A cloud computing-based urban traffic monitoring and management system

Abstract

The present invention discloses an urban traffic monitoring and management system based on cloud computing, which relates to the technical field of urban traffic monitoring and management, and includes an overflow perception module, an occupancy prediction module, a signal regulation module, a path guidance module, and a model optimization module; the occupancy prediction module, when there is a vehicle queue overflow phenomenon in the U-turn lane, obtains the lane occupancy information of the U-turn lane and its adjacent straight lanes in real time, analyzes it, and predicts the queue occupancy degree of the U-turn lane when a vehicle queue overflow occurs in the U-turn lane; the signal regulation module, according to the prediction result, respectively executes corresponding U-turn signal timing regulation measures. The present invention solves the problem that the prior art cannot dynamically regulate the signal timing based on the degree of vehicle queue overflow in the U-turn lane, realizes the accurate identification and hierarchical response of U-turn behaviors, and improves the intersection traffic efficiency and the accuracy of induction decision-making.

Description

Technical Field

[0001] The present invention relates to the technical field of urban traffic monitoring and management, and particularly to an urban traffic monitoring and management system based on cloud computing. Background Art

[0002] Urban traffic is one of the important basic systems for urban operation, covering multiple elements such as roads, vehicles, traffic signals, public transportation, and pedestrian flow. Its core task is to ensure the efficient and orderly flow of people, vehicles, and goods in the urban space. However, with the accelerating urbanization process and the continuous increase in the number of motor vehicles, traditional traffic management means are difficult to cope with the increasingly complex traffic conditions. Problems such as road congestion, frequent accidents, and traffic information asymmetry have become more prominent, bringing huge pressure to citizens' travel and urban operation. To achieve a comprehensive perception, real-time analysis, and intelligent scheduling of the urban traffic state, it is urgent to digitally upgrade the traffic system with the help of new-generation information technologies. As a technical architecture with powerful computing power, high concurrent processing capabilities, and characteristics of elastic resource allocation, cloud computing can provide centralized storage, high-speed processing, and intelligent decision-making support for the urban traffic system. By deploying a traffic monitoring and management platform in the cloud, it is possible to achieve unified access, dynamic analysis, and visual presentation of data such as traffic flow, road conditions, and equipment status within the whole city. And based on big data and AI algorithms, accurate congestion prediction, signal optimization control, and emergency event response plans can be provided, thus significantly improving the intelligent management level and operation efficiency of urban traffic.

[0003] Existing urban traffic monitoring and management technologies based on cloud computing usually achieve comprehensive monitoring and centralized management of urban traffic by constructing a multi-layer architecture of "perception layer - network layer - platform layer - application layer". In the perception layer, widely deployed front-end perception devices such as video cameras, geomagnetic sensors, radars, and GPS devices collect real-time road traffic flow, vehicle speed, traffic density, congestion conditions, and emergency event information; these data are stably and quickly transmitted to the cloud through the network layer (including wired networks, 5G / 4G communications, the Internet of Things, etc.); relying on the powerful data processing capabilities of the cloud computing platform, the platform layer performs fusion cleaning and formatting processing on multi-source heterogeneous data, and models and predicts the traffic state with the help of big data analysis and artificial intelligence algorithms to achieve functions such as real-time situation perception, congestion trend analysis, and abnormal event identification; finally, in the application layer, through management terminals, command and dispatch platforms, or mobile applications, the analysis results are presented to traffic managers in a visual form, and signal timing plans can be intelligently generated, inducement information can be released, and early warning event response strategies can be implemented, thereby realizing dynamic supervision and precise control of the entire urban traffic system and significantly improving traffic operation efficiency and emergency response capabilities.

[0004] The existing technologies have the following deficiencies:

[0005] At intersections with dedicated U-turn lanes, when the U-turn demand rapidly increases within a short period of time, and the length of the U-turn lane itself is not sufficient to fully accommodate all the vehicles waiting to make a U-turn, the excess vehicles will queue into the adjacent straight-through lanes, resulting in queue spillover. Although such spillover lasts for a short time, it significantly affects the normal traffic flow in the adjacent lanes, causing the straight-through vehicles to be blocked and resulting in a local decrease in traffic efficiency. However, this spatial occupancy shift caused by the U-turn demand is not accurately identified by the existing monitoring systems because the existing cloud computing-based urban traffic monitoring and management technologies cannot dynamically adjust the U-turn signal timing according to the queue occupancy degree of the U-turn lane when vehicle queue spillover occurs in the U-turn lane. The system still evaluates the timing demand based on the traditional lane traffic flow, wrongly judging the spillover vehicles as an increase in the straight-through flow, thus weakening the traffic weight in the U-turn direction in the timing strategy, further aggravating the U-turn waiting and lane conflicts, and then causing the regulation error to accumulate continuously, affecting the traffic organization efficiency and the accuracy of induction judgment of the entire intersection and even the surrounding areas.

[0006] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, and thus it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0007] The object of the present invention is to provide a cloud computing-based urban traffic monitoring and management system to solve the problems in the above background art.

[0008] To achieve the above object, the present invention provides the following technical solution: A cloud computing-based urban traffic monitoring and management system includes a spillover perception module, an occupancy prediction module, a signal regulation module, a path guidance module, and a model optimization module;

[0009] The spillover perception module monitors a intersection with a dedicated U-turn lane in real time by deploying traffic detection devices, obtains the queue lengths, position information, and movement trajectory information of the vehicles in the U-turn lane and its adjacent straight-through lanes, and analyzes them to determine whether there is vehicle queue spillover in the U-turn lane;

[0010] The occupancy prediction module, when there is vehicle queue spillover in the U-turn lane, obtains the lane occupancy information of the U-turn lane and its adjacent straight-through lanes in real time, and analyzes it to predict the queue occupancy degree of the U-turn lane when vehicle queue spillover occurs in the U-turn lane;

[0011] The signal regulation module, according to the prediction result, respectively executes corresponding U-turn signal timing regulation measures;

[0012] A path guidance module applies the prediction result and the turn-around signal timing control measures as feedback information to the traffic-induced path generation process to adjust the navigation path recommendation strategy and reduce the recommendation priority of the intersection as a turn-around point.

[0013] A model optimization module continuously monitors the occurrence frequency, duration of the queue overflow in the turn-around lane, and the execution effect of the timing control measures, optimizes the generation method of the prediction result based on historical data, and constructs a signal timing control template applicable to different time periods.

[0014] Preferably, in the occupancy prediction module, when there is a queue overflow phenomenon in the turn-around lane, the lane occupancy information of the turn-around lane and its adjacent straight lanes is obtained in real time and preprocessed after acquisition; the out-of-position queue space occupancy information and the lateral traffic behavior disturbance information are extracted from the preprocessed lane occupancy information and analyzed after extraction to generate the out-of-position queue compression structure coefficient and the lateral traffic interference index respectively; an occupancy evaluation model is constructed for the generated out-of-position queue compression structure coefficient and the lateral traffic interference index, and the occupancy prediction coefficient is generated by weighted summation; a preset occupancy prediction coefficient threshold interval is determined and compared with the generated occupancy prediction coefficient after determination, and the queue occupancy degree of the turn-around lane when a queue overflow occurs in the turn-around lane is predicted according to the comparison result.

[0015] Preferably, the acquisition logic of the out-of-position queue compression structure coefficient is as follows:

[0016] The out-of-position queue space occupancy information is extracted from the preprocessed lane occupancy information, specifically including the queue segment length of the turn-around vehicles in the adjacent straight lane at different moments within a period of time when there is a queue overflow phenomenon in the turn-around lane, the effective lane width of the adjacent straight lane in the occupied area, and the number of passing vehicles in the occupied area of the adjacent straight lane, and are respectively calibrated as L m 、W m和 V m ,L m represents the queue segment length of the turn-around vehicles in the adjacent straight lane at the m-th moment within a period of time when there is a queue overflow phenomenon in the turn-around lane, W m represents the effective lane width of the adjacent straight lane in the occupied area at the m-th moment within a period of time when there is a queue overflow phenomenon in the turn-around lane, V m represents the number of passing vehicles in the occupied area of the adjacent straight lane at the m-th moment within a period of time when there is a queue overflow phenomenon in the turn-around lane, ,h is a positive integer;

[0017] Calculate the out-of-position queue compression structure coefficient, and the specific calculation formula is as follows:

[0018]

[0019] Wherein, DCC is the off-site queuing compression structure coefficient.

[0020] Preferably, the acquisition logic of the lateral traffic interference index is as follows:

[0021] Extract the lateral traffic behavior disturbance information from the preprocessed lane occupancy information, specifically including the number of vehicles performing lateral lane change operations in adjacent straight lanes at different moments within a period of time when there is a phenomenon of vehicle queue overflow in the U-turn lane, the number of vehicles with a lateral trajectory deviation amplitude greater than a preset threshold in adjacent straight lanes, and the total lane width of adjacent straight lanes, and respectively label them as M m , H m and S, M m represents the number of vehicles performing lateral lane change operations in the adjacent straight lane at the m-th moment within a period of time when there is a phenomenon of vehicle queue overflow in the U-turn lane, H m represents the number of vehicles with a lateral trajectory deviation amplitude greater than a preset threshold in the adjacent straight lane at the m-th moment within a period of time when there is a phenomenon of vehicle queue overflow in the U-turn lane, and S represents the total lane width of the adjacent straight lane, , h is a positive integer;

[0022] Calculate the lateral traffic interference index, and the specific calculation formula is as follows:

[0023] ;

[0024] Wherein, LDI is the lateral traffic interference index.

[0025] Preferably, an occupancy evaluation model is constructed for the generated off-site queuing compression structure coefficient DCC and lateral traffic interference index LDI, and an occupancy prediction coefficient is generated by weighted summation. The specific calculation formula is as follows:

[0026]

[0027] Wherein, OPC is the occupancy prediction coefficient, and are the non-zero weight coefficients of the off-site queuing compression structure coefficient DCC and the lateral traffic interference index LDI respectively, and .

[0028] Preferably, determine the preset occupancy prediction coefficient threshold interval , and compare it with the generated occupancy prediction coefficient OPC after determination. According to the comparison result, predict the queue occupancy degree of the U-turn lane when there is a vehicle queue overflow situation in the U-turn lane. The specific comparison analysis is as follows:

[0029] If , when there is a vehicle queue overflow in the U-turn lane, the queue occupancy level of the U-turn lane is at a low occupancy level;

[0030] If , when there is a vehicle queue overflow in the U-turn lane, the queue occupancy level of the U-turn lane is at a medium occupancy level;

[0031] If , when there is a vehicle queue overflow in the U-turn lane, the queue occupancy level of the U-turn lane is at a high occupancy level.

[0032] Preferably, in the signal control module, according to the prediction result, corresponding U-turn signal timing control measures are respectively executed, specifically:

[0033] When the prediction result is at a low occupancy level, the U-turn signal timing control measures executed are specifically: maintaining the signal release duration and phase arrangement order of the current U-turn direction unchanged, and keeping the existing signal cycle structure and timing parameters;

[0034] When the prediction result is at a medium occupancy level, the U-turn signal timing control measures executed are specifically: increasing the signal release duration of the U-turn direction by a fixed value within the current signal cycle, executing this timing strategy for multiple consecutive cycles, and updating the release duration of the next cycle according to the length of the U-turn vehicle queue segment after each cycle ends;

[0035] When the prediction result is at a high occupancy level, the U-turn signal timing control measures executed are specifically: increasing the signal release duration of the U-turn direction to the longest passing phase level within the current cycle, advancing the U-turn phase to the front position of the overall phase sequence, and inserting an additional U-turn phase as a compensation phase at the end of this cycle to quickly eliminate the vehicle queue overflow phenomenon of U-turn vehicles.

[0036] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0037] 1. By constructing an overflow perception module and an occupancy prediction module, the present invention can realize the real-time monitoring and detailed identification of the vehicle queue overflow phenomenon of U-turn vehicles in intersections with dedicated U-turn lanes, overcoming the misjudgment problem caused by the rough judgment based only on the lane traffic flow in the prior art. The system can not only accurately capture whether the U-turn vehicle crosses the boundary and occupies the adjacent straight lane, but also collect its specific interference manifestations in the spatial dimension and behavior dimension, further extract the spatial occupancy information of out-of-position queuing and the disturbance information of lateral traffic behavior, and realize the quantitative expression of the overflow intensity by constructing the out-of-position queuing oppression structure coefficient and the lateral traffic interference index. This fine-grained data modeling method greatly improves the recognition sensitivity and description integrity of the system for U-turn abnormal behaviors, providing a solid foundation for subsequent dynamic control and prediction.

[0038] 2. The present invention introduces an occupancy prediction coefficient model and a comparison mechanism with the threshold interval, which can predict and divide the occupancy levels into low, medium and high levels according to the actual occupancy degree and disturbance intensity of the U-turn queue, breaking the limitation of the traditional signal timing mechanism relying on a single flow indicator. On this basis, the system can execute a multi-level progressive signal timing strategy according to the prediction level, including operations such as release duration adjustment, phase sequence rearrangement and compensation phase insertion, to achieve precise release control of the U-turn direction. This dynamic timing mechanism significantly enhances the flexibility of U-turn traffic at the intersection, which can maintain system stability in the case of mild overflow and achieve high-intensity intervention in severe congestion, thereby effectively alleviating the blocking effect of U-turn conflicts on the main channel and ensuring the overall traffic efficiency and order continuity of the intersection.

[0039] 3. On the basis of achieving accurate prediction and dynamic timing, the present invention further establishes a full-chain intelligent feedback mechanism through the path guidance module and the model optimization module. On the one hand, the system can link the occupancy prediction results and the control strategy to the path induction engine, dynamically adjust the navigation system's recommended priority for intersections as U-turn nodes, and collaboratively reduce the pressure of U-turns from the demand side; on the other hand, the system continuously monitors the overflow frequency, duration and control effect of the U-turn queue, and constructs a time-based signal control template based on historical data to achieve adaptive optimization of model parameters. The above mechanism connects the four key links of perception, evaluation, control and guidance, and constructs a highly coupled, strongly linked and evolvable urban U-turn traffic management model, which significantly improves the intelligence level of the signal control system and the global traffic management capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0041] Figure 1 This is a module schematic diagram of a cloud computing-based urban traffic monitoring and management system of the present invention. DETAILED DESCRIPTION

[0042] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0043] The present invention provides Figure 1A city traffic monitoring and management system based on cloud computing as shown includes an overflow perception module, an occupancy prediction module, a signal regulation module, a path guidance module, and a model optimization module;

[0044] The overflow perception module conducts real-time monitoring on intersections with dedicated U-turn lanes by deploying traffic detection devices, obtains the queue lengths, position information, and movement trajectory information of vehicles in the U-turn lane and its adjacent straight lanes, and analyzes them to determine whether there is a phenomenon of vehicle queue overflow in the U-turn lane;

[0045] To achieve real-time monitoring of intersections with dedicated U-turn lanes, various types of traffic detection devices can be deployed to obtain the vehicle queue lengths, position information, and movement trajectory information of the U-turn lane and its adjacent straight lanes. Specifically, video surveillance cameras can be installed in front of the intersection, and the head and tail positions can be extracted in real time through image recognition algorithms to calculate the queue lengths in the lanes. Additionally, point sensors combined with geomagnetic induction coils or lidar can be used to obtain the dynamic positioning information of vehicles on each lane. In the cloud platform, data from different detection devices are fused and processed. Through means such as time series synchronization, position matching, and vehicle identification reconstruction, continuous trajectory tracking of each vehicle in the lane is achieved, thereby extracting information such as the stationary position, driving direction, relative position relationship, and lane-changing trajectory of vehicles in the lane. These information, as an important basis for subsequent judgment of the U-turn overflow phenomenon, can reflect the actual running state of vehicles in real time and continuously, with high timeliness and integrity.

[0046] After obtaining the vehicle queue and movement trajectory data, intelligent judgment can be made on whether there is queue overflow in the U-turn lane through software logic. Specifically, first, the spatial range of the U-turn lane is set according to the design boundaries of the U-turn lane (such as start and end coordinate points or physical lengths). Then, it is determined through the trajectory data whether there are vehicles with stationary or low-speed running positions exceeding this spatial range and entering the adjacent straight lane area, and if this behavior lasts for a preset threshold (such as exceeding a certain number of seconds), it can be preliminarily judged as a queue overflow behavior. In addition, the software can also introduce indicators such as vehicle trajectory direction and pre-judged turning behavior. For example, if the path trend of a parked vehicle in the straight lane is consistent with the U-turn direction and it continuously maintains a low speed or a stationary state, the determination accuracy of "abnormal parking + deviation from the functional lane" can be further enhanced. The above analysis completely relies on the obtained trajectory data, position information, and lane topology structure, and without manual intervention, high-frequency, automated, and low-error-rate U-turn overflow identification can be achieved at the software level.

[0047] The core purpose of implementing this spillover perception is to solve the blind spot problem of traditional signal control strategies based on traffic flow and speed in spatial occupancy recognition. Although the spillover phenomenon in the U-turn lane is transient in the time dimension, it significantly interferes with the adjacent straight lanes in the spatial dimension. Especially in scenarios with high-frequency U-turn intersections or short U-turn lanes, the spillover of U-turn vehicles may continuously occupy the straight lane, resulting in a decline in the straight-through traffic capacity and even triggering minor congestion. Since existing systems generally rely on traffic flow counting and average vehicle speed to evaluate traffic efficiency and cannot identify the offset behavior of "vehicles in the lane but not supposed to be in this lane", it leads to misjudgment of control strategies and imbalance in resource allocation. By constructing a software-based spillover judgment mechanism, it is possible to achieve the conversion judgment from "logical lane" to "actual lane occupancy behavior", so that the control logic no longer simply relies on static traffic models but truly reflects the immediate availability of lane resources. This not only provides a more accurate decision-making basis for subsequent signal control but also lays a key data foundation for route guidance strategies and system prediction optimization, thus greatly enhancing the dynamic adaptability and traffic efficiency of the entire intersection and surrounding roads.

[0048] The occupancy prediction module, in the case of vehicle queue spillover in the U-turn lane, obtains the lane occupancy information of the U-turn lane and its adjacent straight lanes in real time, analyzes it, and predicts the queue occupancy degree of the U-turn lane when vehicle queue spillover occurs in the U-turn lane.

[0049] In this embodiment, in the occupancy prediction module, in the case of vehicle queue spillover in the U-turn lane, the lane occupancy information of the U-turn lane and its adjacent straight lanes is obtained in real time and preprocessed after acquisition; the out-of-position queue space occupancy information and lateral traffic behavior disturbance information are extracted from the preprocessed lane occupancy information and analyzed after extraction to generate the out-of-position queue oppression structure coefficient and the lateral traffic interference index respectively; an occupancy evaluation model is constructed for the generated out-of-position queue oppression structure coefficient and the lateral traffic interference index, and the occupancy prediction coefficient is generated by weighted summation; the preset occupancy prediction coefficient threshold interval is determined and compared with the generated occupancy prediction coefficient after determination, and the queue occupancy degree of the U-turn lane when vehicle queue spillover occurs in the U-turn lane is predicted according to the comparison result.

[0050] In the case of vehicle queue overflow in the U-turn lane, multi-type traffic detection devices (including video surveillance cameras, millimeter-wave radars, lidar, geomagnetic coils, etc.) installed at intersections can be used to monitor the U-turn lane and its adjacent straight lanes at high frequencies. The software platform accesses the data interfaces of these sensing devices to collect raw data such as the vehicle position, length, head / tail coordinates, lane number, and relative position within the lane on each lane in real time. By constructing a road topology map and a lane boundary model and combining the time-series data collected by the detection devices, the system can automatically calculate the space segments occupied by vehicles on the U-turn lane and the adjacent straight lanes in each monitoring cycle, and further form a structured "lane occupancy dataset". This dataset usually contains information such as the lane number where the vehicle is located, the absolute start and end positions of the vehicle within the lane, the vehicle type identifier, the lane-changing mark, and the preliminary judgment of the passing intention, which serves as the basic raw material for "lane occupancy information" and can continuously and dynamically reflect the actual usage status of lane resources, and is the premise for subsequent behavior modeling and occupancy analysis.

[0051] As the raw traffic data obtained by multi-source sensing devices, lane occupancy information usually has problems such as heterogeneous data formats, inconsistent coordinate systems, missed detections and misidentifications, and inconsistent data granularity. If directly used for behavior analysis and parameter calculation, it may lead to cumulative deviation, distorted results, and affect the model accuracy. Therefore, it is necessary to perform unified formatting and semantic cleaning on the data before it enters the analysis model, that is, "preprocessing". The preprocessing operations mainly include three categories: one is time-series alignment, which ensures that the data from different devices is in the same time reference system through timestamp calibration and interpolation filling; the second is spatial mapping reconstruction, which projects the vehicle position information onto the lane topology coordinate system to establish an accurate attribution relationship between the vehicle and the logical lane; the third is data denoising and completion, which identifies and repairs problems such as outliers, missing segments, and duplicate records through trajectory continuity judgment and position rationality detection. The above preprocessing process is completely completed by the software platform and is usually automatically executed based on a multi-threaded data cleaning framework and a spatial matching algorithm, and finally outputs a set of standardized lane occupancy data with unified semantics, standardized structure, and high accuracy for subsequent information extraction and behavior modeling modules to call.

[0052] Based on the acquired and preprocessed lane occupancy information, an automatic extraction of "off-site queuing space occupancy information" and "lateral traffic behavior disturbance information" can be realized in the software platform by constructing a semantic recognition and behavior annotation rule model. Specifically, the system first judges, based on the lane number, the position coordinates of the vehicle in the lane, and the vehicle movement trajectory in the preprocessed data, whether there is a phenomenon that a vehicle in the U-turn direction remains stationary or moves slowly for a long time but its position has exceeded the boundary of the U-turn lane and entered the adjacent straight lane. If so, such a vehicle is marked as an "off-site queuing vehicle", and information such as the occupied space range, the corresponding time, the width of the lane where it is located, and the relative traffic density is extracted, and combined to form the "off-site queuing space occupancy information". At the same time, the system combines the lateral coordinate change trend of the vehicle at consecutive moments to identify whether there is a lateral trajectory that significantly deviates from the original traffic path, and further analyzes the occurrence frequency of lateral lane-changing behavior within a unit time period, the space occupied by the lane-changing vehicle, and whether there is a behavior with a lateral deviation amplitude exceeding the set threshold, to form the "lateral traffic behavior disturbance information". This extraction process is based on trajectory fitting, spatial rule judgment, and event label generation algorithms, and can be automatically completed without manual intervention, ensuring the real-time, consistency, and structural standardization of data extraction, and providing standardized input for subsequent behavior modeling and parameter generation.

[0053] To determine the threshold interval of the occupancy prediction coefficient, a distribution fitting and model training mechanism based on historical traffic behavior data can be constructed to realize dynamic threshold setting at the software level. Specifically, the system first screens out a large number of sample data sets related to the U-turn lane queuing spillover events from the long-term monitoring data, and constructs a complete coefficient value distribution map based on the occupancy prediction coefficients generated in each sample period. Then, based on clustering algorithms (such as K-means or DBSCAN), cluster analysis is performed on these prediction coefficients to identify the naturally occurring aggregation intervals in the data, that is, the typical coefficient ranges corresponding to different occupancy degrees. Further, the software conducts an associated modeling with the actual traffic performance corresponding to this period (such as the queuing length of the U-turn lane, the degree of traffic obstruction in the adjacent lane, whether signal intervention is triggered, etc.), so as to logically establish a mapping relationship between each coefficient clustering segment and the three spillover intensity labels of "low occupancy", "medium occupancy", and "high occupancy". Finally, the software automatically extracts the maximum and minimum values of the coefficient intervals corresponding to each type of label, and constructs a three-segment occupancy prediction coefficient threshold interval. This threshold interval can be automatically updated with the scale and pattern evolution of historical data, and has statistical sufficiency, behavior interpretability, and algorithm adaptability, which is the core premise for implementing a hierarchical control strategy based on predictive evaluation.

[0054] In this embodiment, the acquisition logic of the off-site queuing compression structure coefficient is as follows:

[0055] The occupancy information of the heterogeneous queue space is extracted from the preprocessed lane occupancy information, including the queue length of the U-turn vehicles in the adjacent through lane at different times within a period of time when the vehicle queue overflow occurs in the U-turn lane, the effective lane width of the adjacent through lane in the occupied area, and the number of vehicles passing through the adjacent through lane in the occupied area, and they are marked as L m , W m和 V m , L m W represents the length of the queue of vehicles turning around in the adjacent straight lane at time m within a period of time when the vehicle queue overflow occurs in the U-turn lane. m V represents the effective lane width of the adjacent straight lane in the occupied area at time m when the U-turn lane has overflow. m It represents the number of vehicles passing through the adjacent straight lane within the occupied area at time m when the U-turn lane has overflow. , h is a positive integer;

[0056] In the event of overflow of vehicle queues in the U-turn lane, the system can perform high-frequency and continuous status monitoring of the relevant lanes through the video surveillance equipment, lidar or millimeter-wave radar and other sensing hardware deployed around the intersection, combined with the data fusion and trajectory recognition algorithms of the software platform. First, in order to obtain the queue length of U-turn vehicles in the adjacent straight lane, the system automatically extracts the stationary or low-speed motion trajectory of vehicles belonging to the U-turn path from the camera image or radar point cloud through image recognition algorithms or trajectory tracking models, and compares it with the road topology information to identify the set of vehicles that have deviated from the U-turn lane and actually stayed in the adjacent straight lane area. Based on the coordinate positions of the front and rear of the vehicle in the set, the system can accurately calculate the actual linear projection distance of the queue in the lane, which is the queue length L at the corresponding moment. m Secondly, the effective lane width W of the adjacent through lane in the occupied area m , which can be obtained through CAD design drawings, electronic maps or static road metadata in the platform in the road structure information, or scanned and entered once by sensor equipment during system initialization, and used as standard static parameters for subsequent regional mapping and coordinate conversion. In addition, to obtain the number of vehicles passing through the area V m, within each monitoring moment, the system counts the trajectory counts of all passing vehicles within the occupied area of the adjacent straight lane. Combining with the spatial range of the occupied area, the instantaneous traffic volume data of this area can be automatically generated on the software side. The above three data fields will be dynamically integrated and updated in real time through software logics such as spatial calibration, logical lane matching, vehicle path prediction, and behavior classification, and will be automatically classified into "out-of-position queuing space occupancy information" in the platform, forming the core input for subsequent calculation of the oppression structure coefficient. This entire process does not rely on manual marking, has a high degree of automation and real-time nature, and is the key foundation for the system to achieve dynamic prediction and strategy-driven.

[0057] Calculate the out-of-position queuing oppression structure coefficient, and the specific calculation formula is as follows:

[0058]

[0059] In the formula, DCC is the out-of-position queuing oppression structure coefficient.

[0060] To accurately measure the degree of spatial oppression caused by U-turn vehicles to the adjacent straight lane when queuing spillover occurs, an out-of-position queuing oppression structure coefficient DCC model constructed in a weighted average form is adopted. First, the formula calculates the square ratio of the queuing section length L m of U-turn vehicles in the adjacent straight lane to the effective width W m of the corresponding lane, that is , which is used to measure the proportion degree of the queuing behavior to the spatial structure of the straight lane at the m-th moment. The square processing is to enhance the non-linear amplification effect of the queuing length on the sense of oppression, reflecting the trend of "the longer, the more oppressive". Secondly, multiply by item, which is used to construct an inhibition factor for the traffic density V m of the adjacent straight lane. When V m is small (that is, the traffic capacity is weak), the value of this logarithmic term is large, indicating that the lane is more vulnerable to occupation interference; while when V m is large, this term converges rapidly, reflecting that high-traffic lanes have a stronger tolerance for short-term occupation. "+1" is used to avoid the division-by-zero problem when, ensuring the stability of the calculation logic. Finally, perform arithmetic averaging on the results of all h moments, then an overall trend evaluation of the oppression caused by U-turn vehicle queuing spillover to the adjacent straight lane is obtained in the time dimension, taking into account the balance between instantaneous anomalies and global stability, making this coefficient both spatially sensitive and temporally robust, and can be used as an important basis for triggering signal control strategies.

[0061] The numerical value of the off-site queuing pressure structure coefficient is significantly positively correlated with the queue occupancy degree when predicting the situation of vehicle queuing spillover in the U-turn lane. That is, the larger the coefficient, the more serious the occupation behavior of U-turn vehicles in the non-functional lane area, the stronger the spatial interference of the queue on the adjacent straight lane, and the higher the actual queue occupancy degree reflected. Specifically, if within multiple monitoring moments, U-turn vehicles continuously cross the boundary of the U-turn lane and form a relatively long queuing section (L m is large), while the width of the adjacent straight lane is small (W m is small), and the unit traffic capacity is weak (V m is small), then it will obtain a higher coefficient value in the DCC calculation, indicating that the U-turn queue has strong invasiveness and pressure in space. This state often indicates an obvious mismatch between the U-turn traffic demand and the physical lane resources, which is the core triggering condition for signal timing adjustment and traffic guidance intervention. Therefore, DCC can be used as an important parameter for predicting the queue occupancy degree. The increase in its value means that the spatial occupancy pressure of U-turn vehicles at the intersection increases, and the system should give priority to improving the signal release capacity in the U-turn direction to alleviate local traffic conflicts. This coefficient realizes the quantitative bridging between physical lane behavior and dynamic control strategies, and has high application value and response interpretability.

[0062] In this embodiment, the acquisition logic of the lateral traffic interference index is as follows:

[0063] Extract the lateral traffic behavior disturbance information from the preprocessed lane occupancy information, specifically including the number of vehicles performing lateral lane changes in the adjacent straight lane at different moments within a period of time when there is a vehicle queuing spillover phenomenon in the U-turn lane, the number of vehicles with a lateral trajectory offset amplitude greater than a preset threshold on the adjacent straight lane, and the total lane width of the adjacent straight lane, and calibrate them as M m , H m and S respectively. M m represents the number of vehicles performing lateral lane changes in the adjacent straight lane at the m-th moment within a period of time when there is a vehicle queuing spillover phenomenon in the U-turn lane. H m represents the number of vehicles with a lateral trajectory offset amplitude greater than a preset threshold on the adjacent straight lane at the m-th moment within a period of time when there is a vehicle queuing spillover phenomenon in the U-turn lane. S represents the total lane width of the adjacent straight lane, , h is a positive integer;

[0064] When there is a phenomenon of vehicle queuing and overflow in the U-turn lane, the system can use high-definition surveillance cameras, lidar, millimeter-wave radars and other sensing devices deployed at intersections and above the lanes to sample the traffic trajectories of vehicles in adjacent straight lanes at a high frequency, and analyze and extract lateral traffic behaviors through the trajectory recognition and trajectory reconstruction algorithms of the software platform. First, to obtain the number of vehicles M that perform lateral lane-changing behaviors at each monitoring moment in the adjacent straight lane m , the software platform judges whether there is a trajectory segment that crosses from one logical lane to another by comparing the coordinate change trends of vehicles in the lateral direction (i.e., perpendicular to the lane center line) in consecutive frames. Once the trajectory crosses the lane boundary and remains in a continuous moving state, the system marks it as an "effective lane-changing behavior" and counts the total number of lane-changing vehicles at that moment. Second, to judge which vehicles' lateral trajectory offset amplitudes exceed the preset threshold to identify serious interference behaviors, the system will construct a "lane reference center line" based on the original trajectory and calculate the offset distance between the lateral position of each vehicle and the center line at that moment. If this distance exceeds the set physical threshold (such as 1.5 meters), it is considered that the running trajectory of this vehicle has significantly disturbed the traffic order, and it will be recorded in the offset vehicle set, and counting this set gives H m . As for the total lane width S of the adjacent straight lane, it can be obtained through the static road structure data embedded in the system, including design drawing data, electronic map structure or initial radar scan results, as the standard scale for cross-lane calculation and trajectory projection of the adjacent straight lane. The above three types of data are automatically extracted and updated in real time through the trajectory recognition module, offset recognition model and spatial geometry matching engine in the software system, and finally archived in the "lateral traffic behavior disturbance information" dataset to support the real-time calculation and dynamic evaluation of the lateral traffic interference index (LDI).

[0065] Calculate the lateral traffic interference index, and the specific calculation formula is as follows:

[0066]

[0067] In the formula, LDI is the lateral traffic interference index.

[0068] To quantify the lateral behavior interference intensity caused by the queuing and overflow of U-turn vehicles to the normal traffic path of adjacent straight lanes, a parameter of the lateral traffic interference index LDI is constructed. Its calculation formula adopts a method of cumulative evaluation by time period. At multiple monitoring moments within a period of time, two types of key data are respectively counted and a non-linear function modeling logic is introduced. The first term in the formula is used to describe the vehicle density of lateral lane-changing operations per unit lane width at the Mth monitoring moment, where M mrepresents the number of lane-changing vehicles, and S is the total width of adjacent straight lanes. The introduction of the exponent 1.5 is to amplify the weight of high-density lateral interference behaviors, reflecting the non-linear trend that the denser the traffic flow and the more concentrated the lane-changing behaviors, the more severe the impact on traffic order. The second term in the formula is used to reflect the structural disturbances caused by vehicle groups with abnormally large deviation amplitudes, where H m represents the number of vehicles with trajectory deviation amplitudes exceeding the preset threshold. Using the logarithmic function for processing can avoid extreme values from dominating the model while retaining the characteristics of their behavior intensities. Finally, by taking the arithmetic mean of all h moments, a stability assessment of the overall disturbance trend over a period of time is constructed. Overall, this formula achieves a dynamic fusion between the spatial density dimension and the behavior complexity dimension, enabling LDI to have a high responsiveness to the intensity of lateral interference and temporal robustness, and can effectively support the hierarchical classification of subsequent control strategies and the decision-making of execution priorities.

[0069] There is a significant positive correlation between the numerical value of the lateral traffic interference index LDI and the queue occupancy degree when predicting the vehicle queuing and overspill situation in the turning lane. The larger the LDI value, the stronger the lateral disturbance caused by the overspill behavior to the adjacent straight-lane traffic path, and thus it can be inferred that the actual queue occupancy degree of the turning lane is higher. Specifically, when more overspill turning vehicles force the vehicles in the adjacent straight lane to perform lateral lane-changing operations, the lane-changing density increases, resulting in an increase in M m increases, and then term is significantly amplified, reflecting the instability of the local traffic order; at the same time, if the driving trajectory deviation amplitudes of a large number of vehicles exceed the set threshold, it indicates that the lateral traffic path is structurally distorted, and H m increases, term also rises, representing an increase in the intensity level of interference behaviors. These two changes together drive the increase of LDI, forming an abnormal discrimination signal at the "behavior level" in addition to spatial oppression. Therefore, the higher the LDI, the more it means that the overspill turning queuing behavior not only occupies lane resources spatially, but also causes complex interference to the traffic flow at the behavior level. Based on this, the system can judge that the queue occupancy degree in this situation is relatively high, and a higher release priority should be set for the turning direction in the signal timing strategy to maintain the stability of the traffic flow and the continuity of order.

[0070] In this embodiment, an occupancy assessment model is constructed for the generated abnormal queuing oppression structure coefficient DCC and lateral traffic interference index LDI, and an occupancy prediction coefficient is generated through weighted summation. The specific calculation formula is as follows:

[0071]

[0072] In the formula, OPC is the occupancy prediction coefficient, and are non-zero weight coefficients for the out-of-position queuing oppression structure coefficient DCC and the lateral traffic interference index LDI, respectively, and .

[0073] After calculating the out-of-position queuing oppression structure coefficient DCC and the lateral traffic interference index LDI, the system can construct the final occupancy prediction coefficient OPC in the software platform by weighted summation based on the occupancy evaluation model. This calculation process is automatically executed by the model synthesis module. Among them, DCC is used to reflect the queuing oppression intensity caused by U-turn vehicles on adjacent straight lanes in the spatial dimension, and LDI is used to reflect the behavior interference intensity of the lateral traffic path. The two respectively measure the occupancy performance in the two dimensions of spatial occupancy and behavior perturbation. The two weight coefficients and introduced in the weighted summation formula are quantitative expressions of the contribution degrees of DCC and LDI in the overall occupancy evaluation, and are both non-zero positive numbers, ensuring that both dimensional factors are taken into account in the model. Usually, and can be set according to the multi-round learning results of historical traffic data. For example, by minimizing the prediction error or fitting the actual occupancy level label, the optimal weight configuration can be determined using gradient descent, genetic algorithm or grid search method, so that the model has higher generalization ability and prediction accuracy in different traffic scenarios. At the same time, these two weight coefficients satisfy the constraint condition that their sum is 1, , so as to maintain the numerical scale consistency of OPC and make the model results have good interpretability and comparability. The finally generated OPC value, as a quantitative index of the occupancy degree of the U-turn lane queue, will participate in the subsequent grade division and signal control strategy selection process.

[0074] In this embodiment, a pre-set occupancy prediction coefficient threshold interval is determined, and after determination, it is compared with the generated occupancy prediction coefficient OPC. According to the comparison result, the occupancy degree of the U-turn lane queue when vehicle queuing overflows occurs in the U-turn lane is predicted. The specific comparison and analysis are as follows:

[0075] If , the occupancy degree of the U-turn lane queue when vehicle queuing overflows occurs in the U-turn lane is low occupancy degree;

[0076] This situation means that although there is a short-term queuing overflow phenomenon in the U-turn lane, its queuing length is short, the occupied area is limited, and the adjacent straight lane has strong traffic capacity and infrequent lane-changing behavior. The overall traffic operation still has good self-recovery ability. In such cases, there is no need to immediately adjust the U-turn signal timing, and the current strategy can be maintained to ensure the overall intersection traffic efficiency. If the frequency is not high, there is no need to adjust the path guidance either, so as to avoid unnecessary frequent changes to the navigation guidance system.

[0077] If , when there is a spillover of queuing vehicles in the U-turn lane, the occupancy level of the queue in the U-turn lane is at a medium occupancy level;

[0078] This situation indicates that the occupancy of the adjacent through lane caused by the spillover of the U-turn lane queue is at a medium level. At this time, the length of the spillover queue has reached a certain scale, the occupied area has had a significant impact on through traffic, and local lane-changing behaviors have started to be intensive, disturbing the continuity of the traffic flow path, which is likely to lead to a decrease in traffic efficiency or local congestion fluctuations. This state belongs to the "sensitive regulation area". It is recommended that the system appropriately increase the signal release time in the U-turn direction according to the overall traffic flow situation at the intersection to relieve the queuing backlog. At the same time, link the traffic guidance module and set the U-turn recommendation priority of this intersection to medium level to balance local intervention and network-wide stability and prevent the continuous expansion of local congestion into a full-network congestion chain.

[0079] If , when there is a spillover of queuing vehicles in the U-turn lane, the occupancy level of the queue in the U-turn lane is at a high occupancy level.

[0080] This situation shows that a large number of U-turn vehicles have overflowed into the adjacent through lane, and the occupied space area has significantly expanded. At the same time, it is accompanied by high-frequency and large-number lane-changing interference behaviors, seriously disrupting the normal traffic order. This state often triggers internal traffic conflicts and increased delays at the intersection, and may induce rear queuing and affect other traffic phases. At this time, it is necessary to immediately trigger the adjustment of the U-turn signal timing strategy, such as significantly increasing the release time in the U-turn direction, or adopting the U-turn priority phase mode to quickly relieve the queue. At the same time, the recommended weight of this intersection as a U-turn node should be reduced through the route guidance strategy to guide some U-turn demands to be diverted to downstream intersections, suppressing the spillover pressure from the source and ensuring the overall operation safety and stability of the road network.

[0081] The signal regulation module, according to the prediction results, respectively executes the corresponding U-turn signal timing regulation measures;

[0082] In this embodiment, in the signal regulation module, according to the prediction results, the corresponding U-turn signal timing regulation measures are respectively executed, specifically:

[0083] When the prediction result is a low occupancy level, the U-turn signal timing regulation measure executed is: maintaining the signal release duration and phase arrangement order in the current U-turn direction unchanged, and keeping the existing signal cycle structure and timing parameters;

[0084] When the predicted result indicates a low occupancy level, the software platform reads the timing parameter table in the current signal cycle. After detecting that the release time for the U-turn direction matches the current traffic state and there is no significant increasing trend in the queue, it does not trigger any parameter update operations and maintains the duration and phase arrangement order of the original U-turn signal phase. This operation can be achieved by setting a "low occupancy lock" flag bit in the policy execution module, keeping the signal timing static within the current cycle. The purpose of this design is to avoid frequent system adjustments to signals caused by slight fluctuations in the predicted value, thereby interfering with the stability of the overall signal timing plan. At the same time, it reduces algorithm resource consumption and ensures the efficient operation of the system in non-critical situations. Through this method, the system not only maintains its regulatory response ability but also has static adaptability, enhancing the robustness of the overall timing strategy.

[0085] When the predicted result indicates a medium occupancy level, the specific U-turn signal timing control measures are as follows: within the current signal cycle, increase the signal release duration for the U-turn direction by a fixed value, execute this timing strategy for multiple consecutive cycles, and update the release duration for the next cycle based on the length of the U-turn vehicle queue segment at the end of each cycle.

[0086] When the predicted result indicates a medium occupancy level, the software platform automatically increases the signal release duration for the U-turn direction in the current signal cycle by a preset fixed time value on the existing basis through the policy scheduling module. This fixed value can be set during system initialization or through the historical learning process and is stored in the policy parameter configuration file. Subsequently, at the end of each signal cycle, the system uses the real-time monitoring module to obtain the length of the U-turn vehicle queue segment at the end of the current cycle and dynamically adjusts the U-turn release duration for the next cycle based on the change trend of the queue length. If the queue length increases, the enhanced release duration is maintained; if the queue length decreases, it gradually returns to the initial value. The entire control process is implemented through the cycle policy update logic in the software platform, supporting a dual adaptive control method of "in-cycle increment + inter-cycle feedback". This design ensures that in the case of medium occupancy, it can quickly respond to queue growth while avoiding excessive release and wasting traffic resources, balancing the control intensity and signal resource allocation efficiency.

[0087] When the predicted result indicates a high occupancy level, the specific U-turn signal timing control measures are as follows: increase the signal release duration for the U-turn direction to the longest passing phase level within the current cycle, move the U-turn phase to the front position of the overall phase sequence, and insert an additional U-turn phase as a compensation phase at the end of this cycle to quickly eliminate the overflow phenomenon of the U-turn vehicle queue.

[0088] When the prediction result indicates a high occupancy level, the software platform triggers the advanced regulation mode and strengthens the U-turn signal control in a three-step manner. First, the platform automatically increases the signal release duration in the U-turn direction to the maximum duration among all the current cycle's traffic phases. This value can be obtained by extracting the maximum release record from historical cycles or other phase parameters in the current cycle. Second, through the signal phase reconstruction algorithm, the position of the U-turn phase is adjusted to the foremost position in the entire signal sequence to ensure its priority release. Finally, a compensation phase containing only the U-turn direction is appended at the end of the signal cycle for a short period to release the high-pressure queue a second time. The insertion of such a compensation phase is achieved through dynamic phase stack operations, and the software platform jointly coordinates the time and order of each phase through the period controller and scheduler. These multiple strengthening measures can significantly improve the efficiency of clearing the U-turn queue, reduce the risk of intersection blockage, and prevent the main traffic path from being interrupted due to U-turn queues. It is a strong intervention mechanism for severe high-occupancy situations to ensure rapid evacuation under extreme conditions.

[0089] The path guidance module uses the prediction result and the U-turn signal timing regulation measures as feedback information in the process of generating traffic guidance paths to adjust the navigation path recommendation strategy and reduce the recommendation priority of intersections as U-turn points.

[0090] In the process of generating traffic guidance paths, the software platform embeds the occupancy prediction results (such as low, medium, and high levels) and the corresponding U-turn signal timing regulation measures as real-time feedback variables into the weight correction model of the path generation engine by constructing a dynamic path weight update module. Specifically, the platform first identifies each intersection node with a U-turn lane in the road network topology map and creates a "U-turn recommendation weight" field that can be updated in real time. When there is an overflow of U-turn vehicles queuing at a certain intersection, the system dynamically calculates the recommendation inhibition factor of this node based on the OPC value of this intersection and the type of timing strategy currently being executed. For example, if the occupancy level is high and the timing strategy has executed a high-intensity regulation action, this factor is set to a negative offset value, and the priority of the path segment corresponding to this node is reduced through the path cost function. When the path engine calculates the shortest path or the optimal travel time, the recommendation priority of such U-turn points is automatically lowered, reducing their appearance frequency in the navigation results. The entire process is achieved through the feedback weight mapping model, the path scoring system, and the navigation output generator to ensure that the prediction-regulation-feedback forms a closed-loop linkage.

[0091] The reason for taking the U-turn signal timing strategy and prediction results as the dynamic feedback of the route guidance system is that simply relying on static navigation or average travel time data cannot accurately reflect the instantaneous congestion status of intersections during the current period. Especially in scenarios where U-turn queue spillover problems frequently occur, it is easy to have "misleading deviation" in navigation guidance, concentrating a large number of U-turn demands on nodes that are already under high pressure, leading to further deterioration of intersections and forming a vicious cycle of induction-congestion-misjudgment. By feeding OPC and timing response measures back to the route engine, it is possible to achieve the dynamic perception of the current traffic risk of intersections by the navigation strategy, and avoid misidentifying nodes in a high-occupation state as path choices with smooth traffic. Especially in areas with strong substitutability of U-turn points, this mechanism can prompt the navigation to preferentially guide vehicles to U-turn intersections with lower occupancy, realizing the coordinated control of demand-side guidance and supply-side release strategies, thereby optimizing the allocation of road network resources in a higher dimension, improving the overall operation efficiency, reducing conflict incentives, and enhancing the intelligence and response ability of the traffic guidance system.

[0092] The model optimization module continuously monitors the occurrence frequency, duration of queue spillover in the U-turn lane, and the execution effect of timing control measures, optimizes the generation method of prediction results based on historical data, and constructs signal timing control templates applicable to different time periods.

[0093] The core function of the model optimization module can be realized in the software platform by constructing a "historical behavior record database" and a "multi-dimensional strategy evaluation engine". After each signal cycle ends, the system automatically records whether queue spillover occurs in the U-turn lane, the duration of spillover, the U-turn signal timing control measures taken during the spillover period, and the change trend of the U-turn queue length under the action of these measures, whether the queue is cleared, whether there is another spillover, and other key status data. These information are uniformly encoded and stored to form a historical data set that can be traced by time period, located by lane, and labeled by strategy. Subsequently, through the strategy evaluation engine, a comparative analysis is carried out on the traffic response situations after implementing different timing control strategies under the same prediction level. For example, compare the spillover mitigation efficiency after adopting different U-turn release durations or phase insertion strategies under the prediction result of the same "high occupancy degree", and extract the "optimal control path". The system can also extract frequently occurring patterns in different time periods and different traffic flow densities, construct a signal timing control template library indexed by "time label + prediction level", and push this template to the real-time control module to achieve signal control preset that can be called on demand and respond quickly.

[0094] The core objective of building the model optimization module is to achieve closed-loop learning between the identification of overflow at U-turn queues and the regulation strategy, and to use historical behavior data to feed back the prediction model and control model, enabling them to have the ability of continuous evolution. In the actual traffic environment, there are significant differences in traffic flow intensity, U-turn demand distribution, and adjacent lane availability at different time periods. If the signal timing strategy only relies on static rules or empirical parameters, it is very likely to lead to slow response or insufficient regulation intensity. By continuously monitoring the overflow behavior of the U-turn lane and the feedback effect after regulation, the system can optimize the predicted generated path based on the real traffic evolution results (such as retraining weight coefficients, correcting parameter sensitivity, etc.), and improve the response accuracy of the OPC index. At the same time, by constructing the optimal timing templates for different time periods in combination with historical regulation effects, not only can the regulation response time be significantly shortened, but also strategies such as "quick release template" or "strong suppression template" can be automatically invoked under abnormal traffic conditions, enhancing the spatio-temporal adaptability and steady-state control ability of the signal regulation strategy, thus promoting the evolution of the entire intersection management from "static scheduling" to "data-driven dynamic optimal control".

[0095] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the real situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0096] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center in a wired or wireless (such as infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium that the computer can access or a data storage device such as a server or data center containing one or more available medium sets. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0097] It should be understood that in various embodiments of the present application, the sequence numbers of the above processes do not indicate the order of execution, and the order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0098] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present application.

[0099] In several embodiments provided by the present application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the above-described embodiments are merely illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be an indirect coupling or communication connection through some interfaces, devices or units, and can be in an electrical, mechanical or other form.

[0100] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0101] In addition, the functional units in various embodiments of the present application can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0102] As described above, it is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present application, and all should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. An urban traffic monitoring and management system based on cloud computing, characterized in that, It includes a spillover perception module, an occupancy prediction module, a signal regulation module, a path guidance module, and a model optimization module; The spillover perception module uses traffic detection devices to monitor in real time the intersections with dedicated U-turn lanes, obtains the queue lengths, position information, and movement trajectory information of vehicles in the U-turn lane and its adjacent straight-through lanes, analyzes them, and determines whether there is a vehicle queue spillover phenomenon in the U-turn lane; The occupancy prediction module, when there is a vehicle queue spillover phenomenon in the U-turn lane, obtains in real time the lane occupancy information of the U-turn lane and its adjacent straight-through lanes, analyzes it, and predicts the queue occupancy degree of the U-turn lane when a vehicle queue spillover occurs in the U-turn lane; In the occupancy prediction module, an occupancy evaluation model is constructed for the generated out-of-position queue oppression structure coefficient and lateral traffic interference index, and an occupancy prediction coefficient is generated through weighted summation; a pre-set threshold interval of the occupancy prediction coefficient is determined and compared with the generated occupancy prediction coefficient after determination, and the queue occupancy degree of the U-turn lane when a vehicle queue spillover occurs in the U-turn lane is predicted according to the comparison result; Indicates the queuing section length of U-turn vehicles in the adjacent straight lane for a period of time when there is an overflow of queuing vehicles in the U-turn lane at a certain moment, Indicates the effective lane width of the adjacent straight lane within the occupied area for a period of time when there is an overflow of queuing vehicles in the U-turn lane at a certain moment, Indicates the number of passing vehicles in the adjacent straight lane within the occupied area at moment m for a period of time when there is an overflow of queuing vehicles in the U-turn lane , is a positive integer; Calculate the out-of-position queue oppression structure coefficient, and the specific calculation formula is as follows: ， In the formula, DCC is the out-of-position queue oppression structure coefficient; Indicates the number of vehicles performing lateral lane-changing operations in adjacent straight lanes within a period of time when there is a phenomenon of vehicle queue overflow in the U-turn lane at a certain moment, Indicates the number of vehicles with a lateral trajectory deviation amplitude greater than a preset threshold on adjacent straight lanes within a period of time when there is a phenomenon of vehicle queue overflow in the U-turn lane at a certain moment, Indicates the total lane width of adjacent straight lanes , where \(n\) is a positive integer; Calculate the lateral traffic interference index, and the specific calculation formula is as follows: , In the formula, LDI is the lateral traffic interference index; The signal regulation module, according to the prediction result, respectively executes corresponding U-turn signal timing regulation measures; The path guidance module applies the prediction result and the U-turn signal timing regulation measures as feedback information to the traffic induction path generation process to adjust the navigation path recommendation strategy and reduce the recommendation priority of the intersection as a U-turn point; The model optimization module continuously monitors the occurrence frequency, duration of the queue spillover in the U-turn lane, and the execution effect of the timing regulation measures, optimizes the generation method of the prediction result based on historical data, and constructs a signal timing regulation template applicable to different time periods.

2. The urban traffic monitoring and management system based on cloud computing according to claim 1, characterized in that, In the occupancy prediction module, when there is a vehicle queue spillover phenomenon in the U-turn lane, the lane occupancy information of the U-turn lane and its adjacent straight-through lanes is obtained in real time and pre-processed after acquisition; Extract the out-of-position queue space occupancy information and lateral traffic behavior disturbance information from the pre-processed lane occupancy information, analyze them after extraction, and generate the out-of-position queue oppression structure coefficient and the lateral traffic interference index respectively.

3. The urban traffic monitoring and management system based on cloud computing according to claim 2, characterized in that, The acquisition logic of the out-of-position queue oppression structure coefficient is as follows: Extract the occupancy information of the out-of-position queue space from the preprocessed lane occupancy information, specifically including the queue segment lengths of U-turn vehicles in adjacent straight lanes at different times within a certain period when there is a phenomenon of vehicle queue overflow in the U-turn lane, the effective lane width of the adjacent straight lane within the occupied area, and the number of passing vehicles in the adjacent straight lane within the occupied area, and calibrate them respectively as , and .

4. A cloud computing-based urban traffic monitoring and management system according to claim 3, characterized in that, The acquisition logic of the lateral traffic interference index is as follows: Extract the lateral passing behavior disturbance information from the pre-processed lane occupancy information, specifically including the number of vehicles performing lateral lane change operations in adjacent straight lanes at different times within a period when there is a phenomenon of vehicle queue overflow in the U-turn lane, the number of vehicles with a lateral trajectory deviation amplitude greater than a preset threshold on the adjacent straight lanes, and the total lane width of the adjacent straight lanes, and respectively calibrate them as , and .

5. The urban traffic monitoring and management system based on cloud computing according to claim 4, characterized in that An occupancy evaluation model is constructed for the generated out-of-position queue oppression structure coefficient DCC and lateral traffic interference index LDI, and an occupancy prediction coefficient is generated through weighted summation. The specific calculation formula is as follows: , where OPC is the occupancy prediction coefficient, and are the non-zero weight coefficients of the out-of-position queuing compression structure coefficient DCC and the lateral traffic interference index, respectively, and .

6. The urban traffic monitoring and management system based on cloud computing according to claim 5, characterized in that Determine the pre-set occupancy prediction coefficient threshold interval , and after determination, compare it with the generated occupancy prediction coefficient OPC, and predict the queue occupancy degree of the U-turn lane when vehicle queue overflows occur in the U-turn lane according to the comparison result. The specific comparison and analysis are as follows: If , when there is a situation where vehicles queue up and overflow on the U-turn lane, the occupancy level of the queue on the U-turn lane is at a low occupancy level; If , when there is a spillover of queuing vehicles in the U-turn lane, the occupancy degree of the queue in the U-turn lane is at a medium occupancy level; If , when there is a queue overflow of vehicles in the U-turn lane, the occupancy level of the queue in the U-turn lane is a high occupancy level.

7. A cloud computing-based urban traffic monitoring and management system according to claim 6, characterized in that, In the signal regulation module, according to the prediction result, corresponding U-turn signal timing regulation measures are respectively executed, specifically: When the prediction result is a low occupancy degree, the U-turn signal timing regulation measures executed are specifically: maintaining the signal release duration and phase arrangement order in the current U-turn direction unchanged, and keeping the existing signal cycle structure and timing parameters; When the predicted result is medium occupancy, the specific turn-around signal timing control measures are as follows: within the current signal cycle, increase the signal release duration in the turn-around direction by a fixed value, execute this timing control measure for multiple consecutive cycles, and update the release duration of the next cycle according to the length of the turn-around vehicle queue section after the end of each cycle; When the predicted result is high occupancy, the specific turn-around signal timing control measures are as follows: increase the signal release duration in the turn-around direction to the longest passing phase level within the current cycle, advance the turn-around phase to the front position of the overall phase sequence, and insert an additional turn-around phase as a compensation phase at the end of this cycle to quickly eliminate the overflow phenomenon of the turn-around vehicle queue.

Images