Multi-source high-precision sensing intelligent road network cooperative control method and platform

Abstract

This invention discloses a method and platform for intelligent road network cooperative control based on multi-source high-precision sensing, relating to the field of intelligent transportation. The method includes: acquiring multi-source traffic sensing data of a target road network; performing spatiotemporal alignment and denoising fusion on the multi-source traffic sensing data to obtain a traffic sensing fusion dataset; constructing a road network traffic situation model based on the traffic sensing fusion dataset; performing target detection and trajectory tracking on the road network traffic situation model to obtain multiple traffic anomaly event feature vectors; and generating a road network cooperative control strategy based on the multiple traffic anomaly event feature vectors. This application solves the technical problem of existing road network management modes being unable to achieve unified perception and dynamic linkage across the entire road network, achieving the technical effect of dynamic linkage and precise response in road network traffic management through high-precision perception of the entire road network data.

Description

Technical Field

[0001] This application relates to the field of intelligent transportation, and in particular to a method and platform for collaborative control of intelligent road networks based on multi-source high-precision sensing. Background Technology

[0002] The coordinated control of intelligent road networks is a core means to ensure efficient urban traffic operation, improve road network throughput, and reduce traffic congestion and energy consumption. Its technological implementation directly impacts the precision of urban traffic management and the travel experience of citizens, making it a crucial component of modern intelligent transportation systems. Currently, the industry primarily relies on manual scheduling and infrastructure linkage based on single data sources for road network traffic control. This involves fixed timing schemes for individual roadside traffic signal equipment, feedback from manual on-site inspections, and a limited number of roadside cameras for manual video monitoring, achieving passive response and basic control of road network traffic conditions. These methods only cover basic traffic condition monitoring in localized areas, failing to achieve comprehensive perception of the overall road network traffic condition. Furthermore, the reliance on manual intervention results in slow response times, making it difficult to adapt to dynamically changing road network traffic flow. It also cannot predict the evolution of traffic conditions, easily leading to control lags and mismatches between control measures and actual traffic demands.

[0003] Currently, the road network management and control model has the technical problem of failing to achieve unified perception and dynamic linkage across the entire road network. Summary of the Invention

[0004] This application provides a multi-source high-precision sensing intelligent road network collaborative control method and platform. By acquiring multi-source traffic sensing data from vehicle-mounted sensors, roadside cameras, millimeter-wave radar, lidar, and traffic signal equipment of the target road network, the method performs spatiotemporal alignment and denoising fusion on the above data to obtain a traffic sensing fusion dataset. Based on this dataset, a road network traffic situation model is constructed. By performing target detection and trajectory tracking on the model, multiple traffic anomaly event feature vectors are obtained. Based on these feature vectors, a road network collaborative control strategy is generated. This application solves the technical problem of existing road network management modes being unable to achieve unified perception and dynamic linkage across the entire road network. It achieves the technical effect of dynamic linkage and precise response in road network traffic management through high-precision perception of the entire road network data.

[0005] This application provides a method for intelligent road network cooperative control based on multi-source high-precision perception, comprising: acquiring multi-source traffic perception data of a target road network; performing spatiotemporal alignment and denoising fusion on the multi-source traffic perception data to obtain a traffic perception fusion dataset; constructing a road network traffic situation model based on the traffic perception fusion dataset; performing target detection and trajectory tracking on the road network traffic situation model to obtain multiple traffic anomaly event feature vectors; and generating a road network cooperative control strategy based on the multiple traffic anomaly event feature vectors.

[0006] In a possible implementation, the multi-source traffic sensing data is spatiotemporally aligned and denoised to obtain a traffic sensing fusion dataset. The following processing is then performed: time synchronization of the multi-source traffic sensing data is performed according to a unified time reference to generate a multi-source time-aligned data sequence; spatial coordinate mapping and matching of the multi-source time-aligned data sequence are performed according to a road network spatial coordinate system to generate multi-source spatially aligned data; anomaly detection and noise filtering are performed on the multi-source spatially aligned data to obtain multi-source denoised traffic sensing data; and sensing fusion processing is performed on the multi-source denoised traffic sensing data to obtain the traffic sensing fusion dataset.

[0007] In a possible implementation, a road network traffic situation model is constructed based on the traffic perception fusion dataset, and the following processes are performed: traffic state feature parameters of each road segment are extracted based on the traffic perception fusion dataset to obtain multiple road segment traffic state feature vectors; a road segment traffic state matrix is ​​constructed based on the multiple road segment traffic state feature vectors; a topological association mapping is performed on the road segment traffic state matrix according to the target road network topology to establish road network traffic association relationships; and traffic state propagation modeling is performed on the road segment traffic state matrix based on the road network traffic association relationships to generate the road network traffic situation model.

[0008] In a possible implementation, the road network traffic situation model is subjected to target detection and trajectory tracking to obtain multiple traffic anomaly event feature vectors, and the following processing is performed: target detection is performed on the road network traffic situation model to obtain traffic target detection results; trajectory tracking is performed on the road network traffic situation model based on the traffic target detection results to obtain multiple target trajectory feature vectors; anomaly identification is performed on the multiple target trajectory feature vectors to generate the multiple traffic anomaly event feature vectors.

[0009] In a possible implementation, a road network cooperative control strategy is generated based on the feature vectors of the multiple traffic anomaly events, and the following processes are performed: event type identification and event impact range assessment are performed on the feature vectors of the multiple traffic anomaly events to obtain each traffic anomaly level and each traffic anomaly propagation impact area; based on each traffic anomaly level and each traffic anomaly propagation impact area, a road network cooperative control trigger node is determined; a control strategy candidate set is constructed based on the road network cooperative control trigger node; and the control strategy candidate set is subjected to cooperative matching and conflict resolution to obtain the road network cooperative control strategy.

[0010] In a possible implementation, the following processing is performed: the multi-source traffic sensing data includes vehicle-mounted sensing data, roadside camera data, millimeter-wave radar data, lidar data, and traffic signal equipment data.

[0011] In a possible implementation, the following processing is performed: the control strategy candidate set includes signal timing adjustment strategy, lane passage guidance strategy, route diversion guidance strategy, and roadside prompt release strategy.

[0012] This application also provides a multi-source high-precision sensing intelligent road network cooperative control platform, including: a multi-source traffic sensing data acquisition module for acquiring multi-source traffic sensing data of a target road network; a data fusion module for performing spatiotemporal alignment and denoising fusion on the multi-source traffic sensing data to obtain a traffic sensing fusion dataset; a road network traffic situation model construction module for constructing a road network traffic situation model based on the traffic sensing fusion dataset; a traffic anomaly detection module for performing target detection and trajectory tracking on the road network traffic situation model to obtain multiple traffic anomaly event feature vectors; and a road network cooperative control strategy generation module for generating a road network cooperative control strategy based on the multiple traffic anomaly event feature vectors.

[0013] The proposed method and platform for intelligent road network cooperative control, based on multi-source high-precision sensing, first acquires multi-source traffic sensing data of the target road network. Then, it performs spatiotemporal alignment and denoising fusion on the multi-source traffic sensing data to obtain a traffic sensing fusion dataset. Next, based on this dataset, a road network traffic situation model is constructed. Then, target detection and trajectory tracking are performed on the model to obtain multiple traffic anomaly event feature vectors. Finally, a road network cooperative control strategy is generated based on these feature vectors. Through this process, the proposed method and platform achieve the technical effect of dynamic linkage and precise response in road network traffic management by using high-precision sensing of the entire road network data. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0015] Figure 1 This is a flowchart illustrating the intelligent road network collaborative control method based on multi-source high-precision sensing provided in an embodiment of this application.

[0016] Figure 2 This is a schematic diagram of the structure of the intelligent road network collaborative control platform with multi-source high-precision sensing provided in the embodiments of this application.

[0017] Figure labeling: 10 Multi-source traffic perception data acquisition module, 20 Data fusion module, 30 Road network traffic situation model construction module, 40 Traffic anomaly detection module, 50 Road network collaborative control strategy generation module. Detailed Implementation

[0018] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0019] This application provides a method for intelligent road network cooperative control using multi-source high-precision sensing, such as... Figure 1 As shown, the method includes:

[0020] Step S100: Obtain multi-source traffic perception data of the target road network, including vehicle-mounted perception data, roadside camera data, millimeter-wave radar data, lidar data, and traffic signal equipment data.

[0021] Specifically, vehicle-mounted sensing terminals, roadside high-definition cameras, roadside millimeter-wave radar equipment, roadside lidar equipment, and traffic signal controllers deployed within the target road network are used as data acquisition tools. The system takes a preset acquisition cycle from the target road network as input and reads real-time data from the corresponding devices. Vehicle-mounted sensing data includes vehicle location, speed, direction of travel, and vehicle type information; roadside camera data includes road segment video streams and intersection image frames; millimeter-wave radar data includes target distance, relative speed, and reflection intensity data; lidar data includes 3D point cloud coordinates and target contour information; and traffic signal equipment data includes traffic light phase, traffic light status, remaining traffic light duration, and signal timing scheme data. The output is a raw multi-source traffic sensing dataset containing all of the above data types.

[0022] Step S200: Perform spatiotemporal alignment and denoising fusion on the multi-source traffic perception data to obtain a traffic perception fusion dataset.

[0023] Specifically, taking the original multi-source traffic perception dataset output in step S100 as input, the time reference of the data from different sources is first unified, then the spatial coordinates are unified, then abnormal and noisy data are removed, and finally the effective data are fused to form a traffic perception fusion dataset for modeling.

[0024] In one possible implementation, the multi-source traffic sensing data is spatiotemporally aligned and denoised to obtain a traffic sensing fusion dataset. Step S200 further includes step S210, which involves time synchronization processing of the multi-source traffic sensing data based on a unified time reference to generate a multi-source time-aligned data sequence. Specifically, a Global Positioning System (GPS) clock is used as the unified time reference. The original multi-source traffic sensing data output in step S100 is used as input, and the vehicle-mounted sensing data, roadside camera data, millimeter-wave radar data, lidar data, and traffic signal equipment data are timestamped to align the acquisition times of different devices to the same time axis. For example, camera data with an acquisition period of 100 milliseconds, millimeter-wave radar data with an acquisition period of 50 milliseconds, and lidar data with an acquisition period of 20 milliseconds are all calibrated to whole 100-millisecond time nodes, outputting a multi-source time-aligned data sequence with a consistent time dimension.

[0025] Step S220 involves performing spatial coordinate mapping and spatial matching processing on the multi-source time-aligned data sequence according to the road network spatial coordinate system to generate multi-source spatially aligned data. Specifically, the Gaussian Cartesian coordinate system corresponding to the target road network is used as the unified spatial coordinate system. Taking the multi-source time-aligned data sequence output in step S210 as input, the latitude and longitude coordinates in the vehicle perception data, the relative detection coordinates of millimeter-wave radar and lidar, and the pixel coordinates of camera images are all converted into absolute coordinates in the road network spatial coordinate system. For example, the vehicle pixel positions in the camera images are mapped to the actual mileage coordinates of the corresponding road segments, and the lidar point cloud coordinates are mapped to the horizontal and vertical coordinates in the road network plane, completing the matching of data from different sources in the same space, and outputting multi-source spatially aligned data.

[0026] Step S230 involves performing anomaly detection and noise filtering on the multi-source spatially aligned data to obtain multi-source denoised traffic perception data. Specifically, a threshold judgment method and a sliding window filtering method are used as processing algorithms. The multi-source spatially aligned data output in step S220 is used as input to perform anomaly detection on the speed, position, and signal status fields. For example, speed values ​​exceeding the normal driving range, position coordinates exceeding the road network range, and signal light states with logical inconsistencies in signal cycle are identified as anomaly data. The mean of neighboring data is used to replace noise points, and duplicate and erroneous data are eliminated, resulting in effective and accurate multi-source denoised traffic perception data.

[0027] Step S240: Perform perception fusion processing on the multi-source denoised traffic perception data to obtain the traffic perception fusion dataset. Specifically, a weighted fusion method is used as the data fusion approach. Taking the multi-source denoised traffic perception data output in step S230 as input, different sources of perception data for the same target, location, and time are assigned corresponding weights. For example, LiDAR data has a higher weight than camera data, and millimeter-wave radar velocity data has a higher weight than vehicle-mounted perception data. Through weighted calculation, unique and stable target information and road condition information are obtained, and the output is a traffic perception fusion dataset containing complete traffic target and road condition information.

[0028] Step S300: Construct a road network traffic situation model based on the traffic perception fusion dataset.

[0029] Specifically, taking the traffic perception fusion dataset output in step S240 as input, the state features of each road segment are first extracted, then a state matrix is ​​constructed, and the correlation is established in combination with the road network topology to model the propagation of traffic state within the road network, and a road network traffic situation model that can reflect the overall operating state of the road network is output.

[0030] In one possible implementation, a road network traffic situation model is constructed based on the traffic perception fusion dataset. Step S300 further includes step S310, which involves extracting traffic state feature parameters for each road segment based on the traffic perception fusion dataset to obtain multiple road segment traffic state feature vectors. Specifically, a feature statistics method is used as the extraction method. Taking the traffic perception fusion dataset output in step S240 as input, and according to the preset road network segment division results, average vehicle speed, traffic flow, vehicle density, queue length, and occupancy rate feature parameters are extracted for each road segment. These parameters are combined to form the traffic state feature vector of the corresponding road segment, and the output is multiple road segment traffic state feature vectors.

[0031] Step S320: Construct a road segment traffic state matrix based on the multiple road segment traffic state feature vectors. Specifically, a matrix construction method is used, taking the multiple road segment traffic state feature vectors output in step S310 as input, and arranging the feature vector of each road segment as a row of the matrix, sequentially according to the road segment number, to form a road segment traffic state matrix with road segments as rows and traffic state feature parameters as columns. For example, a road network containing 20 road segments and 5 feature parameters will generate a 20-row, 5-column road segment traffic state matrix.

[0032] Step S330: Based on the target road network topology, perform topological association mapping on the road segment traffic state matrix to establish road network traffic association relationships. Specifically, based on a preset target road network topology file and using the road segment traffic state matrix output in step S320 as input, match the interconnected road segments according to their connection relationships, intersection turning relationships, and upstream / downstream relationships. For example, establish association markers between upstream straight lanes and downstream straight lanes, and between left-turn lanes and their corresponding merging road segments to form traffic influence relationships between road segments. The output is a road network traffic association relationship containing road segment association markers.

[0033] Step S340: Based on the road network traffic correlation, traffic state propagation modeling is performed on the road segment traffic state matrix to generate the road network traffic situation model. Specifically, traffic wave theory combined with segmented recursive calculation is used as the modeling method. Taking the road network traffic correlation and road segment traffic state matrix obtained in step S330 as input, the calculation method for traffic wave propagation speed is first determined: taking the average vehicle speed, vehicle density, and road capacity of the road segment as input, the traffic wave propagation speed of each road segment is calculated according to the formula: Traffic wave speed = (free-flow vehicle speed - current average vehicle speed) × current vehicle density ÷ (road saturation density - current vehicle density). Based on the upstream and downstream road segment connection markers in the road network traffic correlation, taking the traffic state of the upstream road segment and the calculated traffic wave propagation speed as input, the traffic state change is recursively calculated towards the downstream road segment according to the time step. The traffic state changes of all road segments at different time steps are integrated to form a road network traffic situation model that can dynamically reflect the overall traffic operation status of the road network.

[0034] Step S400: Target detection and trajectory tracking are performed on the road network traffic situation model to obtain feature vectors of multiple traffic anomaly events.

[0035] Specifically, taking the road network traffic situation model output in step S340 as input, traffic targets are first identified from the road network situation, then the target's trajectory is tracked, and the presence of anomalies is determined based on the trajectory characteristics, and multiple traffic anomaly event feature vectors are output to characterize traffic anomalies.

[0036] In one possible implementation, target detection and trajectory tracking are performed on the road network traffic situation model to obtain multiple traffic anomaly event feature vectors. Step S400 further includes step S410, which involves performing target detection on the road network traffic situation model to obtain traffic target detection results. Specifically, target contour matching and location clustering algorithms are used as the detection method. Taking the road network traffic situation model output in step S340 as input, the target position, size, and motion information within the model are identified, distinguishing between motor vehicles, non-motor vehicles, and pedestrian traffic target types. The number, position, size, and type information of each target are determined, and the output is a traffic target detection result containing basic information of all traffic targets.

[0037] Step S420: Based on the traffic target detection results, trajectory tracking is performed on the road network traffic situation model to obtain multiple target trajectory feature vectors. Specifically, a Kalman filter tracking algorithm is used as the tracking method. Taking the traffic target detection results output in step S410 as input, position association and trajectory continuation are performed on traffic targets with the same number within consecutive time periods. The position, velocity, acceleration, and direction of travel of the target at different time periods are recorded to form a continuous motion trajectory. The trajectory-related parameters are combined into target trajectory feature vectors, and multiple target trajectory feature vectors are output.

[0038] Step S430 involves anomaly identification of the multiple target trajectory feature vectors to generate multiple traffic anomaly event feature vectors. Specifically, a trajectory feature threshold judgment method is used as the identification method. The multiple target trajectory feature vectors output in step S420 are used as input, and the actual target trajectory is compared with the normal driving trajectory. For example, sudden deceleration, emergency stopping, illegal lane changing, driving in the wrong direction, and prolonged lingering are identified as anomaly events. Anomaly type, location, time of occurrence, and impact length feature parameters are extracted to generate corresponding feature vectors for the anomaly events, and the output is multiple traffic anomaly event feature vectors.

[0039] Step S500: Generate a road network cooperative control strategy based on the feature vectors of the multiple traffic anomaly events.

[0040] Specifically, taking the feature vectors of multiple traffic anomalies output in step S430 as input, the anomaly level and impact range are first assessed, then the control trigger location and timing are determined, a set of candidate strategies is constructed, and a road network collaborative control strategy suitable for the current road network is output through collaborative optimization.

[0041] In one possible implementation, a road network collaborative control strategy is generated based on the feature vectors of the multiple traffic anomalies. Step S500 further includes step S510, which involves identifying the event type and assessing the impact range of the multiple traffic anomaly feature vectors to obtain the traffic anomaly level and the propagation impact area of ​​each traffic anomaly. Specifically, a pre-defined rule base combined with grid diffusion calculation is used as the evaluation method. Taking the feature vectors of multiple traffic anomalies output in step S430 as input, the following steps are performed: First, event type identification is performed: the anomaly type field in the anomaly event feature vector is matched with type entries in the pre-defined rule base, such as vehicle breakdown, traffic accident, illegal parking, and road congestion, to determine the specific event type. Second, anomaly level classification is performed: based on the level judgment criteria corresponding to different event types in the pre-defined rule base, the anomaly level is calculated in combination with the event duration and the number of affected vehicles. For example, vehicle breakdowns with an event duration of less than 10 minutes and affecting less than 10 vehicles are judged as Level 1 anomalies, and traffic accidents with an event duration of more than 30 minutes and affecting more than 50 vehicles are judged as Level 3 anomalies. Finally, the influence range diffusion calculation is performed: the target road network is divided into 100m × 100m grid units. Taking the grid where the anomaly event occurs as the center, the number of grids in which the anomaly influence spreads per unit time is calculated in combination with the road segment capacity and the current traffic flow. Finally, the total grid range of the anomaly propagation influence is determined as the traffic anomaly propagation influence area.

[0042] Step S520: Determine the road network collaborative control trigger nodes based on the traffic anomaly levels and the affected areas of each traffic anomaly propagation. Specifically, using the traffic anomaly levels and affected areas output in step S510 as input, set control trigger thresholds. For example, trigger collaborative control when the anomaly level reaches level two or above and the affected area covers three or more grid units. Determine the range of control trigger nodes, designating the road segments and intersections corresponding to the grid units where the anomaly occurred as core trigger nodes, the road segments and intersections corresponding to the boundary grid units of the affected area as peripheral trigger nodes, and the traffic signal controllers, roadside guidance screens, and vehicle information push terminals corresponding to the above nodes as specific trigger device nodes, forming a list of road network collaborative control trigger nodes that includes node number, node type, trigger device type, and trigger conditions.

[0043] Step S530: Construct a control strategy candidate set based on the road network cooperative control trigger nodes. The control strategy candidate set includes signal timing adjustment strategies, lane passage guidance strategies, path diversion guidance strategies, and roadside prompt release strategies. Specifically, using the road network cooperative control trigger nodes output in step S520 as input, corresponding control strategies are generated according to the type, location, and anomaly level of the trigger nodes. For example, for congested intersections, a signal timing adjustment strategy that extends green light duration and adjusts phase sequence is generated; for congested road sections, a lane passage guidance strategy that prohibits lane changes and allows the use of turning lanes is generated; for large-scale congestion, a path diversion guidance strategy that recommends early diversion and detours is generated; and for nearby vehicles, a roadside prompt release strategy that provides speed limits and detour prompts is generated. The above strategies are summarized to form a control strategy candidate set, which is then output as the control strategy candidate set.

[0044] Step S540 involves collaborative matching and conflict resolution of the control strategy candidate set to obtain the road network collaborative control strategy. Specifically, a priority ranking method combined with conflict matrix verification is used as the processing method. Taking the control strategy candidate set output in step S530 as input, strategy priority ranking is performed. For example, a preset priority rule is: ensuring life safety > alleviating congestion > improving traffic efficiency > providing auxiliary prompts. All strategies in the candidate set are ranked from highest to lowest priority according to this rule. Conflict matrix verification is performed, constructing a strategy conflict matrix. The matrix rows and columns represent different strategy types, and matrix elements indicate whether strategies conflict. The ranked strategies are sequentially substituted into the conflict matrix to verify whether the current strategy conflicts with the retained strategies. If a conflict is found, the strategy with the lower priority is removed. All strategies that pass the verification are integrated to form a complete scheme including strategy type, execution node, execution parameters, and execution time.

[0045] This application embodiment acquires multi-source traffic perception data from the target road network, including vehicle-mounted sensors, roadside cameras, millimeter-wave radar, lidar, and traffic signal equipment. It then performs spatiotemporal alignment and denoising fusion on this data to obtain a traffic perception fusion dataset. Based on this dataset, a road network traffic situation model is constructed. Multiple traffic anomaly event feature vectors are obtained through target detection and trajectory tracking of the model. Based on these feature vectors, a road network collaborative control strategy is generated. This addresses the technical problem of existing road network management models failing to achieve unified perception and dynamic linkage across the entire road network. It achieves the technical effect of dynamic linkage and precise response in road network traffic management through high-precision perception of the entire road network data.

[0046] In the above text, refer to Figure 1 This paper describes in detail a multi-source high-precision sensing intelligent road network cooperative control method according to embodiments of the present invention. Next, reference will be made to... Figure 2 This invention describes a multi-source high-precision sensing intelligent road network cooperative control platform according to embodiments of the present invention.

[0047] The multi-source high-precision sensing intelligent road network collaborative control platform according to embodiments of the present invention addresses the technical problem of existing road network management and control models failing to achieve unified perception and dynamic linkage across the entire road network. It achieves high-precision perception of the entire road network data, enabling dynamic linkage and precise response in road network traffic management and control. The multi-source high-precision sensing intelligent road network collaborative control platform includes: a multi-source traffic perception data acquisition module 10, a data fusion module 20, a road network traffic situation model construction module 30, a traffic anomaly detection module 40, and a road network collaborative control strategy generation module 50.

[0048] The system includes a multi-source traffic perception data acquisition module 10, used to acquire multi-source traffic perception data of the target road network; a data fusion module 20, used to perform spatiotemporal alignment and denoising fusion on the multi-source traffic perception data to obtain a traffic perception fusion dataset; a road network traffic situation model construction module 30, used to construct a road network traffic situation model based on the traffic perception fusion dataset; a traffic anomaly detection module 40, used to perform target detection and trajectory tracking on the road network traffic situation model to obtain multiple traffic anomaly event feature vectors; and a road network cooperative control strategy generation module 50, used to generate a road network cooperative control strategy based on the multiple traffic anomaly event feature vectors.

[0049] The detailed description of the specific configuration of the data fusion module 20 is explained as follows: As mentioned above, the multi-source traffic perception data is spatiotemporally aligned and denoised to obtain a traffic perception fusion dataset. The data fusion module 20 may further include: a time synchronization processing unit for performing time synchronization processing on the multi-source traffic perception data according to a unified time reference to generate a multi-source time-aligned data sequence; a spatial coordinate mapping unit for performing spatial coordinate mapping and spatial matching processing on the multi-source time-aligned data sequence according to the road network spatial coordinate system to generate multi-source spatially aligned data; an abnormal data detection unit for performing abnormal data detection and noise data filtering processing on the multi-source spatially aligned data to obtain multi-source denoised traffic perception data; and a perception fusion processing unit for performing perception fusion processing on the multi-source denoised traffic perception data to obtain the traffic perception fusion dataset.

[0050] The detailed configuration of the road network traffic situation model construction module 30 is explained as follows: As mentioned above, a road network traffic situation model is constructed based on the traffic perception fusion dataset. The road network traffic situation model construction module 30 may further include: a traffic state feature parameter extraction unit for extracting traffic state feature parameters of each road segment based on the traffic perception fusion dataset, and obtaining multiple road segment traffic state feature vectors; a road segment traffic state matrix construction unit for constructing a road segment traffic state matrix based on the multiple road segment traffic state feature vectors; a topology association mapping unit for performing topology association mapping on the road segment traffic state matrix based on the target road network topology structure, and establishing road network traffic association relationships; and a traffic state propagation modeling unit for performing traffic state propagation modeling on the road segment traffic state matrix based on the road network traffic association relationships, and generating the road network traffic situation model.

[0051] The specific configuration of the traffic anomaly detection module 40 is described in detail below: As mentioned above, the traffic anomaly detection module 40 performs target detection and trajectory tracking on the road network traffic situation model to obtain multiple traffic anomaly event feature vectors. The traffic anomaly detection module 40 may further include: a target detection unit for performing target detection on the road network traffic situation model and obtaining traffic target detection results; a trajectory tracking unit for performing trajectory tracking on the road network traffic situation model based on the traffic target detection results and obtaining multiple target trajectory feature vectors; and an anomaly identification unit for performing anomaly identification on the multiple target trajectory feature vectors and generating the multiple traffic anomaly event feature vectors.

[0052] The detailed configuration of the road network cooperative control strategy generation module 50 is explained below: As described above, a road network cooperative control strategy is generated based on the feature vectors of the multiple traffic anomaly events. The road network cooperative control strategy generation module 50 may further include: an event type identification unit for identifying the event type and assessing the impact range of the multiple traffic anomaly event feature vectors to obtain the traffic anomaly level and the propagation impact area of ​​each traffic anomaly; a road network cooperative control trigger node determination unit for determining the road network cooperative control trigger node based on the traffic anomaly level and the propagation impact area of ​​each traffic anomaly; a control strategy candidate set construction unit for constructing a control strategy candidate set based on the road network cooperative control trigger node; and a cooperative matching unit for performing cooperative matching and conflict resolution on the control strategy candidate set to obtain the road network cooperative control strategy.

[0053] The detailed description of the specific configuration of the multi-source traffic perception data acquisition module 10 is explained as follows: As mentioned above, the multi-source traffic perception data acquisition module 10 may further include: the multi-source traffic perception data includes vehicle-mounted perception data, roadside camera data, millimeter-wave radar data, lidar data, and traffic signal equipment data.

[0054] The control strategy candidate set construction unit may further include: the control strategy candidate set includes signal timing adjustment strategy, lane traffic guidance strategy, route diversion guidance strategy and roadside prompt release strategy.

[0055] The intelligent road network cooperative control platform with multi-source high-precision sensing provided in the embodiments of the present invention can execute the intelligent road network cooperative control method with multi-source high-precision sensing provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method.

[0056] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0057] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for collaborative control of intelligent road networks based on multi-source high-precision sensing, characterized in that, The method includes: Acquire multi-source traffic perception data of the target road network; The multi-source traffic perception data is spatiotemporally aligned and denoised and fused to obtain a traffic perception fusion dataset. Based on the traffic perception fusion dataset, a road network traffic situation model is constructed; Target detection and trajectory tracking are performed on the road network traffic situation model to obtain feature vectors of multiple traffic anomalies; Based on the feature vectors of the multiple traffic anomaly events, a road network cooperative control strategy is generated.

2. The intelligent road network cooperative control method based on multi-source high-precision sensing as described in claim 1, characterized in that, The multi-source traffic perception data is spatiotemporally aligned and denoised and fused to obtain a traffic perception fusion dataset, including: The multi-source traffic sensing data is time-synchronized according to a unified time reference to generate a multi-source time-aligned data sequence. Based on the road network spatial coordinate system, the multi-source time-aligned data sequence is subjected to spatial coordinate mapping and spatial matching processing to generate multi-source spatially aligned data. The multi-source spatially aligned data is subjected to anomaly detection and noise filtering to obtain multi-source denoised traffic perception data. The traffic perception fusion dataset is obtained by performing perception fusion processing on the multi-source denoised traffic perception data.

3. The intelligent road network cooperative control method based on multi-source high-precision sensing as described in claim 1, characterized in that, Based on the traffic perception fusion dataset, a road network traffic situation model is constructed, including: Based on the traffic perception fusion dataset, traffic state feature parameters of each road segment are extracted to obtain multiple road segment traffic state feature vectors. Based on the traffic state feature vectors of the multiple road segments, a road segment traffic state matrix is ​​constructed; Based on the target road network topology, the traffic state matrix of the road segment is topologically correlated and mapped to establish road network traffic correlation relationships. Based on the traffic correlation of the road network, the traffic state matrix of the road segment is modeled for traffic state propagation to generate the road network traffic situation model.

4. The intelligent road network cooperative control method based on multi-source high-precision sensing as described in claim 1, characterized in that, The road network traffic situation model is subjected to target detection and trajectory tracking to obtain feature vectors of multiple traffic anomaly events, including: Target detection is performed on the road network traffic situation model to obtain traffic target detection results; Based on the traffic target detection results, the road network traffic situation model is tracked to obtain multiple target trajectory feature vectors; Anomaly identification is performed on the feature vectors of the multiple target trajectories to generate the feature vectors of the multiple traffic anomaly events.

5. The intelligent road network cooperative control method based on multi-source high-precision sensing as described in claim 1, characterized in that, Based on the feature vectors of the multiple traffic anomaly events, a road network cooperative control strategy is generated, including: The event type is identified and the impact range is assessed based on the feature vectors of the multiple traffic anomalies, and the levels of traffic anomalies and the areas affected by the propagation of each traffic anomaly are obtained. Based on the traffic anomaly levels and the areas affected by the propagation of each traffic anomaly, the trigger nodes for road network collaborative control are determined; A set of candidate control strategies is constructed based on the road network cooperative control triggering nodes; The candidate control strategy set is matched and conflict resolved collaboratively to obtain the road network collaborative control strategy.

6. The intelligent road network cooperative control method based on multi-source high-precision sensing as described in claim 1, characterized in that, The multi-source traffic perception data includes vehicle-mounted perception data, roadside camera data, millimeter-wave radar data, lidar data, and traffic signal equipment data.

7. The intelligent road network cooperative control method based on multi-source high-precision sensing as described in claim 5, characterized in that, The candidate set of control strategies includes signal timing adjustment strategies, lane traffic guidance strategies, route diversion guidance strategies, and roadside prompt release strategies.

8. A multi-source high-precision sensing intelligent road network collaborative control platform, characterized in that, The platform is used to implement the intelligent road network cooperative control method with multi-source high-precision sensing as described in any one of claims 1-7, and the platform comprises: The multi-source traffic perception data acquisition module is used to acquire multi-source traffic perception data of the target road network; The data fusion module is used to perform spatiotemporal alignment and denoising fusion on the multi-source traffic perception data to obtain a traffic perception fusion dataset. The road network traffic situation model construction module is used to construct a road network traffic situation model based on the traffic perception fusion dataset. The traffic anomaly detection module is used to perform target detection and trajectory tracking on the road network traffic situation model and obtain feature vectors of multiple traffic anomaly events. The road network cooperative control strategy generation module is used to generate a road network cooperative control strategy based on the feature vectors of the multiple traffic anomaly events.

Images