Intelligent traffic control method and system, roadside edge computing device and storage medium

Abstract

The invention relates to the technical field of intelligent traffic, and discloses an intelligent traffic control method and system, roadside edge computing equipment and a storage medium, and the method comprises the following steps: collecting traffic state information obtained by video monitoring equipment, radar equipment and manual input equipment which are arranged at a traffic intersection and a surrounding area of the traffic intersection; the traffic state information comprises vehicle queuing length, average speed and waiting time; performing normalization processing on the traffic state information, calculating an entropy value of each data source, and calculating and generating a weight of each data source based on the entropy value; and performing weighted fusion on the traffic state information based on the weight to obtain a fused traffic state estimation vector, and constructing Bayesian belief distribution based on the vector to represent a belief value of the traffic state. According to the method, the Bayesian belief distribution and the game model are combined, so that real-time intelligent optimal scheduling of traffic signals is realized, traffic flow management is effectively improved, and signal waste is reduced.

Description

Technical Field

[0001] With the acceleration of urbanization, traffic management has become a critical issue in urban development. Optimizing traffic signal scheduling, especially during peak hours, directly impacts road efficiency and safety. However, traditional traffic signal control systems often fail to effectively respond to fluctuations in traffic flow, leading to congestion and unnecessary waiting, wasting significant time and resources. Therefore, intelligently controlling traffic signals to maximize traffic flow has become a challenging technical challenge.

[0002] Existing technologies often rely on preset time periods and traffic statistics, switching signals according to fixed cycles. This approach is simple and easy to implement, ensuring basic traffic control and regulation, and is particularly suitable for areas with relatively stable traffic flows. For example, some intelligent traffic signal systems can adjust signal cycles based on traffic flow data, avoiding the inefficiencies associated with consistent signal cycles throughout the day. Furthermore, some existing systems are capable of adjusting signal durations for different directions within an intersection by setting priority policies, thereby achieving a certain degree of traffic optimization.

[0003] However, existing technologies still have some shortcomings. First, traditional signal control schemes are often based on fixed models or traffic counts and cannot adapt to complex traffic flow fluctuations. Especially during peak hours or when emergencies occur, the system's response speed is slow and cannot be adjusted immediately, resulting in increased traffic congestion. Secondly, most existing systems do not take into account the mutual constraints and coordinated regulation between various directions of traffic, resulting in the priority of one direction affecting the traffic capacity of other directions, causing unnecessary traffic conflicts and delays. Finally, signal control in existing technologies mostly relies on manual settings or historical data deduction, lacks the ability to make real-time intelligent decisions, and cannot respond to sudden traffic conditions in a timely manner. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides an intelligent traffic control method, system, roadside edge computing device and storage medium, which solves the problem in the existing technology that traffic signal control cannot adapt to dynamic traffic flow changes in real time.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: an intelligent traffic control method, comprising the following steps: Collecting traffic status information obtained by video surveillance equipment, radar equipment, and manual input equipment installed at traffic intersections and their surrounding areas, the traffic status information including vehicle queue length, average speed, and waiting time; Normalize the traffic status information, calculate the entropy value of each data source, and calculate and generate the weight of each data source based on the entropy value; Performing weighted fusion on the traffic state information based on the weights to obtain a fused traffic state estimation vector, and constructing a Bayesian belief distribution based on the vector to represent a belief value of the traffic state; According to the Bayesian belief distribution, a game model including a strategy set of left-turn traffic direction and straight-through traffic direction is constructed, and the optimal strategy of the game participants is obtained by reasoning; A signal phase switching control instruction is generated based on the optimal strategy, and the control instruction is sent to a signal control device to execute corresponding phase switching.

[0006] Preferably, the step of collecting video surveillance equipment located at the traffic intersection and its surrounding areas includes: Obtain real-time video streams of traffic intersections and analyze vehicle queue information, vehicle types, and flow status in the images; Use radar equipment to measure the speed of traffic flow and synchronize the speed measurement data with video data; Real-time traffic control instructions and emergency information are input through manual input devices.

[0007] Preferably, the step of normalizing the traffic status information includes: Normalize the collected vehicle queue length, average speed and waiting time to a unified value range; The traffic status information is processed using the minimum to maximum normalization method, and the formula is: Among them, x is the original data; x ′ is the normalized data; min(x) and max(x) are the minimum and maximum values ​​of the data dimension respectively; The normalized data were normalized by standard deviation.

[0008] Preferably, the step of performing standard deviation normalization processing includes: For the normalized traffic status information, the standard deviation of each data dimension is calculated to quantify the degree of dispersion of its data distribution; based on the calculated standard deviation, each normalized data value is adjusted to meet the requirements of standard deviation normalization; during the standard deviation normalization process, an appropriate scaling factor is used based on the actual fluctuation of the data to balance the fluctuation range of each traffic status data.

[0009] Preferably, the step of performing weighted fusion on the traffic status information based on the weights includes: According to the entropy weight of each data source, the traffic state information is weighted averaged to obtain the fused traffic state estimation vector. The formula is: Among them, x i is the normalized traffic status data; w i is the weight obtained by entropy calculation; f(x) is the traffic state estimation vector after weighted fusion; n is the total number of traffic state indicators; During the fusion process, different weights are assigned according to different sources of traffic status information.

[0010] Preferably, the step of constructing a game model including a strategy set for a left-turn traffic direction and a straight-through traffic direction comprises: In the game model, the left-turn direction and the straight-through direction are respectively considered as participants in the game, and the strategy set includes two strategies: "request phase switching" and "maintain current phase"; The revenue function of the game model considers the queue length and traffic efficiency of each direction of travel; the waiting time of each direction of travel; and the calculation of the potential conflict cost when two parties request a phase switch at the same time.

[0011] Preferably, the step of generating a signal phase switching control instruction based on the optimal strategy includes: The optimal strategy inferred by the game model is mapped into traffic light control instructions, specifically determining the time window for phase switching based on the optimal strategy; The optimal strategy is "request phase switching", which generates a control instruction to change the current signal light phase; The optimal strategy is "maintain the current phase", which means that the current traffic light state remains unchanged until the next decision cycle.

[0012] It also provides intelligent traffic control systems, including: An information collection and normalization module is used to receive traffic status information from video surveillance equipment, radar equipment, and manual input equipment at the traffic intersection and its surrounding areas, and to normalize the information; a weight determination and fusion module, which calculates the entropy value of each data source based on the normalized traffic state information and performs a weighted fusion operation based on the weight to generate a fused traffic state estimation vector for representing the overall traffic state; Traffic state inference module, which constructs a Bayesian belief distribution based on the fused traffic state estimation vector to represent the uncertainty and belief level of the current traffic state; The game reasoning and instruction generation module is used to construct a game model containing a set of strategies for left-turn and straight-through directions based on the belief distribution, infer the optimal strategy of the game participants, and then generate the corresponding signal phase switching control instructions; The signal control execution module is used to receive the control instruction and drive the signal control device to execute the phase switching operation corresponding to the optimal strategy to achieve dynamic adjustment of signal timing.

[0013] A roadside edge computing device is also provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, an intelligent traffic control method can be implemented.

[0014] A storage medium is also provided, which stores a computer program. When the computer program is executed by a processor, the intelligent traffic control method can be implemented.

[0015] The present invention provides an intelligent traffic control method, system, roadside edge computing device and storage medium. Beneficial effects: 1. The present invention adopts a technical solution that combines Bayesian belief distribution with a game model, achieving the ability to accurately infer traffic flow and priority in complex traffic environments. Compared with traditional timing control methods in the prior art, the present invention can adjust signal timing in real time according to traffic conditions, solving the problem that static signal control cannot adapt to dynamic traffic flow changes.

[0016] 2. This invention improves the adaptability and real-time performance of traffic signal control by weightedly fusing multiple traffic data sources and generating signal control instructions. Compared to traditional signal control technologies based on simple traffic flow counting, this invention solves the problem of slow response to large traffic fluctuations, enabling more flexible response to traffic peaks and emergencies.

[0017] 3. This invention significantly improves signal control accuracy and scheduling efficiency by generating signal phase switching control instructions based on an optimal strategy. Unlike existing fixed-cycle scheduling methods that rely on manual intervention or historical data, this invention optimizes phase switching decisions through intelligent reasoning, avoiding signal waste and traffic congestion.

[0018] 4. The present invention introduces game theory to optimize traffic signal control strategies, thereby reducing traffic conflicts and improving traffic efficiency. Compared with the existing technology that lacks consideration of the mutual influence between traffic strategies in different directions, the present invention dynamically adjusts signal cycles and priorities through a game model, thereby solving the problem of insufficient coordination of multi-directional traffic and effectively improving the traffic capacity of intersections. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 Schematic diagram of the method flow of the present invention; Figure 2 This is a system architecture diagram of the present invention; Figure 3Schematic diagram of an electronic device of the present invention. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the present specification. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0021] Please see the attached Figure 1 , an embodiment of the present invention provides an intelligent traffic control method, comprising the following steps: S1. Collect traffic status information obtained by video surveillance equipment, radar equipment, and manual input equipment installed at traffic intersections and their surrounding areas. Traffic status information includes vehicle queue length, average speed, and waiting time; The acquisition and processing of traffic status information is fundamental to supporting the subsequent generation of control strategies and the issuance of control instructions. Accurate perception of traffic conditions directly impacts the efficiency of model inference and the relevance of control strategies. To achieve refined, real-time control of intersection traffic conditions, the present invention incorporates multiple types of sensing devices, including video surveillance equipment, radar equipment, and manual input terminals, located at intersections and their surrounding areas. These devices operate collaboratively to capture key traffic status information, including queue lengths, average speeds, and wait times.

[0022] In this embodiment, the video surveillance equipment is mainly used to obtain real-time images of traffic flow. With the help of image recognition algorithms, the vehicle positions can be calibrated in real time, thereby identifying the number of vehicles in the queue and calculating the length of the vehicle queue. Generally, by detecting the relative position of the vehicle bounding box and the road area, the vehicle density can be estimated, and the queue length can be obtained. The queue length L q It can be calculated using the following formula: L q =N v ·l v +(N v -1)·d s ; Among them, L q Indicates the length of the vehicle queue; N v Indicates the number of vehicles currently queuing; l v Indicates the average vehicle length, which can be set as a fixed value or dynamically estimated based on the recognition results; d s Indicates the average distance between vehicles (unit: meter).

[0023] Specifically, the number of queued vehicles N vThis can be obtained through real-time statistics of target detection models (such as the YOLO series). Combined with the geographically marked areas, it can be used to determine whether the vehicle is in a queue state, thus avoiding misjudgment of passing vehicles.

[0024] In one possible implementation, the radar device is installed above the intersection or on the side of the road. Its working mode does not rely on visible light and is applicable to all-weather conditions, especially in low light or bad weather conditions. The radar measures the instantaneous speed of the vehicle through the Doppler effect and calculates the average speed of the vehicle in the section. The average speed V avg It can be given by the following formula: Among them, V avg represents the average speed during the sampling period; N r Indicates the number of detected vehicles; v i represents the instantaneous speed of the i-th vehicle.

[0025] Alternatively, radar data can be fused with video image recognition results to improve speed calculation accuracy. In areas where radar beams overlap, multi-source data verification can be performed by matching vehicle IDs with their paths, eliminating anomalous samples.

[0026] In this embodiment, manual input devices are primarily used to facilitate the collection of emergency or local intervention signals, such as on-site traffic police control information, dispatch instructions, or temporary construction records. Using a mobile terminal or portable input panel, on-duty personnel can directly enter abnormal status flags. For example, inputting "severe left-turn lane congestion" can trigger the model to adjust the duration of the left-turn phase. This input information enters the model processing flow in a structured format to correct or supplement the automated perception results.

[0027] In some embodiments, in order to avoid the impact of data errors on the strategy, the present invention introduces a time-weighted filtering mechanism to make a revised estimate of the vehicle waiting time. w The calculation is as follows: Among them, T w Indicates the average waiting time (unit: seconds); t j represents the actual waiting time of the jth vehicle; w j Indicates the time weighting factor corresponding to the vehicle; M represents the number of statistical vehicle samples.

[0028] In one practical application, if the average waiting time on a main road exceeds a threshold, the system can trigger priority traffic flow and adjust traffic light timing. The threshold setting can be adaptively learned based on the fluctuation range of historical average values.

[0029] Furthermore, to enhance system stability and redundancy, the present invention supports redundant sampling of sensor data. If any device fails or experiences inaccurate data, the system automatically uses a backup data source to fill the gap, ensuring overall data continuity.

[0030] Furthermore, before entering the dispatch model, multi-source traffic status data undergoes unified feature standardization, such as min-max normalization and Z-score transformation, to adapt to the data format requirements of the language model input interface. This step ensures the consistency and comparability of data acquired by the subsequent semantic reasoning module, improving the accuracy and stability of the output strategy.

[0031] S2. Normalize the traffic status information, calculate the entropy value of each data source, and calculate and generate the weight of each data source based on the entropy value; While acquiring traffic status information constitutes the foundation of data input, for this multi-source, heterogeneous data to provide reliable support for subsequent reasoning and decision-making, it must be standardized and weighted for optimal fusion. Therefore, after completing the initial perception and extraction of traffic status information collected by video surveillance equipment, radar equipment, and manual input terminals, the present invention further incorporates a normalization processing and information entropy weight calculation module to numerically normalize data from different data sources. This module also measures data distribution characteristics using entropy theory to generate data source weights that can be used for fusion calculations, ensuring the rationality of the input data distribution, the salience of its features, and its decision-making relevance.

[0032] In this embodiment, in order to make traffic status data (such as vehicle queue length, average speed, and waiting time) from different data sources comparable, each raw observation data is first normalized. Generally, the minimum-maximum normalization method is used. The raw data x is normalized and the calculation formula is as follows: Among them, x is the original data; x ′ is the normalized data; min(x) and max(x) are the minimum and maximum values ​​of the data dimension respectively.

[0033] As an option, the above normalization process can be applied to the same traffic indicators from different data sources, such as average speed data collected by radar and video equipment, and then input into the next entropy analysis module after being scaled uniformly.

[0034] In a possible implementation, the system calculates the information entropy of each indicator data after normalization to measure the information distribution state of the indicator in the overall sample. ij The calculation is as follows: Among them, pij represents the normalized proportion of sample i on the jth indicator; m represents the total number of samples; x represents the original data.

[0035] In some embodiments, the system calculates the discriminative power of each data indicator based on the entropy value results and generates a corresponding weight. The smaller the entropy value, the more concentrated the indicator distribution, the greater its difference, the higher the amount of information, and the higher the weight should be given. The final weight calculation formula is: Among them, w j Indicates the time weighting factor corresponding to the vehicle; Typically, the system stores the calculated weights for each indicator and uses them as weighting factors in the feature aggregation process during the multi-source data fusion phase. This weight control mechanism effectively balances the reference status of different perception channels in complex traffic scenarios, suppresses the impact of noise redundancy, and improves the perceptual robustness of the strategy model.

[0036] As an extension, the present invention supports a dynamic entropy update mechanism. Within a set period (e.g., every 5 minutes), the system automatically recalculates the entropy and weights of each indicator within the current time period, thereby adapting to the time-varying characteristics of traffic conditions. Furthermore, for abnormal samples (such as data mutations caused by local congestion due to emergencies), the system can introduce a sliding window averaging process to smooth single-point data and prevent mutations from affecting overall entropy fluctuations.

[0037] S3. Perform weighted fusion on the traffic state information based on the weights to obtain a fused traffic state estimation vector, and construct a Bayesian belief distribution based on the vector to represent the belief value of the traffic state; After normalizing multi-source traffic status data and determining the weights of various indicators based on information entropy theory, a unified, structured state estimation vector must be constructed to support the subsequent reasoning and execution of the intelligent traffic control model. The core of this stage is to fuse the normalized and weighted data into a unified vector representing the current traffic state and, based on this, establish a Bayesian belief distribution to quantitatively express the system's confidence in the current traffic state. This invention integrates the state data obtained from multiple perception sources through a weighted fusion method and constructs a belief expression model using Bayesian theory, achieving robust modeling of traffic states under uncertain conditions.

[0038] In this embodiment, after completing the normalization processing of each traffic state indicator and the entropy value weight calculation, the weight w corresponding to each indicator is calculated. i , weighted fusion is performed on the normalized data to obtain the fused traffic state estimation vector f(x). This estimation vector represents the integration of information from multiple data sources on key states in the current traffic scenario (such as queue length, average speed, waiting time, etc.). The calculation formula is as follows: Among them, x i is the normalized traffic status data; w i is the weight obtained by entropy calculation; f(x) is the traffic state estimation vector after weighted fusion; n is the total number of traffic state indicators.

[0039] In one possible implementation, the fused vector f(x) is further used to construct a Bayesian belief distribution. This belief distribution expresses the system's confidence in different traffic status classifications (e.g., "unimpeded," "mildly congested," "heavily congested," etc.). This belief distribution is based on Bayesian inference, using the fused vector as observational evidence and combining it with prior knowledge to derive a posteriori belief values.

[0040] In general, the fusion state vector is taken as f(x), which is k The likelihood function under is expressed as: Where P(f(x)|θ k ) is in state θ k The observation probability of the fused state vector; Indicates that the mean is μ k , covariance is Σ k The multivariate normal distribution of .

[0041] As an option, the present invention supports a multi-round fusion mechanism. By accumulating f(x) vectors at multiple moments, a time series belief propagation model can be constructed, further improving the robustness of traffic state prediction. In dynamic scenarios, the introduction of a hidden Markov model (HMM) for Bayesian updating helps model state transition probabilities and compensates for inference bias caused by short-term anomalies or data interruptions.

[0042] S4. Based on the Bayesian belief distribution, a game model is constructed that includes a set of strategies for left-turn and straight-through directions, and the optimal strategies of the game participants are derived by reasoning. After obtaining the fused traffic state estimate vector and constructing a Bayesian belief distribution based on it, the system must still rationally select the possible traffic control strategies under the current traffic conditions. To this end, the present invention further proposes a strategic game model that, based on belief-driven reasoning, infers the optimal control behavior of each game participant. This model uses the traffic state characterized by the belief distribution as the decision-making environment, sets a set of strategies including left-turn and straight-through directions, constructs a game framework for traffic participant behavior, and implements strategy evolution reasoning under competitive or cooperative game mechanisms.

[0043] In this example, the game model focuses on key intersection directions, primarily including left turns and straight-ahead traffic. Each direction is abstracted as a player, with a strategy set consisting of control options such as "release," "delay," and "restricted release." The goal of the game is to achieve a balance between traffic efficiency and conflict minimization under different traffic conditions.

[0044] Typically, the system first determines the current traffic status category and its confidence level based on the aforementioned Bayesian belief distribution, and then infers the traffic pressure for each direction. Higher belief values ​​indicate greater traffic density in that direction and a higher degree of urgency for traffic strategy. Optionally, this pressure information can be used to construct a payoff function for participants in each direction.

[0045] Specifically, the strategy choices for turning left and going straight have potential conflicts. For example, when both directions receive the "go" control signal at the same time, there will be a high probability of traffic conflict at the intersection, which will affect traffic safety. To this end, in the game model, each strategy combination is set to correspond to a payoff matrix. The matrix elements are constructed by the following factors: the belief value of the current traffic state; The expected number of vehicles passing in each direction; Potential delay costs and conflict penalties under each strategy combination; Control strategy switching costs.

[0046] The dependence is jointly affected by the delay and the release status of the adjacent through-going direction. This dependence is described by constructing a conditional benefit model and dynamically updated based on the Bayesian belief value.

[0047] In some embodiments, the system uses Nash equilibrium analysis to solve the constructed strategic game model, inferring the optimal strategy combination for each participant under the current traffic conditions. If a strategy combination remains stable without any unilateral deviation incentives from any participant, it is considered a Nash equilibrium solution. This optimal solution serves as the input strategy for signal control in the next cycle, used to update the traffic signal timing status.

[0048] As a specific operational approach, the system incorporates an incomplete information game theory framework into its strategy reasoning process, combining historical strategy execution records to construct a hybrid strategy model. Within this model, each direction can probabilistically choose to execute a strategy based on the uncertainty of the opposing strategy. For example, when traffic is high in the left-turn direction and the belief value in the straight-ahead direction is low, the system will execute the "left-turn priority" strategy with a higher probability, improving overall traffic efficiency.

[0049] Furthermore, to enhance the model's adaptability and real-time performance, the present invention introduces a dynamic game evolution mechanism. During continuous traffic cycles, the system adjusts the weight coefficients of the payoff function based on feedback from each round of strategy implementation (such as changes in queue length and number of vehicles passing through), achieving dynamic iteration and self-optimization of the strategy game framework.

[0050] S5. Generate a signal phase switching control instruction based on the optimal strategy, and send the control instruction to the signal control device to execute the corresponding phase switching; After selecting the optimal strategy based on Bayesian belief distribution inference, the next key task is to translate the optimal strategy into signal control instructions and transmit these instructions to the signal control equipment to execute the corresponding signal phase switching in real time. This process requires close integration with the aforementioned game reasoning module to ensure that each decision effectively reflects the traffic flow state, optimizes traffic signal timing in a timely manner, and avoids traffic congestion or signal waste.

[0051] In this embodiment, the system first converts the optimal strategy's output into signal control instructions. These control instructions specify the switching order and duration of each traffic signal phase within the current cycle, as well as priority settings. For example, if the belief value for straight travel is high during a certain time period, the system will generate a "straight ahead priority" strategy, instructing the signal control device to activate the straight travel phase and extend its duration. Simultaneously, the signal duration for other phases (such as left and right turns) will be shortened accordingly, avoiding unnecessary traffic congestion.

[0052] Specifically, the system extracts the corresponding signal control parameters from the optimal strategy output obtained by game model reasoning, including the signal duration in each direction (for example, T left Indicates the time of the left turn phase, T straight According to these parameters, the system generates control instructions through the signal control algorithm, which involves the signal light cycle P cycle The setting formula is as follows: P cycle =T left +T straight +T turnright ; Among them, P cycle is the total duration of the signal cycle; T left is the duration of the left turn phase; T straight is the duration of the straight phase; T turnright is the duration of the right turn phase.

[0053] As an option, the system can dynamically adjust the duration of each phase based on traffic flow, belief value and conflict weight of each phase to ensure that traffic flow is optimized to the greatest extent possible.

[0054] Next, the generated signal control instructions are sent to the signal control device via a communication interface. This process requires consideration of the data transmission protocol between the system and the signal control device. Generally, the system uses standardized traffic signal control protocols (such as SCOOT and ITS) to ensure that signal instructions are transmitted to the signal control device in a timely and accurate manner. In actual operation, the signal control device will interpret the received instructions and perform the corresponding signal phase switching and duration adjustments based on the instructions.

[0055] Specifically, after the signal control device interprets the instruction, it will control the status of the traffic lights in each traffic direction according to the time allocation specified in the instruction. The control process of the device usually includes the following steps: Start the corresponding signal phase and keep it for a specified time; At the end of the phase, switch to the next signal phase; At the end of the cycle, the next signal cycle begins again.

[0056] In certain embodiments, to enhance system responsiveness and stability, signal control equipment can also adjust command execution based on feedback mechanisms. For example, when traffic flow changes significantly, the equipment can dynamically adjust signal cycles or phase switching sequences by monitoring traffic conditions in real time, further optimizing traffic efficiency.

[0057] As an extension, the system can further optimize generated signal control commands by combining real-time traffic data and historical traffic patterns. By incorporating machine learning algorithms or predictive models, the system can predict the most likely traffic flow changes based on traffic patterns over different time periods and adjust signal phases accordingly, improving traffic flow smoothness and responding to emergencies.

[0058] The intelligent traffic control system described below and the intelligent traffic control method described above may refer to each other.

[0059] Please see the attached Figure 2 The present invention also provides an intelligent traffic control system, comprising: The information collection and normalization module is used to receive traffic status information from video surveillance equipment, radar equipment and manual input equipment at traffic intersections and their surrounding areas, and normalize the information; A weight determination and fusion module calculates the entropy value of each data source based on the normalized traffic state information and performs a weighted fusion operation based on the weights to generate a fused traffic state estimation vector used to represent the overall traffic state; Traffic state inference module, which constructs a Bayesian belief distribution based on the fused traffic state estimation vector to represent the uncertainty and belief level of the current traffic state; The game reasoning and instruction generation module is used to construct a game model containing a set of strategies for left-turn and straight-through directions based on the belief distribution, infer the optimal strategy of the game participants, and then generate the corresponding signal phase switching control instructions; The signal control execution module is used to receive control instructions and drive the signal control device to execute the phase switching operation corresponding to the optimal strategy to achieve dynamic adjustment of signal timing.

[0060] The system of this embodiment can be used to execute the above method embodiments, and its principles and technical effects are similar, so they will not be repeated here.

[0061] The roadside edge computing device described below and the intelligent traffic control method described above can be referenced to each other.

[0062] Please see the attached Figure 3 The present invention also provides a roadside edge computing device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it can execute the above method.

[0063] The present invention also provides a storage medium on which a computer program is stored. When the computer program is run by a processor, the above method is executed.

[0064] Among them, the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0065] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent traffic control method, characterized in that: The following steps are involved: Collecting traffic status information obtained by video surveillance equipment, radar equipment, and manual input equipment installed at traffic intersections and their surrounding areas, the traffic status information including vehicle queue length, average speed, and waiting time; Normalize the traffic status information, calculate the entropy value of each data source, and calculate and generate the weight of each data source based on the entropy value; Performing weighted fusion on the traffic state information based on the weights to obtain a fused traffic state estimation vector, and constructing a Bayesian belief distribution based on the vector to represent a belief value of the traffic state; According to the Bayesian belief distribution, a game model including a strategy set of left-turn traffic direction and straight-through traffic direction is constructed, and the optimal strategy of the game participants is obtained by reasoning; A signal phase switching control instruction is generated based on the optimal strategy, and the control instruction is sent to a signal control device to execute corresponding phase switching.

2. The intelligent traffic control method according to claim 1, characterized in that: The step of collecting video from a video surveillance device located at a traffic intersection and its surrounding area includes: Obtain real-time video streams of traffic intersections and analyze vehicle queue information, vehicle types, and flow status in the images; Use radar equipment to measure the speed of traffic flow and synchronize the speed measurement data with video data; Real-time traffic control instructions and emergency information are input through manual input devices.

3. The intelligent traffic control method according to claim 1, characterized in that: The step of normalizing the traffic status information comprises: Normalize the collected vehicle queue length, average speed and waiting time to a unified value range; The traffic status information is processed using the minimum to maximum normalization method, and the formula is: Among them, x is the original data; x ′ is the normalized data; min(x) and max(x) are the minimum and maximum values ​​of the data dimension respectively; The normalized data were normalized by standard deviation.

4. The intelligent traffic control method according to claim 3, characterized in that: The step of performing standard deviation normalization processing includes: For the normalized traffic status information, the standard deviation of each data dimension is calculated to quantify the degree of dispersion of its data distribution; based on the calculated standard deviation, each normalized data value is adjusted to meet the requirements of standard deviation normalization; During the standard deviation normalization process, an appropriate scaling factor is used based on the actual fluctuation of the data.

5. The intelligent traffic control method according to claim 1, characterized in that: The step of performing weighted fusion on the traffic status information based on the weights comprises: According to the entropy weight of each data source, the traffic state information is weighted averaged to obtain the fused traffic state estimation vector. The formula is: Among them, x i is the normalized traffic status data; w i is the weight obtained by entropy calculation; f(x) is the traffic state estimation vector after weighted fusion; n is the total number of traffic state indicators; During the fusion process, different weights are assigned according to different sources of traffic status information.

6. The intelligent traffic control method according to claim 1, characterized in that: The step of constructing a game model including a strategy set of left-turn traffic direction and straight-through traffic direction includes: In the game model, the left-turn direction and the straight-through direction are respectively considered as participants in the game, and the strategy set includes two strategies: "request phase switching" and "maintain current phase"; The revenue function of the game model considers the queue length and traffic efficiency of each direction of travel; the waiting time of each direction of travel; and the calculation of the potential conflict cost when two parties request a phase switch at the same time.

7. The intelligent traffic control method according to claim 1, characterized in that: The step of generating a signal phase switching control instruction based on the optimal strategy includes: The optimal strategy inferred by the game model is mapped into traffic light control instructions, specifically determining the time window for phase switching based on the optimal strategy; The optimal strategy is "request phase switch", which generates a control instruction to change the current signal light phase; The optimal strategy is "maintain current phase", which means that the current traffic light state remains unchanged until the next decision cycle.

8. An intelligent traffic control system, applied to the intelligent traffic control method according to any one of claims 1 to 7, characterized in that: include: An information collection and normalization module is used to receive traffic status information from video surveillance equipment, radar equipment, and manual input equipment at the traffic intersection and its surrounding areas, and to normalize the information; a weight determination and fusion module, which calculates the entropy value of each data source based on the normalized traffic state information and performs a weighted fusion operation based on the weight to generate a fused traffic state estimation vector for representing the overall traffic state; Traffic state inference module, which constructs a Bayesian belief distribution based on the fused traffic state estimation vector to represent the uncertainty and belief level of the current traffic state; The game reasoning and instruction generation module is used to construct a game model containing a set of strategies for left-turn and straight-through directions based on the belief distribution, infer the optimal strategy of the game participants, and then generate the corresponding signal phase switching control instructions; The signal control execution module is used to receive the control instruction and drive the signal control device to execute the phase switching operation corresponding to the optimal strategy to achieve dynamic adjustment of signal timing.

9. A roadside edge computing device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, it implements the intelligent traffic control method according to any one of claims 1 to 7.

10. A storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a processor, the intelligent traffic control method according to any one of claims 1 to 7 is implemented.

Images