Electronic information traffic flow automatic regulation and control system based on wireless sensor network

Abstract

The invention relates to the technical field of traffic flow regulation and control, and discloses an electronic information traffic flow automatic regulation and control system based on a wireless sensor network. The multi-source sensing module collects road network multi-dimensional traffic data through a wireless sensing network, and traffic situation characteristics are generated through a specific data fusion method. The spatio-temporal feature fusion module utilizes a spatio-temporal attention mechanism to associate cross-modal features, and the collaborative decision-making module inputs fusion features into a pre-trained distributed decision-making model to generate a regulation and control instruction set. The dynamic game optimization module constructs a multi-target game optimization model, and traffic signal parameters are optimized by adopting a dynamic game strategy decomposition algorithm. And the hierarchical execution module executes regulation and control instructions in a distributed manner through a three-level control architecture of a central decision-making layer, a regional coordination layer and an intersection execution layer. The system can comprehensively collect and fuse traffic data, realizes scientific decision making and accurate regulation and control, effectively improves the road passing efficiency, balances the road network load, and relieves traffic jam.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic flow regulation, and particularly to an automatic traffic flow regulation system for electronic information based on a wireless sensor network. Background Art

[0002] With the acceleration of the urbanization process and the sharp increase in the number of motor vehicles, the problem of urban traffic congestion has become increasingly serious. Traditional traffic flow regulation methods, such as fixed-time signal control and manual intervention, have been difficult to meet the complex needs of modern traffic and have exposed many drawbacks.

[0003] In terms of traffic data collection, the previous methods have obvious deficiencies. In the early stage, only a single type of sensor, such as a geomagnetic sensor, was relied on, which could only obtain simple traffic flow data and could not comprehensively reflect the traffic conditions. For example, information such as the specific location and driving speed of vehicles could not be obtained, and it was even more difficult to detect sudden traffic events such as traffic accidents and road construction in real time. This has led to a lack of sufficient and accurate data support for traffic management departments when formulating regulation strategies, and the regulation measures are often blind. Although video monitoring devices have been introduced later, various types of sensors are independent of each other, and the data has not been effectively integrated, forming individual "data islands". The data from different sources have differences in time and space, making it difficult to comprehensively utilize them and unable to provide a complete and accurate traffic situation perception for traffic regulation.

[0004] In terms of traffic flow regulation strategies, the traditional fixed-time signal control method switches the signal lights according to a preset fixed time period without considering the real-time changes in the actual traffic flow. During the traffic peak period, the traffic flow at some intersections increases sharply, but the signal light duration is not adjusted in time, resulting in vehicles queuing for a long time and extremely low road traffic efficiency; while during the traffic low period, there are few vehicles at the intersection, but the signal lights still switch according to the fixed duration, causing a waste of road resources. Although manual intervention regulation can make adjustments to a certain extent according to the on-site situation, it relies on manpower, has a slow response speed, and is easily affected by human factors, making it impossible to achieve large-scale and efficient traffic control.

[0005] In addition, the traditional traffic regulation system lacks the overall grasp and collaborative processing ability of the global traffic situation. The regulation of each intersection and section is independent, without fully considering the coherence and mutual influence of the traffic flow within the region. When congestion occurs at a certain intersection, it is impossible to coordinate with the surrounding intersections in time, resulting in the continuous expansion of the congestion range and further affecting the traffic fluency of the entire region. For example, in the commercial areas of some cities, multiple intersections are connected to each other, and the traffic flow is complex and changeable. The traditional regulation method cannot effectively coordinate the signal lights of these intersections, making the traffic congestion situation even more serious.

[0006] At the same time, traditional systems also do not comprehensively consider traffic environmental factors. Meteorological conditions, such as heavy rain, fog, etc., will have a significant impact on traffic flow and vehicle driving speed, but traditional control systems rarely incorporate these factors into control decisions. In bad weather, road visibility decreases, vehicle driving speed slows down, and traffic flow distribution changes, while traditional control methods fail to make timely adaptive adjustments, further exacerbating traffic congestion and safety hazards. Summary of the Invention

[0007] The purpose of the present invention is to provide an automatic traffic flow control system for electronic information based on a wireless sensor network to solve the problems raised in the above background technology.

[0008] To achieve the above purpose, the present invention provides the following technical solutions: An automatic traffic flow control system for electronic information based on a wireless sensor network, the system includes: Multi-source perception module: used to collect multi-dimensional traffic data of the road network through a wireless sensor network, and the wireless sensor network includes in-vehicle positioning terminals, geomagnetic traffic sensors, video monitoring units, and meteorological monitoring stations; Spatio-temporal feature fusion module: based on a spatio-temporal attention mechanism, perform cross-modal feature association on multi-dimensional traffic data to generate traffic situation features integrating spatio-temporal characteristics; Collaborative decision-making module: input the traffic situation features into a pre-trained distributed decision-making model to generate a regional traffic flow control instruction set; Dynamic game optimization module: construct a multi-objective game optimization model according to the control instruction set, and the multi-objective game optimization model takes the maximization of traffic efficiency and the balance of road network load as optimization objectives, and uses a dynamic game strategy decomposition algorithm to collaboratively optimize traffic signal parameters; Hierarchical execution module: based on the multi-objective game optimization model, output the optimal control strategy, and realize the distributed execution of control instructions through a three-level control architecture. The three-level control architecture includes a central decision-making layer, a regional coordination layer, and an intersection execution layer. Among them, the central decision-making layer generates a global phase timing plan based on the road network topology, the regional coordination layer uses a sliding window prediction algorithm to dynamically adapt sub-region traffic parameters, and the intersection execution layer realizes the tracking of signal cycle and green signal ratio based on a fuzzy model predictive control algorithm.

[0009] Preferably, the data fusion method of the multi-source perception module includes: Perform spatio-temporal calibration on in-vehicle positioning terminal data and geomagnetic traffic sensor data to construct a real-time traffic flow distribution heat map of the road network; perform multi-scale wavelet decomposition on video monitoring unit data and meteorological monitoring station data to extract traffic event feature vectors; Construct a dual-branch feature fusion network. The first branch uses 3D convolutional kernels to extract the spatial correlation features of the traffic flow distribution heat map, and the second branch uses a gated time series network to extract the dynamic evolution law of the traffic event feature vectors; Fuse the spatial correlation features and dynamic evolution features through a cross-channel attention mechanism to generate a spatio-temporal joint matrix; perform multi-granularity feature modeling on the spatio-temporal joint matrix based on residual dilated convolution, and output traffic situation features including traffic flow density, event influence range, and meteorological interference intensity.

[0010] Preferably, the distributed decision-making model adopts a spatio-temporal graph convolutional network architecture and generates multi-level control instructions based on a dynamic node aggregation mechanism; the spatio-temporal graph convolutional network architecture includes: Construct a road network-vehicle interaction graph. The nodes in the graph include intersection nodes, road section nodes, vehicle nodes, and meteorological nodes, and the node attributes include queue length, average vehicle speed, and visibility parameters; Adopt a dual-modal graph attention mechanism. In the first stage, calculate the association weights between intersection nodes and adjacent road sections through a spatial graph convolutional layer, and in the second stage, perform importance screening on the historical traffic states through a time series graph convolutional layer; Iteratively optimize the node features based on a multi-head spatio-temporal attention module. Each attention head fuses static road network attributes and dynamic traffic parameters; improve the network training stability through skip connections and layer normalization mechanisms, and finally output a set of control instructions including priority passage for emergency vehicles and regional congestion evacuation.

[0011] Preferably, the dynamic game strategy decomposition algorithm integrating Nash equilibrium solution and constraint relaxation mechanism includes: Model the traffic signal optimization problem as a multi-objective mixed integer dynamic game problem. The decision variables include discrete signal phase switching variables and continuous green light duration adjustment variables; Initialize the relaxed game sub-problem and calculate the approximate set of equilibrium solutions. Adopt a dynamic revenue adjustment mechanism to update the utility functions of the game participants according to the real-time traffic demand; In the decomposition stage, divide the global game problem into multiple sub-game problems based on Pareto front analysis technology; in the coordination stage, introduce a virtual game participant to balance the interest conflicts between sub-problems; Use a parallel game solver to alternately optimize the sub-problems, update the local equilibrium solutions in each iteration, and synchronize the global game strategy through coordination variables.

[0012] Preferably, the sliding window prediction algorithm dynamically adapts to the traffic parameters of sub-regions including: Construct a time-varying traffic flow prediction model, discretize the vehicle kinematic equation into a state transition equation, and the state transition equation includes following distance model error, lane change decision delay, and signal control influence terms; Design a dynamic window optimization objective function, including a queuing length suppression term, a travel delay balancing term, and an emergency vehicle priority passage guarantee term.

[0013] Preferably, the intersection execution layer realizes the tracking of the signal cycle and the green signal ratio based on the fuzzy model predictive control algorithm, including: Design a three-layer fuzzy prediction inference system to map the phase deviation and traffic flow fluctuation into fuzzy rule confidence levels; construct an adaptive universe adjustment mechanism to dynamically adjust the coverage interval of the fuzzy set according to the traffic state change rate.

[0014] Preferably, the three-dimensional convolution kernel adopts a multi-scale dilated convolution structure to improve the feature extraction efficiency, including: Divide the real-time traffic flow distribution heat map of the road network into multi-level grid cells, and each cell stores the variance of the vehicle flow speed and the density gradient statistic; In the feature extraction stage, use a dilated convolution kernel to expand the spatial perception range, and in the feature reconstruction stage, restore the local detailed features through a transposed convolution layer; Introduce a spatial attention gating mechanism to dynamically weight and fuse the feature maps.

[0015] Preferably, the spatial graph convolution layer adopts a dynamic adjacency encoding mechanism, including: Define a dynamic association vector between intersection nodes and adjacent road segments, including vehicle turning probability, lane connection relationship, and historical passage efficiency; Convert the dynamic association vector into the bias weight parameter of the graph convolution kernel through a non-linear mapping layer; Fuse the bias weight parameter into the graph convolution operation process to generate a node feature representation with spatial adaptability.

[0016] Preferably, the dynamic revenue adjustment mechanism is realized based on an online reinforcement learning strategy, including: Collect the strategy selection records and equilibrium solution distributions in the historical game process as the training data set; Construct a deep Q-learning network to fit the change law of the revenue function of the game participants; Online update the network parameters through the asynchronous policy gradient descent method, and adjust the weight of the utility function of each participant in real time; When a sudden traffic event is detected, trigger an emergency reconstruction operation of the revenue function.

[0017] Preferably, the error modeling of the state transition equation adopts the interval envelope analysis technology, including: modeling the vehicle motion uncertainty as interval parameters, and the upper and lower bounds are determined by the standard deviation of the sensor measurement error; performing orthogonal decomposition on the interval parameters to separate the deterministic component of the system and the environmental random interference component.

[0018] Compared with the prior art, the beneficial effects of the present invention are: At the data collection and fusion level, the multi-source perception module uses the wireless sensor network to comprehensively collect multi-dimensional traffic data with devices such as vehicle-mounted positioning terminals, geomagnetic traffic sensors, video monitoring units, and meteorological monitoring stations. Through a unique data fusion method, it conducts spatio-temporal calibration and feature extraction on different types of data to construct a more comprehensive and accurate traffic situation feature. For example, after spatio-temporal calibration of the vehicle-mounted positioning terminal data and the geomagnetic traffic sensor data, a heat map of the real-time traffic flow distribution of the road network is constructed, which can clearly present the real-time vehicle flow dynamics; through multi-scale wavelet decomposition of the video monitoring unit data and the meteorological monitoring station data to extract traffic event feature vectors, potential traffic anomaly information can be effectively mined. Compared with the traditional single data collection method, this greatly improves the richness and accuracy of the data, providing a solid data foundation for subsequent regulation decisions.

[0019] The spatio-temporal feature fusion module conducts cross-modal feature association on multi-dimensional traffic data based on the spatio-temporal attention mechanism. This innovative fusion method can accurately capture the internal connections of traffic data in the time and space dimensions, generating traffic situation features that integrate spatio-temporal characteristics. Compared with the traditional data processing method, it is no longer limited to the analysis of single-dimensional data, but starts from the comprehensive perspective of time and space to more comprehensively and deeply understand the change rules of traffic situations. For example, when analyzing the traffic congestion situation during morning and evening rush hours, it can not only consider the spatial distribution of the traffic flow at the current intersection, but also combine the time change trend in historical data, providing a more forward-looking and targeted basis for traffic regulation.

[0020] The collaborative decision-making module inputs the fused traffic situation features into a pre-trained distributed decision-making model to generate a regional traffic flow regulation instruction set. This distributed decision-making model adopts the spatio-temporal graph convolutional network architecture. By constructing a road network-vehicle interaction graph and a dual-modal graph attention mechanism, etc., it can fully consider the mutual relationships of various factors in the road network, such as the associations between intersection nodes, road section nodes, vehicle nodes, and meteorological nodes. This makes the generated regulation instruction set more scientific and reasonable, capable of comprehensively balancing the traffic demands in different regions and different scenarios, realizing the effective regulation of regional traffic flow, and improving the overall traffic operation efficiency.

[0021] The dynamic game optimization module constructs a multi-objective game optimization model according to the regulation instruction set. Aiming at maximizing the traffic efficiency and balancing the road network load, it uses the dynamic game strategy decomposition algorithm to synergistically optimize the traffic signal parameters. This optimization method breaks the limitation of only focusing on a single objective in traditional regulation. By integrating the Nash equilibrium solution and the constraint relaxation mechanism, it can dynamically adjust the traffic signal phase switching variables and the green light duration adjustment variables, better adapting to the complex and changeable traffic flow changes. For example, in the case of large differences in traffic flow at different times and on different sections of the road, it can reasonably allocate the green light time, not only ensuring the traffic efficiency of the main road, but also avoiding excessive congestion on some branch roads, and achieving the load balance of the entire road network.

[0022] The hierarchical execution module outputs the optimal regulation strategy based on the multi-objective game optimization model, and realizes the distributed execution of the regulation instructions through a three-level control architecture. The central decision-making layer generates a global phase timing plan based on the road network topology, controlling the overall traffic rhythm from a macroscopic level; the regional coordination layer uses a sliding window prediction algorithm to dynamically adapt the traffic parameters of the sub-region, and can flexibly adjust the regulation strategy within the region according to the real-time traffic conditions; the intersection execution layer realizes the tracking of the signal cycle and the green signal ratio based on the fuzzy model predictive control algorithm, and conducts refined regulation for the specific traffic conditions of each intersection. This hierarchical execution architecture has clear division of labor and coordinated cooperation, greatly improving the accuracy and timeliness of the execution of the regulation instructions, and ensuring that the traffic regulation strategy can be efficiently implemented in the actual scenario.

[0023] In addition, the key algorithms of each module in the system have been optimized and improved. For example, the three-dimensional convolution kernel adopts a multi-scale dilated convolution structure to improve the feature extraction efficiency; the spatial graph convolution layer adopts a dynamic adjacency encoding mechanism to enhance the spatial adaptability; the dynamic benefit adjustment mechanism realizes more flexible adjustment of the utility function based on the online reinforcement learning strategy; the error modeling of the state transition equation adopts the interval envelope analysis technology to improve the prediction accuracy, etc. The optimization of these algorithms further improves the performance and adaptability of the system, enabling the system to operate more stably and efficiently in the face of complex traffic scenarios and sudden traffic events, providing strong technical support for alleviating urban traffic congestion, improving road traffic efficiency, and ensuring traffic safety. Brief Description of the Drawings

[0024] Figure 1 It is the working principle diagram of the electronic information traffic flow automatic regulation system based on the wireless sensor network described in the present invention; Figure 2 It is the working flow chart of the data fusion of the multi-source perception module; Figure 3 It is the working flow chart of the distributed decision-making model; Figure 4 It is the working flow chart of the dynamic adjacency encoding mechanism of the spatial graph convolution layer. Detailed Embodiment

[0025] Next, in combination with the accompanying drawings in the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

[0026] Please refer to Figures 1 - 4 , the present invention provides an automatic traffic flow regulation system for electronic information based on a wireless sensor network, and its specific implementation will be elaborated in detail below.

[0027] Multi-source perception module: Use a wireless sensor network to collect multi-dimensional traffic data of the road network. Among them, the vehicle-mounted positioning terminal can obtain information such as the position and speed of vehicles in real time; the geomagnetic traffic sensor accurately counts the traffic flow of a section by sensing the magnetic field changes generated when vehicles pass through; the video monitoring unit takes videos through cameras to visually monitor the traffic scene and can identify traffic events such as traffic accidents and vehicle violations; the meteorological monitoring station collects meteorological data such as weather conditions and visibility. These multi-dimensional traffic data provide comprehensive basic information for subsequent analysis and regulation.

[0028] Spatio-temporal feature fusion module: Based on the spatio-temporal attention mechanism, cross-modal feature association is performed on the multi-dimensional traffic data collected by the multi-source perception module. Different types of traffic data have different characteristics and importance in terms of time and space. The spatio-temporal attention mechanism can automatically learn the association relationships between these features, highlight key information, suppress redundant information, and thus generate traffic situation features that integrate spatio-temporal characteristics, providing a more valuable basis for subsequent decision-making.

[0029] Collaborative decision-making module: Input the traffic situation features generated by the spatio-temporal feature fusion module into a pre-trained distributed decision-making model. This model is trained with a large amount of historical traffic data and has the ability to analyze and make decisions on complex traffic conditions. By inputting the current traffic situation features, the model can generate a set of regulation instructions for regional traffic flow, and these instruction sets contain traffic regulation strategies for each intersection and section.

[0030] Dynamic game optimization module: Construct a multi-objective game optimization model according to the set of regulation instructions generated by the collaborative decision-making module. This model aims to maximize the traffic efficiency and balance the road network load, and uses a dynamic game strategy decomposition algorithm to collaboratively optimize the traffic signal parameters. In the actual traffic scenario, the traffic demands of different intersections and sections affect each other. Through the dynamic game strategy decomposition algorithm, the optimal configuration of traffic signal parameters can be achieved considering the interests and constraints of all parties, thereby improving the operation efficiency of the entire road network.

[0031] Hierarchical execution module: Based on the optimal regulation strategy output by the dynamic game optimization module, it realizes the distributed execution of regulation instructions through a three-level control architecture. The central decision-making layer generates a global phase timing plan according to the road network topology, comprehensively considering the global traffic conditions, and regulates the traffic flow macroscopically; the regional coordination layer uses a sliding window prediction algorithm to dynamically adapt the sub-region traffic parameters and adjusts the local regulation strategy according to the real-time traffic changes; the intersection execution layer realizes the tracking of the signal cycle and the green signal ratio based on the fuzzy model predictive control algorithm, and accurately controls the switching of the traffic lights according to the actual traffic flow and phase deviation at the intersection to ensure the efficient operation of the intersection traffic.

[0032] The following elaborates on other technical features of the present invention in the form of embodiments. Embodiment 1

[0033] In this embodiment, the data fusion method of the multi-source perception module is described in detail. First, the spatio-temporal calibration of the in-vehicle positioning terminal data and the geomagnetic traffic sensor data is performed. There are differences in time and space between the vehicle position and time information obtained by the in-vehicle positioning terminal and the traffic flow data monitored by the geomagnetic traffic sensor. Through spatio-temporal calibration, the data of both are unified into the same spatio-temporal coordinate system. For example, according to the driving trajectory and timestamp of the vehicle, combined with the position information of the geomagnetic traffic sensor, the accurate time and flow data of each vehicle passing through each section are determined, and then a real-time traffic flow distribution heat map of the road network is constructed. The heat map intuitively shows the traffic flow density in different regions of the road network through color changes, with red indicating high traffic flow density and green indicating low traffic flow density.

[0034] For the video monitoring unit data and the meteorological monitoring station data, a multi-scale wavelet decomposition method is used to extract traffic event feature vectors. The video captured by the video monitoring unit contains rich traffic scene information, and the data of the meteorological monitoring station reflects the impact of weather on traffic. Multi-scale wavelet decomposition can analyze the data at different time and space scales, extract the characteristics of traffic events such as traffic accidents and congestion, and the impact characteristics of meteorological factors such as heavy rain and fog on traffic.

[0035] Construct a dual-branch feature fusion network. The first branch uses a three-dimensional convolution kernel to extract the spatial correlation features of the traffic flow distribution heat map. The three-dimensional convolution kernel performs convolution operations on the heat map in three spatial dimensions, which can capture the spatial relationships between traffic flows in different regions, such as the mutual influence of traffic flows on adjacent sections. The second branch uses a gated time series network to extract the dynamic evolution law of traffic event feature vectors. The gated time series network can learn the changes of traffic event features over time according to time series data, such as the impact changes of traffic accidents on traffic from the occurrence to the handling process.

[0036] Fuse the spatial correlation features and dynamic evolution features through a cross-channel attention mechanism to generate a spatio-temporal joint matrix. The cross-channel attention mechanism can automatically allocate weights to features in different channels, highlighting important feature information. Based on residual dilated convolution, multi-granularity feature modeling is performed on the spatio-temporal joint matrix, and traffic situation features including traffic flow density, event influence range, and meteorological interference intensity are output. The residual dilated convolution expands the receptive field by introducing dilated convolution kernels, enabling the extraction of richer multi-granularity features and making the output traffic situation features more comprehensive and accurate.

[0037] In this process, if relevant calculations are involved, taking multi-scale wavelet decomposition as an example, assume the signal , and its wavelet decomposition formula is: ; where, represents the wavelet transform coefficient, t is the time variable, a is the scale parameter, which determines the stretching degree of the wavelet function and controls the decomposed frequency range; b is the translation parameter, which determines the position of the wavelet function on the time axis; represents the conjugate function of the wavelet function. By selecting appropriate wavelet basis functions and different scale and translation parameters, effective multi-scale decomposition can be performed on the data of video monitoring units and meteorological monitoring stations, and valuable feature vectors can be extracted. Embodiment 2

[0038] This embodiment focuses on introducing the distributed decision-making model. The distributed decision-making model adopts a spatio-temporal graph convolutional network architecture. First, a road network-vehicle interaction graph is constructed. The nodes in the graph include intersection nodes, road segment nodes, vehicle nodes, and meteorological nodes. Each node has specific attributes. For example, the queue length of an intersection node reflects the number of vehicles waiting to pass at that intersection; the average vehicle speed of a road segment node reflects the driving speed of vehicles on that road segment; the vehicle node can contain information such as the type and driving direction of the vehicle; the visibility parameter of the meteorological node is used to measure the impact of weather conditions on traffic.

[0039] The dual-modal graph attention mechanism is adopted. In the first stage, the association weights between intersection nodes and adjacent road segments are calculated through the spatial graph convolutional layer. The spatial graph convolutional layer adopts a dynamic adjacency encoding mechanism to define the dynamic association vector between intersection nodes and adjacent road segments. This vector contains information such as vehicle turning probability, lane connection relationship, and historical traffic efficiency. Through the non-linear mapping layer, the dynamic association vector is converted into the bias weight parameter of the graph convolution kernel, and then the bias weight parameter is fused into the graph convolution operation process to generate a node feature representation with spatial adaptability. For example, if the vehicle turning probability to the left at a certain intersection is relatively high, and the connection relationship between the turning lane and the adjacent road segment is close, and the historical traffic efficiency is good, then a higher weight will be given to the adjacent road segment when calculating the association weight, highlighting its important impact on the intersection traffic.

[0040] In the second stage, the importance of historical traffic states is screened through the temporal graph convolutional layer. The temporal graph convolutional layer can analyze historical traffic data according to the time series and screen out the historical state information that has an important impact on the current decision-making. For example, during the morning rush hour, analyze the traffic congestion situation in the same period of the past week to provide reference for the current regulation decision-making.

[0041] Based on the multi-head spatio-temporal attention module, the node features are iteratively optimized. Each attention head fuses the static road network attributes and dynamic traffic parameters. The skip connection and layer normalization mechanism are used to improve the stability of network training. The skip connection can avoid the problem of gradient disappearance, enabling the network to better learn deep features; the layer normalization mechanism normalizes the input of each layer to accelerate the convergence of the network. Finally, the model outputs a set of regulation instructions including priority passage for emergency vehicles and regional congestion evacuation. For example, plan priority passage routes for emergency vehicles such as fire trucks and ambulances, and formulate evacuation strategies for congested areas. Embodiment 3

[0042] This embodiment elaborates in detail the dynamic game strategy decomposition algorithm that integrates the Nash equilibrium solution and the constraint relaxation mechanism. First, the traffic signal optimization problem is modeled as a multi-objective mixed integer dynamic game problem. The decision variables include discrete signal phase switching variables and continuous green light duration adjustment variables. The signal phase switching variable determines the switching timing of the signal lamp between different phases, such as switching from the straight phase to the left turn phase; the green light duration adjustment variable is used to adjust the green light time length of each phase to adapt to different traffic flow demands.

[0043] Initialize the relaxed game sub-problem and calculate the approximate set of equilibrium solutions. In the actual traffic scenario, there are various complex constraint conditions, such as the maximum and minimum green light time limits at intersections, the right-of-way priorities of vehicles in different directions, etc. Through the constraint relaxation mechanism, these constraint conditions are appropriately relaxed, and the complex global game problem is transformed into multiple relatively simple relaxed game sub-problems. The dynamic revenue adjustment mechanism is used to update the utility function of the game participants according to the real-time traffic demand. For example, when the traffic flow on a certain road suddenly increases, the weight of the utility function for vehicle passage in that road direction is correspondingly increased to encourage more green light time to be allocated to that direction.

[0044] In the decomposition stage, the global game problem is divided into multiple sub-game problems based on the Pareto front analysis technique. The Pareto front analysis technique can find the optimal trade-off relationship among multiple optimization objectives, decompose the global problem into multiple sub-problems, and each sub-problem corresponds to a local optimization objective. In the coordination stage, a virtual player is introduced to balance the interest conflicts among sub-problems. The virtual player can be regarded as a coordinator, and by adjusting the decision-making strategies of each sub-problem, the traffic flow of the entire road network can reach the optimal balanced state.

[0045] A parallel game solver is used to alternately optimize the sub-problems, update the local equilibrium solution in each iteration, and synchronize the global game strategy through the coordination variable. The parallel game solver can handle multiple sub-problems simultaneously, improving the solving efficiency. In each iteration process, according to the local equilibrium solutions of each sub-problem, the global game strategy is adjusted through the coordination variable, gradually approaching the optimal traffic signal control scheme.

[0046] Taking the calculation of the Nash equilibrium solution as an example, assume that there are n players in the game, and the strategy set of each player i is , and the utility function is , where . The Nash equilibrium solution satisfies that for any , there is , and for all . That is, each player selects its own optimal strategy when the strategies of other players are fixed, thus reaching a stable balanced state. In the traffic signal optimization problem of the present invention, by continuously adjusting the decision variables of each intersection and road section, the optimal solution that satisfies the Nash equilibrium condition is found to achieve the optimal control of traffic signals. Embodiment 4

[0047] This embodiment details the process of dynamically adapting the sub-region traffic parameters by the sliding window prediction algorithm. First, a time-varying traffic flow prediction model is constructed, and the vehicle kinematic equation is discretized into a state transition equation. The vehicle kinematic equation describes the movement law of vehicles on the road. Considering various factors in actual traffic, such as car-following model error, lane-changing decision delay, and signal control influence terms, etc., it is discretized. Assume that the state of the vehicle at time k is , and the state transition equation can be expressed as: ; where is the state transition matrix, which describes the change relationship of the vehicle state over time, and its elements are related to factors such as the vehicle's dynamic characteristics and road conditions; is the input matrix, which is used to map the control input to the state space, It can represent control variables such as the acceleration and steering angle of a vehicle; is the process noise, which includes uncertain factors such as car-following model errors, lane-changing decision-making delays, and signal control influence terms. The interval envelope analysis technology is used to model its errors. The vehicle motion uncertainty is modeled as interval parameters, and the upper and lower bounds are determined by the standard deviation of sensor measurement errors. The interval parameters are orthogonally decomposed to separate the deterministic components of the system and the environmental random interference components.

[0048] Design a dynamic window optimization objective function, which includes a queue length suppression term, a passing delay balancing term, and an emergency vehicle priority passing guarantee term. The queue length suppression term is used to reduce the queue length of vehicles at intersections and improve the passing capacity of intersections; the passing delay balancing term ensures that the passing delays of vehicles in all directions are as balanced as possible, avoiding excessive waiting time for vehicles in a certain direction; the emergency vehicle priority passing guarantee term provides priority passing guarantees for emergency vehicles such as fire trucks and ambulances. For example, the queue length suppression term can be expressed as ; where is the set of all lanes at the intersection, is the number of queuing vehicles in lane i. By minimizing this term, the queue length can be effectively reduced. By adjusting the weights of these three terms, different optimization objectives can be balanced according to the actual traffic demand, realizing the dynamic adaptation of sub-region traffic parameters. Example 5

[0049] This example focuses on the intersection execution layer to achieve the tracking of signal cycle and green ratio based on the fuzzy model predictive control algorithm. Design a three-layer fuzzy predictive inference system to map the phase deviation and traffic flow fluctuation to the fuzzy rule confidence. The phase deviation refers to the difference between the actual phase and the ideal phase at the current intersection, and the traffic flow fluctuation reflects the change of traffic flow. The fuzzy predictive inference system includes three processes: fuzzification, fuzzy inference, and defuzzification. In the fuzzification process, the exact values of the phase deviation and traffic flow fluctuation are converted into membership degrees in the fuzzy set. For example, the phase deviation is divided into fuzzy sets such as "negative large", "negative small", "zero", "positive small", "positive large", etc., and the membership degree in each fuzzy set is determined according to the actual phase deviation value.

[0050] In the fuzzy inference process, according to the pre-set fuzzy rules, combined with the fuzzy membership degrees of the phase deviation and traffic flow fluctuation, the adjustment strategies of the signal cycle and green ratio are inferred. For example, if the phase deviation is "positive large" and the traffic flow fluctuation is "large", the fuzzy rule may indicate increasing the signal cycle and the green ratio in the corresponding direction. The defuzzification process converts the result obtained by fuzzy inference into an exact control variable, which is used to adjust the signal cycle and green ratio of the signal light.

[0051] Build an adaptive domain adjustment mechanism to dynamically adjust the coverage interval of the fuzzy set according to the traffic state change rate. When the traffic state changes drastically, appropriately expand the coverage interval of the fuzzy set to improve the system's adaptability; when the traffic state is relatively stable, narrow the coverage interval of the fuzzy set to improve the control accuracy. For example, during the traffic peak period, the traffic flow changes greatly. At this time, expand the coverage interval of the fuzzy set so that the fuzzy model can better handle various complex traffic conditions; during the off-peak traffic period, the traffic flow is relatively stable, narrow the coverage interval of the fuzzy set to achieve more accurate signal control. Embodiment 6

[0052] This embodiment details the specific method of using a multi-scale dilated convolution structure for the three-dimensional convolution kernel to improve the feature extraction efficiency. First, divide the real-time traffic flow distribution heat map of the road network into multi-level grid cells, and each cell stores the variance of vehicle flow speed and the density gradient statistic. The variance of vehicle flow speed reflects the degree of dispersion of vehicle speeds, and the density gradient statistic reflects the spatial change of vehicle flow density. By storing these statistics, the traffic characteristics of each grid cell can be more comprehensively described.

[0053] In the feature extraction stage, use a dilated convolution kernel to expand the spatial perception range. When performing convolution operations, the dilated convolution kernel can cover a larger spatial area without increasing the number of parameters by setting the dilation rate. For example, an ordinary convolution kernel can only perceive adjacent pixel points, while the dilated convolution kernel can perceive pixel points in a farther area by setting an appropriate dilation rate, thereby capturing more extensive spatial correlation features. In the feature reconstruction stage, restore the local detailed features through a transposed convolution layer. The transposed convolution layer can be regarded as the inverse operation of the convolution layer. It can restore the feature map that has been reduced after convolution operations to the original size while retaining important detailed information.

[0054] Introduce a spatial attention gating mechanism to perform dynamic weighted fusion on the feature map. The spatial attention gating mechanism can automatically assign weights according to the importance of different regions in the feature map. For regions with large traffic flow changes and great impact on the overall traffic situation, give higher weights to highlight the feature information of these regions; for relatively unimportant regions, reduce the weights to suppress redundant information. Through this dynamic weighted fusion method, traffic situation features can be more effectively extracted, and the efficiency and accuracy of feature extraction can be improved.

[0055] Through the detailed elaboration of the above embodiments, the electronic information traffic flow automatic regulation system based on a wireless sensor network of the present invention has been fully described in all technical details. From multi-source perception, feature fusion, decision optimization to instruction execution, each link cooperates closely to achieve efficient automatic regulation of traffic flow, with remarkable practicality and innovation.

[0056] It should be noted that, in this document, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device.

[0057] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An automatic traffic flow regulation system for electronic information based on a wireless sensor network, characterized in that, include: Multi-source sensing module: used to collect multi-dimensional traffic data of the road network through a wireless sensor network, which includes a vehicle-mounted positioning terminal, a geomagnetic flow sensor, a video monitoring unit and a meteorological monitoring station; Spatiotemporal feature fusion module: Based on the spatiotemporal attention mechanism, the multi-dimensional traffic data is cross-modal feature associated to generate traffic situation features that integrate spatiotemporal characteristics; Collaborative decision-making module: input the traffic situation characteristics into the pre-trained distributed decision-making model to generate a regional traffic flow control instruction set; Dynamic game optimization module: construct a multi-objective game optimization model according to the control instruction set, the multi-objective game optimization model takes maximizing traffic efficiency and balancing road network load as optimization goals, and adopts a dynamic game strategy decomposition algorithm to collaboratively optimize traffic signal parameters; Hierarchical execution module: Based on the multi-objective game optimization model, the optimal control strategy is output, and the distributed execution of control instructions is realized through a three-level control architecture. The three-level control architecture includes a central decision-making layer, a regional coordination layer and an intersection execution layer. The central decision-making layer generates a global phase timing plan based on the road network topology, the regional coordination layer uses a sliding window prediction algorithm to dynamically adapt the traffic parameters of the sub-region, and the intersection execution layer tracks the signal cycle and the green-to-signal ratio based on a fuzzy model prediction control algorithm.

2. The electronic information traffic flow automatic regulation system according to claim 1, characterized in that The data fusion method of the multi-source perception module includes: Perform spatiotemporal calibration on the vehicle positioning terminal data and the geomagnetic flow sensor data to construct a real-time traffic distribution heat map of the road network; perform multi-scale wavelet decomposition on the video monitoring unit data and the meteorological monitoring station data to extract the traffic event feature vector; Construct a dual-branch feature fusion network, wherein the first branch uses a three-dimensional convolution kernel to extract the spatial correlation features of the traffic distribution heat map, and the second branch uses a gated timing network to extract the dynamic evolution law of the traffic event feature vector; The spatial correlation features and dynamic evolution features are fused through the cross-channel attention mechanism to generate a spatiotemporal joint matrix. Multi-granularity feature modeling is performed on the spatiotemporal joint matrix based on residual dilated convolution to output traffic situation characteristics including vehicle flow density, event impact range and meteorological interference intensity.

3. The electronic information traffic flow automatic regulation system according to claim 1, wherein, The distributed decision-making model adopts a spatiotemporal graph convolutional network architecture to generate multi-level control instructions based on a dynamic node aggregation mechanism; the spatiotemporal graph convolutional network architecture includes: Construct a road network-vehicle interaction graph, in which nodes include intersection nodes, road section nodes, vehicle nodes and meteorological nodes, and node attributes include queue length, average vehicle speed and visibility parameters; A bimodal graph attention mechanism is used. In the first stage, the association weights between intersection nodes and adjacent road sections are calculated through a spatial graph convolution layer. In the second stage, the importance of historical traffic status is screened through a temporal graph convolution layer. Based on the multi-head spatiotemporal attention module, node features are iteratively optimized. Each attention head integrates static road network attributes and dynamic traffic parameters. The network training stability is improved through skip connections and layer normalization mechanisms. The final output is a control instruction set that includes emergency vehicle priority passage and regional congestion evacuation.

4. The automatic traffic flow regulation system for electronic information according to claim 1, characterized in that, The dynamic game strategy decomposition algorithm integrates Nash equilibrium solution and constraint relaxation mechanism, including: Model the traffic signal optimization problem as a multi-objective mixed-integer dynamic game problem, where the decision variables include discrete signal phase switching variables and continuous green light duration adjustment variables; Initialize the relaxed game sub-problem and calculate the approximate set of equilibrium solutions, and adopt a dynamic payoff adjustment mechanism to update the utility functions of game participants according to real-time traffic demands; In the decomposition stage, divide the global game problem into multiple sub-game problems based on the Pareto front analysis technique; in the coordination stage, introduce a virtual game player to balance the interest conflicts between sub-problems; Use a parallel game solver to alternately optimize the sub-problems, update the local equilibrium solutions in each iteration, and synchronize the global game strategies through coordination variables.

5. The electronic information traffic flow automatic regulation system according to claim 1, characterized in that, The sliding window prediction algorithm dynamically adapts to traffic parameters in sub-regions, including: Construct a time-varying traffic flow prediction model, discretize the vehicle kinematic equation into a state transition equation, and the state transition equation includes a car-following model error, a lane-changing decision delay, and a signal control influence term; Design a dynamic window optimization objective function, including a queue length suppression term, a travel delay equilibrium term, and an emergency vehicle priority passage guarantee term.

6. The automatic regulation system for electronic information traffic flow according to claim 1, wherein, The intersection execution layer realizes the tracking of signal cycle and green signal ratio based on the fuzzy model predictive control algorithm, including: Design a three-layer fuzzy prediction inference system to map the phase deviation and traffic flow fluctuation to the fuzzy rule confidence; construct an adaptive universe adjustment mechanism to dynamically adjust the coverage interval of the fuzzy set according to the traffic state change rate.

7. The automatic traffic flow regulation system for electronic information according to claim 2, characterized in that The three-dimensional convolution kernel adopts a multi-scale dilated convolution structure to improve the feature extraction efficiency, including: Divide the real-time traffic flow distribution heat map of the road network into multi-level grid cells, and each cell stores the statistical quantities of the vehicle speed variance and density gradient; In the feature extraction stage, use a dilated convolution kernel to expand the spatial perception range, and in the feature reconstruction stage, restore the local detailed features through a transposed convolution layer; Introduce a spatial attention gating mechanism to dynamically weight and fuse the feature maps.

8. The automatic regulation system for electronic information traffic flow according to claim 3, characterized in that The spatial graph convolution layer adopts a dynamic adjacency encoding mechanism, including: Define the dynamic association vector between intersection nodes and adjacent road segments, including vehicle turning probability, lane connection relationship, and historical passing efficiency; Convert the dynamic association vector into the bias weight parameters of the graph convolution kernel through a non-linear mapping layer; Fuse the bias weight parameters into the graph convolution operation process to generate a node feature representation with spatial adaptability.

9. The automatic traffic flow regulation system for electronic information according to claim 4, wherein The dynamic payoff adjustment mechanism is realized based on an online reinforcement learning strategy, including: Collect the strategy selection records and equilibrium solution distributions in the historical game process as the training data set; Construct a deep Q-learning network to fit the change law of the payoff function of game participants; Online update the network parameters through the asynchronous policy gradient descent method, and adjust the utility function weights of each participant in real time; When a sudden traffic event is detected, trigger the emergency reconstruction operation of the payoff function.

10. The automatic traffic flow regulation system for electronic information according to claim 5, characterized in that The error modeling of the state transition equation adopts the interval envelope analysis technique, including: modeling the vehicle motion uncertainty as an interval parameter, and the upper and lower bounds are determined by the standard deviation of the sensor measurement error; performing orthogonal decomposition on the interval parameter to separate the deterministic component of the system and the environmental random interference component.

Images