Traffic flow real-time optimization method and device for vehicle infrastructure collaborative edge calculation

Abstract

The invention discloses a traffic flow real-time optimization method and device for vehicle-road collaborative edge calculation, and relates to the technical field of intelligent traffic, and the method comprises the steps: carrying out the real-time collection of a vehicle-mounted unit and roadside equipment in a target road section, and generating a multi-source heterogeneous traffic data set; performing space-time fusion processing on the multi-source heterogeneous traffic data set, and extracting a dynamic traffic flow parameter set; traffic flow optimization is carried out based on the dynamic traffic flow parameter set, and a local optimization instruction set is constructed; and issuing the local optimization instruction set to a vehicle-mounted unit and roadside equipment in the target road section to realize real-time distributed optimization of traffic flow. The technical problem that the traffic flow optimization response is not timely in the prior art is solved, and the technical effect of improving the real-time performance and precision of traffic regulation and control is achieved by fusing the multi-source data of the vehicle-mounted unit and the roadside equipment and performing real-time processing and optimization instruction generation based on edge calculation.

Description

Technical Field

[0001] The present invention relates to the field of smart transportation technology, and specifically to a method and device for real-time traffic flow optimization based on vehicle-road collaborative edge computing. Background Art

[0002] In traffic management systems, traffic flow optimization relies on a central control platform to centrally collect road information and make unified decisions. Data collected by onboard terminals and roadside equipment must be transmitted to remote servers for processing. Limited by network transmission latency and centralized computing load, this makes it difficult to promptly reflect rapidly changing traffic conditions. This leads to delayed generation of optimization instructions, making it difficult to quickly respond to and precisely control traffic flow. This is especially true in urban scenarios with high traffic density and complex road conditions, where traffic scheduling frequently changes. Traditional solutions are unable to coordinate signal control and vehicle speed guidance in real time, impacting overall traffic efficiency. Summary of the Invention

[0003] The present application provides a real-time traffic flow optimization method and device based on vehicle-road collaborative edge computing, which is used to solve the technical problem of untimely traffic flow optimization response in the existing technology.

[0004] In view of the above problems, the present application provides a real-time traffic flow optimization method and device based on vehicle-road collaborative edge computing.

[0005] The first aspect of the present application provides a real-time traffic flow optimization method for vehicle-road collaborative edge computing, the method comprising:

[0006] Real-time data collection is performed on the on-board units and roadside equipment within the target road section to generate a multi-source heterogeneous traffic data set; the multi-source heterogeneous traffic data set is subjected to spatiotemporal fusion processing to extract a dynamic traffic flow parameter set; traffic flow optimization is performed based on the dynamic traffic flow parameter set to construct a local optimization instruction set; the local optimization instruction set is sent to the on-board units and roadside equipment within the target road section to achieve real-time distributed optimization of traffic flow.

[0007] The second aspect of the present application provides a real-time traffic flow optimization device for vehicle-road collaborative edge computing, the device comprising:

[0008] The acquisition module is used to collect data from the on-board units and roadside equipment in the target section in real time to generate a multi-source heterogeneous traffic data set; the processing module is used to perform spatiotemporal fusion processing on the multi-source heterogeneous traffic data set to extract a dynamic traffic flow parameter set; the instruction construction module is used to optimize the traffic flow based on the dynamic traffic flow parameter set and construct a local optimization instruction set; the optimization module is used to send the local optimization instruction set to the on-board units and roadside equipment in the target section to achieve real-time distributed optimization of the traffic flow.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] This application collects data from onboard units and roadside equipment within a target road section in real time to generate a multi-source heterogeneous traffic data set; performs spatiotemporal fusion processing on the multi-source heterogeneous traffic data set to extract a dynamic traffic flow parameter set; optimizes traffic flow based on the dynamic traffic flow parameter set and constructs a local optimization instruction set; and distributes the local optimization instruction set to the onboard units and roadside equipment within the target road section to achieve real-time distributed optimization of traffic flow. This invention solves the technical problem of the untimely response of traffic flow optimization in the prior art by fusing multi-source data from onboard units and roadside equipment and performing real-time processing and optimization instruction generation based on edge computing, thereby achieving the technical effect of improving the real-time performance and accuracy of traffic control. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0012] Figure 1 A flow chart of a real-time traffic flow optimization method for vehicle-road collaborative edge computing provided in an embodiment of the present application;

[0013] Figure 2 Schematic diagram of the structure of the real-time traffic flow optimization device for vehicle-road collaborative edge computing provided in an embodiment of the present application.

[0014] Description of the accompanying drawings: acquisition module 11, processing module 12, instruction construction module 13, optimization module 14. DETAILED DESCRIPTION

[0015] This application aims to solve the technical problem of untimely traffic flow optimization response in the existing technology by providing a real-time traffic flow optimization method and device based on vehicle-road collaborative edge computing. By integrating multi-source data from on-board units and roadside equipment and performing real-time processing and optimization instruction generation based on edge computing, the technical effect of improving the real-time and accuracy of traffic control is achieved.

[0016] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only some of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0017] It should be noted that any variations of the terms "include" and "have" are intended to cover non-exclusive inclusions. For example, a process, method, apparatus, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or are inherent to these processes, methods, products or devices.

[0018] Example 1, as Figure 1 As shown, the present application provides a real-time traffic flow optimization method for vehicle-road collaborative edge computing, the method comprising:

[0019] Step S100: Real-time data collection is performed on the on-board units and roadside equipment within the target road section to generate a multi-source heterogeneous traffic dataset.

[0020] In this embodiment of the application, real-time traffic flow perception is achieved by deploying onboard units and roadside equipment within the target road section. First, based on the traffic flow analysis results, edge computing nodes with processing capabilities are deployed at key locations on the road section. The onboard units then collect vehicle status data (such as position, speed, and acceleration), while the roadside equipment obtains traffic environment data (such as lane occupancy, signal status, and environmental conditions). Both are uploaded to the edge computing nodes for timestamp alignment and fusion processing, ultimately generating a multi-source heterogeneous traffic dataset.

[0021] Furthermore, in the method provided in the embodiment of the application, real-time data collection is performed on the on-board units and roadside equipment within the target road section to generate a multi-source heterogeneous traffic data set, which also includes:

[0022] The target road section is traversed to perform traffic flow analysis, and roadside edge computing nodes are deployed on the target road section according to the analysis results; vehicle status data is acquired through real-time collection by on-board units within the target road section; traffic environment data is acquired through real-time collection by roadside equipment within the target road section; the vehicle status data and the traffic environment data are transmitted to the roadside edge computing node for timestamp alignment to generate the multi-source heterogeneous traffic dataset.

[0023] In this embodiment, traffic flow analysis is first performed on the target road section. Fixed-time interval statistics are used to extract key indicators such as historical traffic volume, average speed, and congestion duration. The analysis results are used to identify high-frequency congestion points or areas of traffic fluctuation. Based on the identification results, roadside edge computing nodes are deployed at these key locations. These nodes are typically equipped with local processors and communication modules, capable of receiving, processing, and issuing commands. They can calculate and process real-time traffic data without relying on remote central servers.

[0024] Next, real-time data collection is performed on vehicles traveling normally on the target road section using an onboard unit (OBU). The OBU, equipped with a built-in GPS module and speed / acceleration sensor, uses a periodic sampling mechanism (e.g., once per second) to collect data such as the vehicle's current location, speed, acceleration, and deceleration. This process generates vehicle status data.

[0025] Roadside equipment, such as cameras, millimeter-wave radar, or lidar, deployed on both sides of the target road, operates synchronously to collect road environment information. Cameras use image recognition algorithms to identify lane occupancy, traffic signal status, and queue lengths, while radar equipment uses point cloud analysis algorithms to obtain the spatial position and size of surrounding vehicles and obstacles, thereby generating traffic environment data.

[0026] Finally, vehicle status data and traffic environment data are transmitted to a roadside edge computing node for timestamp alignment. By transmitting the vehicle status data to the edge computing node at the sampling frequency, vehicle position and speed information are extracted. Lane video streams and radar point clouds collected by roadside equipment are also analyzed to obtain vehicle trajectory parameters and obstacle information. These multiple data types are aligned and fused in time series at the edge computing node, ultimately generating a multi-source, heterogeneous traffic dataset.

[0027] Furthermore, in the method provided in the embodiment of the application, the vehicle status data and the traffic environment data are transmitted to the roadside edge computing node for timestamp alignment to generate the multi-source heterogeneous traffic dataset, further comprising:

[0028] The vehicle status data is transmitted to the roadside edge computing node for processing according to the sampling frequency to obtain vehicle position data and vehicle speed data; the traffic environment data is parsed to obtain lane video stream data and radar point cloud data; the lane video stream data and the radar point cloud data are transmitted to the roadside edge computing node for processing to obtain vehicle trajectory parameters and obstacle information; the vehicle position data, the vehicle speed data, the vehicle trajectory parameters, and the obstacle information are aligned in time series to construct the multi-source heterogeneous traffic dataset.

[0029] In this embodiment of the present application, the acquired vehicle status data is first sent to a roadside edge computing node via a 5G network at a fixed sampling frequency (e.g., 1 Hz). After receiving the data, the edge node processes the raw data using a data field parsing method to extract the vehicle position and speed data for each vehicle at the corresponding timestamp.

[0030] The collected traffic environment data is then initially analyzed using a sensor decoupling method. The video stream output by the camera is processed into lane video stream data, consisting of a sequence of pixel images in consecutive frames. The raw signals collected by the millimeter-wave radar or lidar are then analyzed to generate corresponding radar point cloud data, which represents the spatial distribution of objects on the road in the form of three-dimensional coordinate points.

[0031] The lane video stream and radar point cloud data generated above are then transmitted via wireless links to the same roadside edge computing node. Within the edge node, the lane video stream is processed using an inter-frame motion extraction method to extract the continuous spatial position of the vehicle and generate vehicle trajectory parameters reflecting the vehicle's movement path. Simultaneously, a spatial clustering algorithm (such as DBSCAN) is used to process the radar point cloud data, identifying and extracting the spatial features of static and dynamic objects in the environment, thereby obtaining corresponding obstacle information.

[0032] Finally, a timestamp matching method is used to align the vehicle location data, vehicle speed data, vehicle trajectory parameters, and obstacle information stored in the edge nodes. By matching and integrating various types of data according to a unified time series, a multi-source heterogeneous traffic dataset is ultimately constructed.

[0033] Step S200: performing spatiotemporal fusion processing on the multi-source heterogeneous traffic dataset to extract a dynamic traffic flow parameter set.

[0034] In the present embodiment, when performing spatiotemporal fusion processing on a multi-source heterogeneous traffic dataset, the target road section is first spatially discretized using a fixed-interval grid partitioning method based on the constructed multi-source heterogeneous traffic dataset. This involves dividing the road into a number of spatial grid cells of equal length. Each grid cell serves as an independent analysis area, used to count and extract traffic flow characteristics within the area. Within each grid cell, the number of vehicle position data occurrences per unit time is counted. A fixed-time counting method is used to calculate the number of vehicles entering the grid within that time period. The traffic density of the road section is then calculated based on the spatial length of the grid.

[0035] The vehicle speed data for each grid cell's corresponding time period is then processed using the arithmetic mean method to calculate the average speed for that grid cell. The sum of all vehicle speeds and then dividing by the number of vehicles reflects the overall speed level of vehicles in that spatial unit, expressed in kilometers per hour.

[0036] The system then uses vehicle trajectory parameters to analyze the spatial position changes of vehicles between consecutive frames at signal-controlled intersections or congested areas. Adjacent trajectory differential detection is used to determine whether vehicles are stationary or moving at low speeds. Combined with head-to-tail distance information, the system identifies queues and calculates the queue length in meters based on the spatial distribution of the preceding and following vehicles.

[0037] Finally, the traffic density, average speed and queue length in each grid unit in the same time period are sorted and summarized, and a dynamic traffic flow parameter set with spatiotemporal distribution characteristics is formed according to spatial order and time series.

[0038] Step S300: Optimizing traffic flow based on the dynamic traffic flow parameter set and constructing a local optimization instruction set.

[0039] In this embodiment of the present application, when optimizing traffic flow based on a dynamic traffic flow parameter set, a reinforcement learning method is first applied to the dynamic traffic flow parameter set to generate corresponding parameter learning results. Subsequently, based on the learning results, signal light control and vehicle speed control are analyzed separately to construct a signal light phase adjustment scheme and a vehicle speed guidance scheme. Traffic simulations are then performed on each of these schemes, yielding two simulation results. Finally, these two simulation results are combined and optimized to form a localized optimization instruction set encompassing both signal light control and vehicle speed guidance.

[0040] Furthermore, in the method provided in the embodiment of the application, traffic flow optimization is performed based on the dynamic traffic flow parameter set, and a local optimization instruction set is constructed, which also includes:

[0041] Reinforcement learning is performed based on a dynamic traffic flow parameter set to obtain parameter learning results; traffic lights in a target road section are analyzed according to the parameter learning results to construct a traffic light phase adjustment plan; simulation and deduction are performed on the target road section according to the traffic light phase adjustment plan to generate a first simulation result; vehicles in the target road section are analyzed according to the parameter learning results to construct a vehicle speed guidance plan; simulation and deduction are performed on the target road section according to the vehicle speed guidance plan to generate a second simulation result; the first simulation result and the second simulation result are combined and optimized to generate a local optimization instruction set.

[0042] In this embodiment, a traffic control strategy is first trained using a deep Q-network approach based on a dynamic traffic flow parameter set. Using traffic density, average speed, and queue length as input, the strategy outputs the corresponding control action for the current traffic state during each iteration. After multiple rounds of training, parameter learning results are obtained that reflect control behaviors under different traffic conditions.

[0043] Based on the parameter learning results, the traffic light control strategy for the target road section is then analyzed. Using a state-action mapping method, the recommended control actions from the learning results are mapped one-to-one to specific traffic light control items. This allows for the development of a traffic light phase adjustment plan, including the duration of each phase and the switching order of each phase.

[0044] To verify the effectiveness of the signal light phase adjustment scheme, a traffic simulation model was constructed. This model, based on a microscopic modeling approach, incorporates historically extracted dynamic traffic flow parameters, including vehicle arrival rates, speed distribution, and queue length trends. The model's behavior logic and road traffic conditions are initialized, creating a simulation environment that reflects real-world traffic conditions. The signal light phase adjustment scheme is then fed into the model for simulation, resulting in the output of the first simulation results, including average vehicle travel time, signal control efficiency, and changes in queue length.

[0045] The parameter learning results are then used to analyze the vehicle operating status within the target road section. Using a zoned speed control method, the road is divided into multiple continuous control sections. The traffic density and average speed data within each section are combined to assign appropriate guidance speeds to vehicles, generating a vehicle speed guidance plan. This vehicle speed guidance plan is then input into the previously constructed traffic simulation model, maintaining the same simulation parameters. A simulation run of the optimized plan is then performed to generate a second simulation result, which includes evaluation metrics such as vehicle traffic efficiency.

[0046] Finally, the first and second simulation results are combined for optimization. Traffic state analysis is performed on the first simulation result, and the first traffic state entropy value is calculated. Based on this, the first optimization instruction is generated. Traffic state analysis is then performed on the second simulation result, and the second traffic state entropy value is calculated. Based on this, the second optimization instruction is generated. A feasibility analysis is then performed on the first and second optimization instructions to determine the combined feasibility coefficient. Finally, the two sets of instructions are combined based on the combined feasibility coefficient to construct a local optimization instruction set.

[0047] Furthermore, in the method provided in the embodiment of the application, the first simulation result and the second simulation result are combined and optimized to generate a local optimization instruction set, and the method further includes:

[0048] Based on the first simulation result, a traffic state analysis is performed to obtain a first traffic state entropy value; based on the first traffic state entropy value, local traffic light optimization is performed to generate a first optimization instruction; based on the second simulation result, a traffic state analysis is performed to obtain a second traffic state entropy value; based on the second traffic state entropy value, local vehicle speed optimization is performed to generate a second optimization instruction; a combined feasibility analysis is performed on the first optimization instruction and the second optimization instruction to generate a combined feasibility coefficient; based on the combined feasibility coefficient, the first optimization instruction and the second optimization instruction are combined to construct the local optimization instruction set.

[0049] In this embodiment, based on the first simulation results, the vehicle speed data within the simulation period is first statistically analyzed using an equidistant speed interval division method. The speed values ​​are divided into several fixed intervals (e.g., 0–10 km / h, 10–20 km / h, etc.), and the number and proportion of vehicles in each interval are counted. Subsequently, the speed distribution is processed using an information entropy calculation method to calculate a first traffic state entropy value, which is used to measure the uniformity of traffic conditions under signal control conditions. A higher entropy value indicates greater speed fluctuations and more unstable traffic flow.

[0050] Based on the first traffic state entropy value, a fixed-step adjustment method is used to optimize the traffic light control parameters. When the entropy value exceeds a set threshold (e.g., 2.5), the green light phase of the traffic light on the fluctuating road section is extended by 5 seconds, and the red light phase is shortened by 5 seconds accordingly, while the total signal cycle remains unchanged. This generates a first optimization instruction, which includes the phase number, the duration before and after the adjustment, and the corresponding direction.

[0051] Based on the second simulation results, the same equally spaced speed interval division method was used to process the vehicle operating data under speed guidance control. Statistical analysis was performed on each speed interval to determine the structural characteristics of the speed distribution. Information entropy was calculated to obtain the second traffic state entropy value, which reflects the consistency of the traffic state under the speed guidance strategy.

[0052] Based on the entropy value of the second traffic state, a speed optimization plan is generated using the segment speed re-engineering method. The target road section is divided into multiple control sections of equal length (e.g., one every 100 meters). If the entropy value of a section exceeds a set threshold (e.g., 2.0), the target speed for that section is reduced by 5 km / h, indicating the risk of speed fluctuation. The generated second optimization instruction includes the section number and the corresponding target speed setting.

[0053] To assess whether there is a control conflict between the first and second optimized instructions, a combination feasibility analysis is performed. Using a control overlap detection method, the control areas involved in the two instruction sets are compared to identify any overlaps in spatial location and action time. The overlap time or segment length of the two control instruction sets in the same location segment is calculated, and the combination feasibility coefficient is calculated accordingly. For example, if the overlapping area length is 20 meters and the total control area length is 60 meters, the combination feasibility coefficient is 1-(20 / 60) = 0.667. A higher coefficient indicates better compatibility between the two instruction sets.

[0054] Finally, based on the combined feasibility coefficient, the two sets of optimization instructions are combined using a priority coverage method to construct a localized optimized instruction set. When the combined feasibility coefficient is higher than 0.7, indicating minimal control conflict, the two sets of instructions are directly merged. When the coefficient is between 0.4 and 0.7, conflicting sections are compared. If the entropy of the first traffic state in a section is lower than the entropy of the second traffic state, the first optimized instruction is retained; otherwise, the second optimized instruction is retained. If the combined feasibility coefficient is lower than 0.4, indicating significant control conflict, only the instructions corresponding to the lower entropy value are retained. Through this method, inconsistent control items are eliminated and integrated to generate a localized optimized instruction set.

[0055] Step S400: Send the local optimization instruction set to the on-board units and roadside equipment in the target road section to achieve real-time distributed optimization of traffic flow.

[0056] In this embodiment, a local optimization instruction set is first sent to the edge computing node of the target road section. The instruction set is then broken down into speed optimization instructions and signal light optimization instructions. The speed optimization instructions are then sent to the onboard units within the road section. The onboard units perform local speed analysis based on the received instructions, generate recommended speed information based on the current vehicle operating status, and display it through a visual interface on the onboard terminal to guide the driver in adjusting speed.

[0057] Simultaneously, traffic light optimization instructions are issued to traffic signal control equipment within the target road section. Upon receiving the instructions, the roadside equipment performs phase analysis and outputs the signal light phase cycle information. Based on this phase cycle information, predictions are made for multiple adjacent signal light nodes, generating prediction data for adjacent signal light nodes. Based on this information, the overall traffic status of the road section is analyzed and the section congestion index is calculated. If an upward trend in the congestion index is detected, the signal light phase cycle information is optimized and updated to improve signal control responsiveness. The coordinated execution of these two sub-processes enables distributed collaborative control between the onboard unit and the roadside equipment, achieving real-time dynamic optimization of traffic flow.

[0058] Furthermore, in the method provided in the embodiment of the application, after the local optimization instruction set is sent to the on-board unit in the target road section, the method further includes:

[0059] Decomposition is performed based on the local optimization instruction set to obtain vehicle speed optimization instructions and traffic light optimization instructions; the vehicle speed optimization instructions are sent to the on-board unit, vehicle speed analysis is performed based on the on-board unit, and recommended vehicle speed information is generated; a visualization interface is triggered based on the recommended vehicle speed information, and the information is displayed through the visualization interface.

[0060] In an embodiment of the present application, the field recognition method is first used to decompose the local optimization instruction set. By identifying the type field in the instruction, the part used for vehicle speed control is extracted as a vehicle speed optimization instruction, and the part used for traffic light control is extracted as a traffic light optimization instruction.

[0061] Next, C-V2X communication technology is used to transmit speed optimization instructions to the onboard unit (OBU) within the target road section. Upon receiving the speed optimization instruction, the OBU reads its current speed and position and uses a difference comparison method to compare the current speed with the target speed included in the speed optimization instruction. If the current speed deviates significantly from the target speed, an adjusted recommended speed is calculated. This recommended speed value is used to assist the vehicle in achieving smoother and more efficient driving within the subsequent control section.

[0062] Finally, the onboard unit displays the recommended speed information in text and graphics on a visual interface within the vehicle, such as the center console or instrument panel. Using a speed text prompt, the recommended speed is displayed directly as a numerical value (e.g., a recommended speed of 45 km / h), and can be combined with color to indicate the speed. For example, green indicates a reasonable speed, while yellow or red indicates a need for adjustment. This display continuously updates, guiding the driver to act accordingly, thus completing the localized closed-loop execution of speed guidance control.

[0063] Furthermore, in the method provided in the embodiment of the application, after the local optimization instruction set is sent to the roadside equipment in the target road section, the method further includes:

[0064] The traffic light optimization instruction is sent to the roadside equipment, and phase analysis is performed based on the roadside equipment to generate traffic light phase cycle information; adjacent multi-node prediction is performed based on the traffic light phase cycle information to generate adjacent traffic light node prediction data; congestion analysis is performed on the target road section according to the adjacent traffic light node prediction data to generate a road section congestion index; when it is detected that the road section congestion index has increased, the traffic light phase cycle information is optimized and updated.

[0065] In this embodiment, a traffic light optimization command is first issued to a roadside device, namely, a traffic signal control device. Upon receiving the command, the device reads the command fields to extract control parameters such as the red light duration, green light duration, yellow light duration, and phase switching sequence. By analyzing these parameters item by item, complete traffic light phase cycle information is generated.

[0066] Next, based on the signal light phase cycle information, a time-delayed shifting method is used to predict the states of adjacent intersections upstream and downstream of the current intersection. Based on the start and end times of each signal light phase and the average vehicle travel time between adjacent sections, the signal cycle of the current intersection is shifted forward or backward. The future signal states of each adjacent intersection are deduced over time, generating prediction data for adjacent signal light nodes.

[0067] The traffic status of the entire target road section is then analyzed based on the predicted data from adjacent traffic light nodes. The number of vehicles entering the road section per unit time is calculated as the arrival rate, while the number of vehicles leaving the road section is calculated as the traffic rate. The difference between the two represents the change in the number of queued vehicles within the current time period. The differences over multiple time periods are accumulated to determine the total number of queued vehicles. This is then compared to the preset maximum number of queued vehicles to determine the road section congestion index.

[0068] Finally, if the congestion index on a road section is detected to be continuously rising and exceeds a threshold pre-set by technical experts, the current signal light phase cycle information is automatically adjusted. This adjustment is performed by extending the green light duration in the direction of congestion while correspondingly shortening the red light duration in other directions. The adjustment is generally in increments of five seconds, ensuring that the total signal cycle length remains unchanged. The updated phase cycle information directly overwrites the original setting and is immediately executed by the traffic signal control equipment, dynamically alleviating vehicle congestion and improving traffic efficiency. This process completes the optimization and update of the signal light phase cycle information.

[0069] In the embodiments of the present application, in summary, the embodiments of the present application have at least the following technical effects:

[0070] This application collects data from onboard units and roadside equipment within a target road section in real time to generate a multi-source heterogeneous traffic data set; performs spatiotemporal fusion processing on the multi-source heterogeneous traffic data set to extract a dynamic traffic flow parameter set; optimizes traffic flow based on the dynamic traffic flow parameter set and constructs a local optimization instruction set; and distributes the local optimization instruction set to the onboard units and roadside equipment within the target road section to achieve real-time distributed optimization of traffic flow. This invention solves the technical problem of the untimely response of traffic flow optimization in the prior art by fusing multi-source data from onboard units and roadside equipment and performing real-time processing and optimization instruction generation based on edge computing, thereby achieving the technical effect of improving the real-time performance and accuracy of traffic control.

[0071] Example 2, based on the same inventive concept as the method for real-time optimization of traffic flow by vehicle-road collaborative edge computing in the previous embodiment, Figure 2 As shown, this application provides a real-time traffic flow optimization device for vehicle-road collaborative edge computing. The device and method embodiments in the embodiments of this application are based on the same inventive concept. Among them, the device includes:

[0072] The acquisition module 11 is used to collect data from the on-board units and roadside equipment in the target section in real time to generate a multi-source heterogeneous traffic data set; the processing module 12 is used to perform spatiotemporal fusion processing on the multi-source heterogeneous traffic data set to extract a dynamic traffic flow parameter set; the instruction construction module 13 is used to optimize the traffic flow based on the dynamic traffic flow parameter set and construct a local optimization instruction set; the optimization module 14 is used to send the local optimization instruction set to the on-board units and roadside equipment in the target section to realize real-time distributed optimization of the traffic flow.

[0073] Furthermore, the device is also used to implement the following functions:

[0074] The target road section is traversed to perform traffic flow analysis, and roadside edge computing nodes are deployed on the target road section according to the analysis results; vehicle status data is acquired through real-time collection by on-board units within the target road section; traffic environment data is acquired through real-time collection by roadside equipment within the target road section; the vehicle status data and the traffic environment data are transmitted to the roadside edge computing node for timestamp alignment to generate the multi-source heterogeneous traffic dataset.

[0075] Furthermore, the device is also used to implement the following functions:

[0076] The vehicle status data is transmitted to the roadside edge computing node for processing according to the sampling frequency to obtain vehicle position data and vehicle speed data; the traffic environment data is parsed to obtain lane video stream data and radar point cloud data; the lane video stream data and the radar point cloud data are transmitted to the roadside edge computing node for processing to obtain vehicle trajectory parameters and obstacle information; the vehicle position data, the vehicle speed data, the vehicle trajectory parameters, and the obstacle information are aligned in time series to construct the multi-source heterogeneous traffic dataset.

[0077] Furthermore, the device is also used to implement the following functions:

[0078] Reinforcement learning is performed based on a dynamic traffic flow parameter set to obtain parameter learning results; traffic lights in a target road section are analyzed according to the parameter learning results to construct a traffic light phase adjustment plan; simulation and deduction are performed on the target road section according to the traffic light phase adjustment plan to generate a first simulation result; vehicles in the target road section are analyzed according to the parameter learning results to construct a vehicle speed guidance plan; simulation and deduction are performed on the target road section according to the vehicle speed guidance plan to generate a second simulation result; the first simulation result and the second simulation result are combined and optimized to generate a local optimization instruction set.

[0079] Furthermore, the device is also used to implement the following functions:

[0080] Based on the first simulation result, a traffic state analysis is performed to obtain a first traffic state entropy value; based on the first traffic state entropy value, local traffic light optimization is performed to generate a first optimization instruction; based on the second simulation result, a traffic state analysis is performed to obtain a second traffic state entropy value; based on the second traffic state entropy value, local vehicle speed optimization is performed to generate a second optimization instruction; a combined feasibility analysis is performed on the first optimization instruction and the second optimization instruction to generate a combined feasibility coefficient; based on the combined feasibility coefficient, the first optimization instruction and the second optimization instruction are combined to construct the local optimization instruction set.

[0081] Furthermore, the device is also used to implement the following functions:

[0082] Decomposition is performed based on the local optimization instruction set to obtain vehicle speed optimization instructions and traffic light optimization instructions; the vehicle speed optimization instructions are sent to the on-board unit, vehicle speed analysis is performed based on the on-board unit, and recommended vehicle speed information is generated; a visualization interface is triggered based on the recommended vehicle speed information, and the information is displayed through the visualization interface.

[0083] Furthermore, the device is also used to implement the following functions:

[0084] The traffic light optimization instruction is sent to the roadside equipment, and phase analysis is performed based on the roadside equipment to generate traffic light phase cycle information; adjacent multi-node prediction is performed based on the traffic light phase cycle information to generate adjacent traffic light node prediction data; congestion analysis is performed on the target road section according to the adjacent traffic light node prediction data to generate a road section congestion index; when it is detected that the road section congestion index has increased, the traffic light phase cycle information is optimized and updated.

[0085] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0086] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

[0087] This specification and drawings are merely illustrative of the present application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, this application is intended to include such modifications and variations as fall within the scope of this application and its equivalents.

Claims

1. A real-time traffic flow optimization method based on vehicle-road collaborative edge computing, characterized by: The method comprises: Carry out real-time data collection of on-board units and roadside equipment within the target road section to generate multi-source heterogeneous traffic data sets; Performing spatiotemporal fusion processing on the multi-source heterogeneous traffic data set to extract a dynamic traffic flow parameter set; Performing traffic flow optimization based on the dynamic traffic flow parameter set and constructing a local optimization instruction set; The local optimization instruction set is sent to the on-board units and roadside equipment within the target road section to achieve real-time distributed optimization of traffic flow.

2. The method for real-time traffic flow optimization based on vehicle-road collaborative edge computing according to claim 1, characterized in that: Real-time data collection of on-board units and roadside equipment within the target road section is performed to generate a multi-source heterogeneous traffic dataset. The method includes: Traverse the target road section to perform traffic flow analysis, and deploy roadside edge computing nodes on the target road section based on the analysis results; Real-time data collection is performed through the on-board unit within the target road section to obtain vehicle status data; Obtain traffic environment data through real-time collection of data via roadside equipment within the target road section; The vehicle status data and the traffic environment data are transmitted to a roadside edge computing node for timestamp alignment to generate the multi-source heterogeneous traffic dataset.

3. The method for real-time traffic flow optimization based on vehicle-road collaborative edge computing according to claim 2, characterized in that: The vehicle status data and the traffic environment data are transmitted to a roadside edge computing node for timestamp alignment to generate the multi-source heterogeneous traffic dataset, the method comprising: Transmitting the vehicle status data to the roadside edge computing node for processing according to the sampling frequency to obtain vehicle position data and vehicle speed data; Analyze the traffic environment data to obtain lane video stream data and radar point cloud data; Transmitting the lane video stream data and the radar point cloud data to a roadside edge computing node for processing to obtain vehicle trajectory parameters and obstacle information; The vehicle position data, the vehicle speed data, the vehicle trajectory parameters, and the obstacle information are aligned in time series to construct the multi-source heterogeneous traffic dataset.

4. The method for real-time traffic flow optimization based on vehicle-road collaborative edge computing according to claim 3, characterized in that: Traffic flow optimization is performed based on the dynamic traffic flow parameter set, and a local optimization instruction set is constructed. The method includes: Perform reinforcement learning based on the dynamic traffic flow parameter set to obtain parameter learning results; Analyze the traffic lights in the target road section according to the parameter learning results and build a traffic light phase adjustment plan; Performing simulation on the target road section according to the signal light phase adjustment scheme to generate a first simulation result; Analyze the vehicles in the target road section according to the parameter learning results and build a vehicle speed guidance plan; Performing simulation on the target road section according to the vehicle speed guidance scheme to generate a second simulation result; The first simulation result and the second simulation result are combined and optimized to generate a local optimization instruction set.

5. The method for real-time traffic flow optimization based on vehicle-road collaborative edge computing according to claim 4, characterized in that: Combining and optimizing the first simulation result and the second simulation result to generate a local optimization instruction set, the method comprising: Performing traffic state analysis based on the first simulation result to obtain a first traffic state entropy value; Performing local traffic light optimization according to the first traffic state entropy value to generate a first optimization instruction; Performing traffic state analysis based on the second simulation result to obtain a second traffic state entropy value; Performing local vehicle speed optimization according to the second traffic state entropy value to generate a second optimization instruction; Performing a combined feasibility analysis on the first optimization instruction and the second optimization instruction to generate a combined feasibility coefficient; The first optimization instruction and the second optimization instruction are combined based on the combined feasible coefficient to construct the local optimization instruction set.

6. The method for real-time traffic flow optimization based on vehicle-road collaborative edge computing according to claim 1, characterized in that: After the local optimization instruction set is sent to the on-board unit within the target road section, the method includes: Decomposing the local optimization instruction set to obtain a vehicle speed optimization instruction and a traffic light optimization instruction; Sending the vehicle speed optimization instruction to the onboard unit, performing vehicle speed analysis based on the onboard unit, and generating recommended vehicle speed information; A visualization interface is triggered based on the recommended vehicle speed information, and the information is displayed through the visualization interface.

7. The method for real-time traffic flow optimization based on vehicle-road collaborative edge computing according to claim 6, characterized in that: After the local optimization instruction set is sent to the roadside equipment within the target road section, the method includes: Sending the signal light optimization instruction to the roadside equipment, performing phase analysis based on the roadside equipment, and generating signal light phase cycle information; Performing adjacent multi-node prediction based on the signal light phase cycle information to generate adjacent signal light node prediction data; Performing congestion analysis on the target road section according to the adjacent traffic light node prediction data to generate a road section congestion index; When it is detected that the road section congestion index increases, the signal light phase cycle information is optimized and updated.

8. A real-time traffic flow optimization device based on vehicle-road collaborative edge computing, characterized in that: The device is used to execute the real-time traffic flow optimization method of vehicle-road cooperative edge computing as described in any one of claims 1 to 7, and the device includes: The acquisition module is used to collect data from on-board units and roadside equipment in the target road section in real time to generate a multi-source heterogeneous traffic data set; a processing module, configured to perform spatiotemporal fusion processing on the multi-source heterogeneous traffic data set and extract a dynamic traffic flow parameter set; An instruction construction module, configured to perform traffic flow optimization based on the dynamic traffic flow parameter set and construct a local optimization instruction set; The optimization module is used to send the local optimization instruction set to the on-board units and roadside equipment in the target road section to achieve real-time distributed optimization of traffic flow.

Images