range, calculate the fuzzy membership according to the traffic volume and speed of the local traffic, and use the dynamic traffic flow weighted model to calculate the weighted relationship of the traffic flow between nodes, and adjust the local green light timing in combination with the weighted relationship of the fuzzy membership and the traffic flow;

[0009] A global-local joint objective function is established based on the global optimization objective function and the local optimization objective function, the traffic status level and the local green light timing are input into the global-local joint objective function, and a plurality of global-local control parameters are output; a control strategy corresponding to the minimum global-local control parameter is selected, and the green light control duration is output according to the control strategy.

[0010] Preferably, the multi-source data analysis model is constructed based on the graph neural network and the gated temporal convolutional network, the global information of the traffic is input into the multi-source data analysis model, and the global traffic flow prediction value, density prediction value and speed prediction value are output, specifically through the following steps:

[0011] Dividing the global information into node features, edge features and global structural features;

[0012] The graph neural network GNN is used to fuse the node feature fusion and the edge feature to obtain the node fusion feature and the edge fusion feature. The node fusion feature is specifically calculated by the following formula:

[0013] A city trunk network G = (V, E), where V = {v1, v2, ..., v N} represents N traffic nodes; E represents the edges connecting these nodes; traffic node v i The input feature is a vector consisting of spatiotemporal data from S different data sources.

[0014]

[0015] Among them, q i (t) represents the node v at time t i Traffic flow at the location; d i (t) represents the vehicle density; v i (t) represents the vehicle speed; f i(t) represents other external data sources, including bus data and video surveillance data; Represents node v i The fusion feature vector of represents the concatenation operation of features; φ(·) represents a nonlinear activation function, which is a ReLU function in the present invention; represents the feature transformation matrix of s data sources; b s represents the bias vector; F' represents the unified dimension of each data source feature after transformation; F = S × F' represents the feature dimension after fusion;

[0016] When the global traffic network connectivity C≥0, edge feature fusion of multi-source data is performed; the global traffic network connectivity is specifically calculated by the following formula:

[0017]

[0018] Among them, the adjacency matrix A between nodes represents the topological structure of the road network. ij Represents node v i and v j Is there an edge between them? If so, then A ij =1, otherwise A ij =0;

[0019] The edge fusion feature is calculated by the following formula:

[0020]

[0021] Among them, e uv ∈R D It is represented as the fused feature vector of edge (u,v); ψ(·) is represented as a nonlinear activation function; Represents the feature transformation matrix of the s'th edge data source; It is represented as a bias vector; D' represents the unified dimension of each edge feature after transformation; D = S' × D' represents the dimension of the fused edge feature;

[0022] The gated temporal convolutional neural network GTCN is used to perform temporal encoding on the node fusion features, and the temporal encoding is performed using the following formula:

[0023]

[0024] in, represents the initial hidden state of node v, which contains timing information; σ(·) represents the activation function, which is ReLU; ζ(·) represents the gating function, which is sigmoid; * represents the convolution operation; K τ ,Q τ ∈R K×FRepresented as a time convolution kernel, K is the convolution kernel size; b,c∈R F Represented as a bias vector; Expressed as an element-wise multiplication algorithm;

[0025] The edge feature weight value is calculated according to the temporal coding state and the edge fusion feature, specifically through the following formula:

[0026]

[0027] Among them, α uv represents the attention weight of edge (u,v); represents the hidden state of nodes u and v in the lth layer; e uv ∈R D Represents the fusion features of edge (u,v); represents the transformation matrix; represents the attention mechanism parameter vector; γ(·) represents the LeakyReLU activation function; || represents the vector concatenation operation; N(v) represents the set of neighbor nodes of node v;

[0028] After calculating the edge weight α uv After that, message passing is performed and the node status is updated, specifically through the following formula:

[0029]

[0030] in, represents the hidden state of the node v in the l+1th layer; W h ,W s ∈R F×F represents the weight matrix of message passing and self-loop; σ(·) is the activation function;

[0031] The temporal and spatial information are integrated to perform spatiotemporal attention fusion to obtain the node integration representation, which is specifically expressed by the following formula:

[0032]

[0033] Among them, β v represents the spatiotemporal attention coefficient of node v; represents the spatiotemporal attention parameter vector; represents the feature transformation matrix; tanh(·) represents the hyperbolic tangent activation function; represents the fused representation of node v;

[0034] Use node output to predict traffic flow Q v , the specific formula is as follows:

[0035]

[0036] in, represents the predicted value of node v; L represents the number of layers of the network; f(·) represents the output mapping function, which refers to the linear layer in this invention; at the same time, the density prediction value K of node v can be obtained through the above steps v and the speed prediction value V v .

[0037] Preferably, the traffic flow prediction value, density prediction value and speed prediction value are mapped to traffic status levels through fuzzy logic, and the traffic status levels include smooth and congested, which are calculated by the following formula:

[0038] S=f(Q v ,K v ,V v );

[0039] Among them, S represents the traffic status output; Q v represents the traffic prediction value of the above node v; K v V represents the predicted value of the density of the node v above; v Represents the predicted speed value of the above node v.

[0040] Preferably, the fuzzy membership is calculated according to the traffic volume and speed of the local traffic, and the weighted relationship of the traffic flow between nodes is calculated using a dynamic traffic flow weighted model, and the local green light timing is adjusted in combination with the weighted relationship of the fuzzy membership and the traffic flow, which specifically includes the following steps:

[0041] The fuzzy membership is calculated according to the traffic volume and speed of the local traffic, using the following formula:

[0042]

[0043] Among them, μ Q , μ V They represent the fuzzy membership functions of vehicle flow Q and vehicle speed V respectively; α and β are the adjustment parameters of flow and speed respectively, Q 0 ,V 0 is the fuzzification threshold;

[0044] The weighted relationship of traffic flow between nodes is calculated using a dynamic traffic flow weighted model, and the dynamic traffic flow weighted model is specifically:

[0045]

[0046] Among them, ω uv represents the weighted traffic flow impact of node u on node v; Q u represents the traffic flow at node u;

[0047] V urepresents the average vehicle speed at node u; Q u V u represents the product of the traffic flow speed on road section u, that is, at the timing moment, the traffic flow and vehicle speed of node u jointly determine its traffic impact on the downstream node v. According to this value and the weighted traffic flow in the neighbor node set, the control strategy of the traffic light is dynamically adjusted; N(v) represents the neighbor node set of node v, that is, all the road sections or intersections connected to node v; ∑ k∈N(v) Q k V k represents the total traffic flow in the neighboring nodes of node v;

[0048] The local green light timing is adjusted by combining the weighted relationship between the fuzzy membership and the traffic flow, and the local green light timing G is output. g The calculation formula is:

[0049] G g =ω uv (μ Q +μ V );

[0050] Where: uv represents the weighted traffic flow impact of node u on node v; μ Q , μ V They represent the fuzzy membership functions of vehicle flow Q and vehicle speed V respectively.

[0051] Preferably, the global-local joint objective function is established based on the global optimization objective function and the local optimization objective function, specifically:

[0052] The global optimization objective function J global for:

[0053]

[0054] Where n is the number of arterial intersections; S i Indicates the traffic status level 1 is smooth, 0 is congested; G i Indicates the green light time of the i-th intersection; Queue i represents the length of the vehicle queue at the i-th intersection; Flow i represents the traffic flow at the i-th intersection; ω1, ω2 represent the global target weights, which measure the importance of green light time and queuing efficiency;

[0055] The local optimization objective function J local for:

[0056]

[0057] Among them, Delay jrepresents the average delay time of vehicles at the jth intersection; S j It also indicates the traffic status level, 1 is smooth and 0 is congested; G j It also represents the green light time; ω3 and ω4 represent the local target weights, which measure the importance of delay time and green light allocation;

[0058] The global-local joint optimization objective function is:

[0059] J total =δ·J global +(1-δ)·J local ;

[0060] Among them, J total represents the global and local joint optimization objective function; δ represents the balance coefficient between global and local optimization, which adjusts the balance between global and local optimization objectives, 0≤δ≤1; J global represents the global traffic optimization objective function, reflecting the efficiency of the entire traffic network; J local It represents the local signal control optimization objective function, reflecting the specific signal control effect of each intersection.

[0061] Preferably, the weight of the global-local traffic flow is also set, specifically:

[0062]

[0063] Among them, A uv represents the traffic flow weight of the global node u and the local node v; Q u represents the traffic flow of the global node u; V u represents the vehicle speed of the global node u; Q v represents the traffic flow of local node v; V v represents the vehicle speed of the local node v; δ represents the balance coefficient of global and local optimization.

[0064] Preferably, the control strategy corresponding to the smallest global-local control parameter is selected, and the green light control duration is output according to the control strategy. Specifically, the control strategy corresponding to the smallest of the multiple global-local control parameters is selected, and the green light duration G of each intersection in the control strategy is output. i Calculate with the weight of global-local traffic flow and output the green light control duration.

[0065] Preferably, the green light time G of each intersection in the control strategy is i The weight of the global-local traffic flow is calculated using the following formula:

[0066]

[0067] Among them, G'g G represents the green light duration of local node v; g represents the initial green light duration of the local node v; ΔG represents the adjustment increment of the green light duration, ΔG=G' g -G g ;Σ u∈N(v) A uv It represents the accumulation of global and local traffic flow weights, reflecting the impact of surrounding traffic flow on the current road section.

[0068] The present invention also provides a traffic arterial coordinated control system, which specifically includes:

[0069] A global data collection and analysis module is used to obtain global information of traffic arteries within a target range; the global information specifically includes vehicle data and road condition data; a multi-source data analysis model is constructed based on a graph neural network and a gated temporal convolutional network, the global information of the traffic is input into the multi-source data analysis model, and global traffic flow prediction values, density prediction values ​​and speed prediction values ​​are output; the traffic flow prediction values, density prediction values ​​and speed prediction values ​​are mapped to traffic status levels through fuzzy logic.

[0070] The local signal control module is used to obtain the traffic volume and speed of the local traffic arteries within the target range, calculate the fuzzy membership according to the traffic volume and speed of the local traffic, and use the dynamic traffic flow weighted model to calculate the weighted relationship of the traffic flow between nodes, and adjust the local green light timing in combination with the weighted relationship between the fuzzy membership and the traffic flow.

[0071] The global-local coordinated scheduling module is used to establish a global-local joint objective function based on the global optimization objective function and the local optimization objective function, input the traffic status level and the local green light timing into the global-local joint objective function, and output multiple global-local control parameters; select the control strategy corresponding to the minimum global-local control parameter, and output the green light control duration according to the control strategy.

[0072] The traffic artery coordinated control method provided by the present invention has the following beneficial effects:

[0073] The present invention constructs a multi-source data analysis model based on a graph neural network and a gated temporal convolutional network, inputs the acquired global information of traffic arteries into the multi-source data analysis model, obtains global traffic flow prediction values, density prediction values ​​and speed prediction values, and realizes the fusion of features in a unified space through feature transformation and splicing of multi-source data, thereby improving the model's processing of complex time and space. The vehicle flow and speed of local traffic are obtained and the membership is calculated, and the weighted relationship of traffic flow between nodes is calculated using a dynamic traffic flow weighted model, and the local green light timing is adjusted in combination with the weighted relationship between the fuzzy membership and traffic flow. According to the predicted value of global traffic flow and the local control strategy, the optimal green light duration control strategy is output based on the global-local joint objective function. The signal light timing scheme is adaptively adjusted to achieve dynamic optimization of global-local signal control, effectively reducing vehicle waiting time and traffic congestion. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the embodiment of the present invention and its design scheme, the following briefly introduces the drawings required for this embodiment. The drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0075] Figure 1 The present invention is a flow chart of a traffic artery coordinated control method.

[0076] Figure 2 This is a global traffic network diagram according to an embodiment of the present invention.

[0077] Figure 3 This is a trunk coordinated transportation network diagram according to an embodiment of the present invention.

[0078] Figure 4 This is a diagram showing the effect of local signal light control according to an embodiment of the present invention.

[0079] Figure 5 This is a diagram of the global-local coordination effect in an embodiment of the present invention.

[0080] Figure 6 This is a diagram showing changes in global and local traffic flow before and after optimization in an embodiment of the present invention.

[0081] Figure 7 This is a diagram showing the change in waiting time before and after signal optimization in an embodiment of the present invention.

[0082] Figure 8 This is a diagram showing the changes before and after the optimization of the signal light timing in an embodiment of the present invention.

[0083] Fig. 9 This is a heat map of traffic flow after global and local coordination in an embodiment of the present invention.

[0084] Fig.10 This is a traffic flow heat map in an embodiment of the present invention. DETAILED DESCRIPTION

[0085] In order to enable those skilled in the art to better understand the technical solution of the present invention and implement it, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the scope of protection of the present invention.

[0086] Example

[0087] The present invention provides a traffic artery coordinated control method, such as Figure 1 As shown, the specific steps include:

[0088] S1. Obtain global information on traffic. Global information on trunk traffic is collected through a variety of data sources, mainly including floating vehicle data, road test sensor data, video surveillance data and public transportation data. Among them: vehicle floating data, which obtains the vehicle's location information, speed, and acceleration through the vehicle-mounted satellite positioning equipment, can reflect the vehicle's driving trajectory in real time, and is an important source of dynamic traffic flow data; roadside sensor data is the number, speed, and lane occupancy of passing vehicles detected by sensors installed along the trunk roads; video surveillance data is extracted from the video stream by cameras installed at major intersections, and the number, type, and driving direction of vehicles are identified using computer vision technology; public transportation data is operational data from transportation tools such as buses, taxis, and shared bicycles, which helps to comprehensively evaluate the status of urban traffic flow.

[0089] S2. Build a multi-source data analysis model based on graph neural network and gated temporal convolutional network, input global traffic information into the multi-source data analysis model, output global traffic flow prediction value, density prediction value and speed prediction value, and map the traffic flow prediction value, density prediction value and speed prediction value into traffic status level through fuzzy logic. The global traffic network diagram is as follows: Figure 2 shown.

[0090] The global traffic optimization objective function is as follows:

[0091]

[0092] Where: global represents the global optimization objective function; T i represents the vehicle passing time of the i-th road section; W i represents the average vehicle waiting time of the i-th road section; C i represents the traffic flow rate of the i-th road section; λ1, λ2, λ3 represent the target weights, which are used to measure the priority between different targets. The traffic flow prediction heat map is as follows: Fig.10 shown.

[0093] S21. Different data sources are heterogeneous and inconsistent, and need to be processed by multi-source data fusion algorithms. Graph neural network (GNN) is used to fuse multi-source spatiotemporal data, establish a multi-source data analysis model, process the topological structure in the transportation network, integrate information from different sensors and data sources, and consider the multivariate characteristics of nodes and edges, spatiotemporal sequence information, and spatial topological structure. The relationship between nodes and traffic in the trunk coordinated transportation network is as follows: Figure 3 shown.

[0094] For a city trunk network G = (V, E), where V = {v1, v2, ..., v N} represents N traffic nodes (intersections in the trunk line or sections between adjacent intersections), E represents the edges connecting these nodes (specifically, the interconnection relationship between the connecting sections), and the traffic node v i The input feature is a vector consisting of spatiotemporal data from S different data sources. The multi-source node fusion formula is as follows:

[0095]

[0096] Among them, q i (t) represents the traffic flow at node vi at time t; d i (t) represents the vehicle density; v i (t) represents the vehicle speed; f i (t) represents other external data sources (such as bus data, video surveillance data, etc.); Represents node v i The fusion feature vector of represents the concatenation operation of features; φ(·) represents a nonlinear activation function, which is a ReLU function in the present invention; represents the feature transformation matrix of s data sources; b s represents the bias vector; F' represents the unified dimension of each data source feature after transformation; F = S × F' represents the feature dimension after fusion.

[0097] The adjacency matrix A between nodes represents the topological structure of the road network. ij Represents node v i and v j Is there an edge between them? If so, then A ij =1, otherwise A ij = 0. Here, we combine complex network theory to measure the degree of influence between different road sections, help optimize the global decision of the entire traffic flow control, and make predictions. The formula is as follows:

[0098]

[0099] Among them, the adjacency matrix A between nodes represents the topological structure of the road network. ij Represents node v i and v j Is there an edge between them? If so, then A ij =1, otherwise A ij =0.

[0100] S22. After node fusion, if the network connectivity C ≥ 0, perform edge feature fusion of multi-source data. When graph neural network fuses multi-source spatiotemporal data, it is necessary to perform multi-source edge feature fusion while performing multi-source node feature fusion. For an edge (u, v) ∈ E, there are edge features from S' data sources. Where s'=1,2,...,S', the edge feature fusion formula is as follows:

[0101]

[0102] Among them, e uv ∈R D It is represented as the fused feature vector of edge (u,v); ψ(·) is represented as a nonlinear activation function; Represents the feature transformation matrix of the s′th edge data source; It is represented as a bias vector; D' represents the unified dimension of each edge feature after transformation; D = S' × D' represents the dimension of the fused edge feature.

[0103] Considering the spatiotemporal dynamics of node features, the data of the last Γ time steps are encoded, and the time window Γ = {t-T+1,...,t} is defined. For each time step τ∈Γ, the feature of node v is The gated temporal convolutional neural network GTCN is used for temporal encoding. The formula is as follows:

[0104]

[0105] in, represents the initial hidden state of node v, including timing information; σ(·) represents the activation function, which is ReLU in the present invention; ζ(·) represents the gating function, which is sigmoid; * represents the convolution operation; K τ ,Q τ ∈R K×F Represented as a time convolution kernel, K is the convolution kernel size; b,c∈R F Represented as a bias vector; Expressed as an element-wise multiplication algorithm.

[0106] S23. Calculate the weight α of the multivariate feature calculation edge (u,v) uv Capture the correlation between nodes, specifically through the following formula:

[0107]

[0108] Among them, α uv represents the attention weight of edge (u,v); represents the hidden state of nodes u and v in the lth layer; e uv ∈R D Represents the fusion features of edge (u,v); represents the transformation matrix; represents the attention mechanism parameter vector; γ(·) represents the LeakyReLU activation function; || represents the vector concatenation operation; N(v) represents the set of neighbor nodes of node v.

[0109] Messages are passed based on weights and node status is updated using the following formula:

[0110]

[0111] in, represents the hidden state of the node v in the l+1th layer; W h ,W s ∈R F×F represents the weight matrix of message passing and self-loop; σ(·) is the activation function.

[0112] S24. Combine temporal and spatial information to perform spatiotemporal attention fusion to obtain the node integration representation, which is specifically achieved through the following formula:

[0113]

[0114] Among them, β v represents the spatiotemporal attention coefficient of node v; represents the spatiotemporal attention parameter vector; represents the feature transformation matrix; tanh(·) represents the hyperbolic tangent activation function; Represents the fused representation of node v.

[0115] Use node output to predict traffic flow Q v , the specific formula is as follows:

[0116]

[0117] in, represents the predicted value of node v; L represents the number of layers of the network; f(·) represents the output mapping function, which refers to the linear layer in the present invention.

[0118] In the multivariate spatiotemporal data fusion model, a loss function is defined to train the model, and the Adam optimizer is used to update the model parameters. The specific formula is as follows:

[0119]

[0120] Where: Q v represents the true value of node v; λ is the regularization coefficient; Θ is the set of all trainable parameters of the model; η is the learning rate; Represents the gradient of the loss function with respect to the parameters.

[0121] S25. By combining traditional data (roadside sensors) with dynamic data (vehicle trajectories), the real-time and accuracy of traffic flow status are improved, and finally the traffic flow prediction value Q of node v is obtained. v Similarly, we can get the density prediction value K of node v v and the speed prediction value V v Then the traffic status is evaluated and output. Based on the analyzed data, the fuzzy logic system is used to evaluate the traffic status. The present invention defines two fuzzy rules: if the predicted flow value is large, the traffic status is "congested", and if the predicted flow value is small, the traffic status is "unblocked". These two rules are combined with the fuzzy membership function to judge the traffic status. The formula is as follows:

[0122] S=f(Q v ,K v ,V v );

[0123] Where: S represents the traffic status output; Q v represents the traffic prediction value of the above node v; K v V represents the predicted value of the density of the node v above; v Represents the predicted speed value of the above node v.

[0124] Through fuzzy logic, the system maps continuous traffic parameters to different traffic status levels ("smooth" or "congested"), providing a basis for global and local control.

[0125] S3. Obtain the traffic volume and speed of local traffic, calculate the fuzzy membership according to the traffic volume and speed of local traffic, and use the dynamic traffic flow weighted model to calculate the weighted relationship of traffic flow between nodes, and adjust the local green light timing in combination with the weighted relationship of fuzzy membership and traffic flow.

[0126] Adjust the traffic light timing strategy according to the traffic status and control the local signal. The local signal control optimization objective function is as follows:

[0127]

[0128] Among them: Jlocal represents the local optimization objective function; W t represents the average waiting time of vehicles at time t; P t represents the average number of vehicle stops at time t; C t The vehicle passing rate at time t; λ1, λ2, λ3 represent weight parameters.

[0129] S31. Calculate the fuzzy membership degree according to the traffic volume and vehicle speed of the local traffic, specifically through the following formula:

[0130]

[0131] Among them, μ Q , μ V They represent the fuzzy membership functions of vehicle flow Q and vehicle speed V respectively; α and β are the adjustment parameters of flow and speed respectively, Q 0 ,V 0 is the threshold of fuzzification, defining the demarcation point of the "normal" state, used to distinguish low, medium and high traffic flow or vehicle speed; the adjustment parameters α and β control the steepness of the fuzzy membership function, that is, the sensitivity of the input variables (traffic flow and vehicle speed) to the membership. In the present invention, due to global-local coordinated control, the values ​​of α = 0.5 and β = 0.03 are determined by trial and error.

[0132] S32. Calculate the weighted relationship of traffic flows between nodes using a dynamic traffic flow weighted model. The dynamic traffic flow weighted model is specifically:

[0133]

[0134] Among them, ω uv represents the weighted traffic flow impact of node u on node v; Q u represents the traffic flow at node u; V u represents the average vehicle speed at node u; Q u V u represents the product of the traffic flow speed on road section u, that is, at the timing moment, the traffic flow and vehicle speed of node u jointly determine its traffic impact on the downstream node v. According to this value and the weighted traffic flow in the neighbor node set, the control strategy of the traffic light is dynamically adjusted; N(v) represents the neighbor node set of node v, that is, all road sections or intersections connected to node v; Σ k∈N(v) Q k V k Represents the total traffic flow in the neighboring nodes of node v.

[0135] S33, combining the weighted relationship between fuzzy membership and traffic flow to adjust the local green light timing, and output the green light duration G g The calculation formula is:

[0136] G g =ω uv (μ Q +μ V );

[0137] Where: uv represents the weighted traffic flow impact of node u on node v; μ Q , μ V They represent the fuzzy membership functions of the traffic flow Q and the vehicle speed V respectively. The control effect of local traffic lights is as follows: Figure 4 shown.

[0138] S4. Coordinate global and local signal timing to ensure smooth overall traffic flow on urban trunk lines. A global-local joint objective function is established based on the global optimization objective function and the local optimization objective function, specifically:

[0139] J total =δ·J global +(1-δ)·J local ;

[0140] Among them, J total represents the global and local joint optimization objective function; δ represents the balance coefficient between global and local optimization, 0≤δ≤1; J global represents the global traffic optimization objective function, reflecting the efficiency of the entire traffic network; J local It represents the local signal control optimization objective function, reflecting the specific signal control effect of each intersection.

[0141] S41, J total It is divided into several levels to evaluate the effect of the optimization results. The following evaluation criteria are defined:

[0142] 0≤J total ≤500: indicates that there is almost no delay in traffic flow, the signal timing plan at the intersection is reasonable, the traffic flow is close to the theoretical maximum, and there is almost no congestion.

[0143] J total >500: Indicates that traffic flow is severely congested. The signal timing at the intersection fails to effectively respond to changes in traffic flow. The system does not run smoothly, causing congestion.

[0144] The data is represented in the form of a matrix, which is used as the input matrix of the joint optimization model. The input matrix is ​​generalized as follows:

[0145]

[0146] Assume that there are four intersections (numbered 1 to 4) on a main road in a city. The input data for each intersection includes: Traffic status level: obtained by mapping from traffic parameters (traffic flow, vehicle speed) through fuzzy logic, "smooth": 1; "congested": 0. Green light time: the green light duration (in seconds) of each intersection provided according to the local traffic timing.

[0147] Table 1 Global-local example dataset table

[0148]

[0149] Dataset structure description: Traffic status level: For example, the status of intersections 1 and 3 is "unblocked" (1), while the status of intersections 2 and 4 is "congested" (0). Green light time: The specific duration of the signal light, such as the green light time of intersection 2 is 45 seconds, and the green light time of intersection 4 is 50 seconds.

[0150] Represent the data in matrix form:

[0151]

[0152] Each row corresponds to the input data of an intersection, and the columns represent different input features: the first column is the traffic state level S i The second column is the green light time G j .

[0153] Expand the joint objective function, the formula is as follows:

[0154]

[0155] The joint objective function needs to satisfy the following constraints:

[0156] Green light time range constraints: Traffic state adjustment consistency constraints: ε is a very small value; queuing flow constraint:

[0157] The different data matrices X in the data source of the same trunk traffic that needs to be coordinated are input into the joint objective function J total In the same coordinated trunk line, different control schemes have different traffic conditions and green light times. The data matrix corresponding to different control schemes is input into the J calculated by the joint objective function. total Corresponding to different values, according to the rules proposed in this invention, which solution calculates J total The solution with the smallest value is the best. Take the second column of the data matrix X of the best solution, which is the green light time G solved by the objective function of coordinating each intersection of the trunk line. i .

[0158] S42. Construct global and local traffic flow coordination by calculating the traffic flow weights based on nodes and edges. Define a global-local traffic flow weight matrix A uv , in order to reflect the traffic flow relationship between the global node u and the local node v, the specific formula is as follows:

[0159]

[0160] Among them, A uv represents the traffic flow weight of the global node u and the local node v; Q u represents the traffic flow of the global node u; V u represents the vehicle speed of the global node u; Q v represents the traffic flow of local node v; V v represents the vehicle speed of the local node v; γ represents the coordination coefficient of the global and local traffic flow; N(v) represents the local node v and the set of neighboring nodes.

[0161] S43. By calculating the global and local weights, the system can dynamically adjust the signal light distribution at each intersection. The final green light duration at each intersection after adjustment is as follows: i In the case of i = 1, 2, ..., n, we respectively substitute them and perform a single final optimization according to the edges and nodes. The final optimization of the green light time of a single intersection on the trunk line is to return to the local coordinated control. Different X are input into the global-local joint objective function to minimize the calculated value of this objective function and obtain the best solution. According to the green light time G of each single intersection in the matrix X i Perform final optimization on a single intersection. The global and local coordination effects are as follows: Figure 5 As shown in the figure, the green light duration is adjusted by the following formula:

[0162]

[0163] Among them, G' g G represents the green light duration of local node v; g represents the initial green light duration of the local node v; ΔG represents the adjustment increment of the green light duration; Σ u∈N(v) A uv It represents the accumulation of global and local traffic flow weights, reflecting the impact of surrounding traffic flow on the current road section.

[0164] like Figure 6 , Figure 7 , Figure 8 and Fig. 9 FIG. 1 is a comparison diagram before and after the optimization of coordinated control of traffic arteries using the present invention.

[0165] In order to solve the shortcomings of existing traffic signal control methods in global coordination, real-time adaptability and multi-objective optimization, and realize efficient and intelligent management of traffic arteries. Construct a multi-level control method that integrates global optimization and local dynamic adjustment. By optimizing the global traffic network at the macro level, a globally coordinated signal timing plan is established to improve the traffic efficiency of the entire traffic artery. At the same time, at the micro level, real-time traffic data is used to dynamically adjust local intersections and sections, quickly respond to changes in traffic flow and emergencies, and ensure the good operation of local traffic conditions. The following beneficial effects are included:

[0166] (1) Based on the joint analysis of multi-source spatiotemporal data using GTCN and graph neural network, the combination of gated temporal convolutional network (GTCN) and graph neural network (GNN) is innovatively introduced in the global data analysis and acquisition module. Through GTCN, the system can efficiently extract the temporal dependency characteristics of traffic data in time series and capture the dynamic changes of global traffic flow. At the same time, the GNN model is used to process complex network topological relationships in the spatial dimension and analyze the traffic flow associations between different road sections and intersections. The combined use of the two enables the system to not only accurately predict the spatiotemporal evolution of traffic flow, but also process complex global traffic network topological structures, greatly improving the prediction accuracy and real-time performance of traffic data.

[0167] (2) Dynamic optimization control of local signals based on reinforcement learning and TCN. Reinforcement learning is combined with a temporal convolutional network (TCN) to form a new local traffic signal optimization strategy. Reinforcement learning helps the system to adjust according to traffic flow dynamics in a real-time environment by adaptively adjusting the timing of traffic lights. TCN, through its powerful time series modeling capabilities, predicts and analyzes historical data of local traffic flow, further improving the accuracy and response speed of traffic light control. This combination realizes dynamic optimization of local signal control and can effectively reduce vehicle waiting time and traffic congestion.

[0168] (3) Global and local joint optimization multi-objective time-series traffic scheduling model: This paper innovatively proposes a global-local joint optimization model, which balances the macro control of global traffic flow and the fine regulation of local signals by jointly optimizing global and local traffic flows. The model integrates the global GTCN and the local TCN, constructs a spatiotemporal joint multi-objective optimization function, and comprehensively optimizes and solves indicators such as global traffic flow, local signal timing, and waiting time. It can adaptively adjust the signal timing plan to achieve global and local coordinated scheduling of traffic flow.

[0169] The present invention improves the overall efficiency and stability of traffic flow. By combining global and local optimization strategies, the present invention can improve the overall efficiency of traffic flow and reduce bottlenecks and congestion in the traffic network. The optimized traffic signal can be dynamically adjusted according to the real-time traffic flow, improve the stability of the traffic network, and reduce traffic delays during peak hours. The delays at local intersections are reduced and the traffic capacity is improved. The present invention implements personalized adjustments at each intersection through local signal optimization, which can reduce traffic delays and queues, especially at high-traffic intersections. Local optimization ensures that the traffic capacity of the intersection is maximized by adaptively adjusting the signal timing, thereby improving the local traffic smoothness. Optimize global-local coordination to avoid the phenomenon of "one rises and the other falls". Through the coordination of global and local signal optimization, the present invention can effectively avoid the phenomenon of "one rises and the other falls", ensuring global traffic smoothness while fully guaranteeing the traffic capacity of local intersections. The optimization of traffic flow is not limited to a certain local area, but realizes the efficient operation of the entire traffic network.

[0170] The present invention also provides a traffic arterial coordinated control system, which specifically includes:

[0171] The global data collection and analysis module is used to obtain the global information of traffic arteries within the target range; the global information specifically includes vehicle data and road condition data; a multi-source data analysis model is constructed based on graph neural networks and gated temporal convolutional networks, and the global information of traffic is input into the multi-source data analysis model to output the global traffic flow prediction value, density prediction value and speed prediction value; the traffic flow prediction value, density prediction value and speed prediction value are mapped to traffic status levels through fuzzy logic.

[0172] The local signal control module is used to obtain the traffic volume and speed of the local traffic arteries within the target range, calculate the fuzzy membership based on the traffic volume and speed of the local traffic, and use the dynamic traffic flow weighted model to calculate the weighted relationship of the traffic flow between nodes, and adjust the local green light timing in combination with the weighted relationship between the fuzzy membership and the traffic flow.

[0173] The global-local coordinated scheduling module is used to establish a global-local joint objective function based on the global optimization objective function and the local optimization objective function, input the traffic status level and the local green light timing into the global-local joint objective function, and output multiple global-local control parameters; select the control strategy corresponding to the minimum global-local control parameter, and output the green light control duration according to the control strategy.

[0174] In the system, the global data collection and analysis module integrates multi-source traffic data to build a global traffic status perception system, which improves the dynamic monitoring and prediction capabilities of the entire traffic network. Through data fusion and complex network modeling, it can perceive the changing trend of global traffic flow in real time, reduce traffic congestion, and improve traffic efficiency. At the same time, by optimizing global traffic stability, it can reduce the uncertainty caused by traffic fluctuations and improve the efficiency and stability of global traffic flow.

[0175] In the local signal control module, the timing strategy of the traffic lights is dynamically adjusted at each intersection or road section according to the real-time traffic status. The adaptive algorithm combining reinforcement learning and fuzzy control can automatically optimize the green light duration of the traffic lights according to key indicators such as traffic flow and vehicle waiting time, reduce the waiting time and number of stops of vehicles, and improve the traffic efficiency of local traffic.

[0176] The global-local coordination and dispatching module organically combines global and local traffic control objectives. It can find a balance between global traffic mobility and local signal control, which can not only ensure the maximization of the overall efficiency of the global traffic network, but also meet the personalized needs of local signal control, and achieve efficient traffic flow regulation. It can also achieve real-time global and local coordination and emergency dispatch to respond to emergencies and traffic anomalies. Through real-time data feedback and dynamic optimization and adjustment, the system can quickly detect abnormal situations, conduct emergency signal dispatch at the global and local levels, ensure the rapid recovery and distribution of traffic flow, effectively respond to emergencies such as traffic accidents and road construction, and achieve adaptive emergency management.

[0177] Each module in the above-mentioned traffic artery coordinated control system can be implemented in whole or in part by software, hardware or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each module.

[0178] It will be appreciated by those skilled in the art that embodiments of the present invention may provide methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0179] The present invention is described with reference to flowcharts and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as a combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0180] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0181] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0182] It should be pointed out that the specific implementation methods described above can enable those skilled in the art to understand the invention more comprehensively, but do not limit the invention in any way. Therefore, although the invention has been described in detail in this specification and embodiments, those skilled in the art should understand that the invention can still be modified or replaced by equivalents; and all technical solutions and improvements that do not deviate from the spirit and scope of the invention are included in the protection scope of the patent for the invention. Any figure mark in the claims should not be regarded as limiting the claims involved. Any simple change or equivalent replacement of the technical solution that can be obviously obtained by any technician familiar with the field within the technical scope disclosed in the present invention belongs to the protection scope of the present invention.

Claims

1. A traffic artery coordinated control method, characterized in that: The following steps are involved: Acquire global information of traffic arteries within the target range; the global information specifically includes vehicle data and road condition data; construct a multi-source data analysis model based on a graph neural network and a gated temporal convolutional network, input the global information of the traffic into the multi-source data analysis model, and output a global traffic flow prediction value, density prediction value, and speed prediction value; map the traffic flow prediction value, density prediction value, and speed prediction value to a traffic state level through fuzzy logic; Obtain the traffic volume and speed of the local traffic artery within the target range, calculate the fuzzy membership according to the traffic volume and speed of the local traffic, and use the dynamic traffic flow weighted model to calculate the weighted relationship of the traffic flow between nodes, and adjust the local green light timing in combination with the weighted relationship of the fuzzy membership and the traffic flow; A global-local joint objective function is established based on the global optimization objective function and the local optimization objective function, the traffic status level and the local green light timing are input into the global-local joint objective function, and a plurality of global-local control parameters are output; a control strategy corresponding to the minimum global-local control parameter is selected, and the green light control duration is output according to the control strategy.

2. A traffic artery coordinated control method according to claim 1, characterized in that: The multi-source data analysis model is constructed based on the graph neural network and the gated temporal convolutional network, the global information of the traffic is input into the multi-source data analysis model, and the global traffic flow prediction value, density prediction value and speed prediction value are output, specifically through the following steps: Dividing the global information into node features, edge features and global structural features; The graph neural network GNN is used to fuse the node feature fusion and the edge feature to obtain the node fusion feature and the edge fusion feature. The node fusion feature is specifically calculated by the following formula: A city trunk network G = (V, E), where V = {v1, v2, ..., v N } represents N traffic nodes; E represents the edges connecting these nodes; traffic node v i The input feature is a vector consisting of spatiotemporal data from S different data sources. Among them, q i (t) represents the node v at time t i Traffic flow at the location; d i (t) represents the vehicle density; v i (t) represents the vehicle speed; f i (t) represents other external data sources, including bus data and video surveillance data; Represents node v i The fusion feature vector of represents the concatenation operation of features; φ(·) represents a nonlinear activation function, which is a ReLU function in the present invention; represents the feature transformation matrix of s data sources; b s represents the bias vector; F' represents the unified dimension of each data source feature after transformation; F = S × F' represents the feature dimension after fusion; When the global traffic network connectivity C≥0, edge feature fusion of multi-source data is performed; the global traffic network connectivity is specifically calculated by the following formula: Among them, the adjacency matrix A between nodes represents the topological structure of the road network. ij Represents node v i and v j Is there an edge between them? If so, then A ij =1, otherwise A ij =0; The edge fusion feature is calculated by the following formula: Among them, e uv ∈R D It is represented as the fused feature vector of edge (u,v); ψ(·) is represented as a nonlinear activation function; represents the feature transformation matrix of the s'th edge data source; d s' ∈R D′ It is represented as a bias vector; D' represents the unified dimension of each edge feature after transformation; D = S' × D' represents the dimension of the fused edge feature; The gated temporal convolutional neural network GTCN is used to perform temporal encoding on the node fusion features, and the temporal encoding is performed using the following formula: in, represents the initial hidden state of node v, which contains timing information; σ(·) represents the activation function, which is ReLU; ζ(·) represents the gating function, which is sigmoid; * represents the convolution operation; K τ ,Q τ ∈R K×F Represented as a time convolution kernel, K is the convolution kernel size; b,c∈R F Represented as a bias vector; Expressed as an element-wise multiplication algorithm; The edge feature weight value is calculated according to the temporal coding state and the edge fusion feature, specifically through the following formula: Among them, α uv represents the attention weight of edge (u,v); represents the hidden state of nodes u and v in the lth layer; e uv ∈R D Represents the fusion feature of edge (u,v); W1,W2∈R F'×F ,W3∈R F'×D Represents the transformation matrix; a∈R 3F' represents the attention mechanism parameter vector; γ(·) represents the LeakyReLU activation function; || represents the vector concatenation operation; N(v) represents the set of neighbor nodes of node v; After calculating the edge weight α uv After that, message passing is performed and the node status is updated, specifically through the following formula: in, represents the hidden state of the node v in the l+1th layer; W h ,W s ∈R F×F represents the weight matrix of message passing and self-loop; σ(·) is the activation function; The temporal and spatial information are integrated to perform spatiotemporal attention fusion to obtain the node integration representation, which is specifically expressed by the following formula: Among them, β v represents the spatiotemporal attention coefficient of node v; u∈R F ” represents the spatiotemporal attention parameter vector; represents the feature transformation matrix; tanh(·) represents the hyperbolic tangent activation function; represents the fused representation of node v; Use node output to predict traffic flow Q v , the specific formula is as follows: in, represents the predicted value of node v; L represents the number of layers of the network; f(·) represents the output mapping function, which refers to the linear layer in this invention; at the same time, the density prediction value K of node v can be obtained through the above steps v and the speed prediction value V v .

3. A traffic artery coordinated control method according to claim 2, characterized in that: The traffic flow prediction value, density prediction value and speed prediction value are mapped to traffic status levels through fuzzy logic. The traffic status levels include smooth traffic and congested traffic, which are calculated by the following formula: S=f(Q v ,K v ,V v ); Among them, S represents the traffic status output; Q v represents the traffic prediction value of the above node v; K v V represents the predicted value of the density of the node v above; v Represents the predicted value of the velocity of the above node v.

4. A traffic artery coordinated control method according to claim 3, characterized in that: The fuzzy membership is calculated according to the traffic volume and speed of the local traffic, and the weighted relationship of the traffic flow between nodes is calculated using a dynamic traffic flow weighted model, and the local green light timing is adjusted in combination with the weighted relationship of the fuzzy membership and the traffic flow, specifically including the following steps: The fuzzy membership is calculated according to the traffic volume and speed of the local traffic, using the following formula: Among them, μ Q , μ V They represent the fuzzy membership functions of vehicle flow Q and vehicle speed V respectively; α and β are the adjustment parameters of flow and speed respectively, Q 0 ,V 0 is the fuzzification threshold; The weighted relationship of traffic flow between nodes is calculated using a dynamic traffic flow weighted model, and the dynamic traffic flow weighted model is specifically: Among them, ω uv represents the weighted traffic flow impact of node u on node v; Q u represents the traffic flow at node u; V u represents the average vehicle speed at node u; Q u V u represents the product of the traffic flow speed on road section u, that is, at the timing moment, the traffic flow and vehicle speed of node u jointly determine its traffic impact on the downstream node v. According to this value and the weighted traffic flow in the neighbor node set, the control strategy of the traffic light is dynamically adjusted; N(v) represents the neighbor node set of node v, that is, all the road sections or intersections connected to node v; ∑ k∈N(v) Q k V k represents the total traffic flow in the neighboring nodes of node v; The local green light timing is adjusted by combining the weighted relationship between the fuzzy membership and the traffic flow, and the local green light timing G is output. g The calculation formula is: G g =ω uv (m Q +m V ); Where: uv represents the weighted traffic flow impact of node u on node v; μ Q , μ V They represent the fuzzy membership functions of vehicle flow Q and vehicle speed V respectively.

5. A traffic artery coordinated control method according to claim 4, characterized in that: The global-local joint objective function is established based on the global optimization objective function and the local optimization objective function, specifically: The global optimization objective function J global for: Where n represents the number of trunk intersections; S i Indicates the traffic status level 1 is smooth, 0 is congested; G i Indicates the green light time of the i-th intersection; Queue i represents the length of the vehicle queue at the i-th intersection; Flow i represents the traffic flow at the i-th intersection; ω1, ω2 represent the global target weights, which measure the importance of green light time and queuing efficiency; The local optimization objective function J local for: Among them, Delay j represents the average delay time of vehicles at the jth intersection; S j It also indicates the traffic status level, 1 is smooth and 0 is congested; G j It also represents the green light time; ω3 and ω4 represent the local target weights, which measure the importance of delay time and green light allocation; The global-local joint optimization objective function is: J total =δ·J global +(1-δ)·J local ; Among them, J total represents the global and local joint optimization objective function; δ represents the balance coefficient between global and local optimization, which adjusts the balance between global and local optimization objectives, 0≤δ≤1; J global represents the global traffic optimization objective function, reflecting the efficiency of the entire traffic network; J local It represents the local signal control optimization objective function, reflecting the specific signal control effect of each intersection.

6. A traffic artery coordinated control method according to claim 5, characterized in that: It also includes setting the weights of global-local traffic flows, specifically: Among them, A uv represents the traffic flow weight of the global node u and the local node v; Q u represents the traffic flow of the global node u; V u represents the vehicle speed of the global node u; Q v represents the traffic flow of local node v; V v represents the vehicle speed of the local node v; δ represents the balance coefficient of global and local optimization.

7. A traffic artery coordinated control method according to claim 6, characterized in that: The control strategy corresponding to the smallest global-local control parameter is selected, and the green light control duration is output according to the control strategy. Specifically, the control strategy corresponding to the smallest global-local control parameter is selected, and the green light duration G of each intersection in the control strategy is output. i Calculate with the weight of global-local traffic flow and output the green light control duration.

8. A traffic artery coordinated control method according to claim 6, characterized in that: The green light time G of each intersection in the control strategy i The weight of the global-local traffic flow is calculated using the following formula: Among them, G' g G represents the green light duration of local node v; g represents the initial green light duration of the local node v; ΔG represents the adjustment increment of the green light duration, ΔG=G' g -G g ;∑ u∈N(v) A uv It represents the accumulation of global and local traffic flow weights, reflecting the impact of surrounding traffic flow on the current road section.

9. A traffic artery coordinated control system, characterized in that: include: A global data acquisition and analysis module is used to obtain global information of traffic arteries within the target range; the global information specifically includes vehicle data and road condition data; a multi-source data analysis model is constructed based on a graph neural network and a gated temporal convolutional network, the global information of the traffic is input into the multi-source data analysis model, and a global traffic flow prediction value, density prediction value and speed prediction value are output; the traffic flow prediction value, density prediction value and speed prediction value are mapped to a traffic state level through fuzzy logic; A local signal control module is used to obtain the traffic volume and speed of the local traffic artery within the target range, calculate the fuzzy membership according to the traffic volume and speed of the local traffic, and use the dynamic traffic flow weighted model to calculate the weighted relationship of the traffic flow between nodes, and adjust the local green light timing in combination with the weighted relationship of the fuzzy membership and the traffic flow; The global-local coordinated scheduling module is used to establish a global-local joint objective function based on the global optimization objective function and the local optimization objective function, input the traffic status level and the local green light timing into the global-local joint objective function, and output multiple global-local control parameters; select the control strategy corresponding to the minimum global-local control parameter, and output the green light control duration according to the control strategy.

Images