Intersection signal advance intervention control method and system based on traffic demand prediction

Abstract

The application relates to the technical field of intelligent traffic, and provides a crossroad signal early intervention control method and system based on traffic demand prediction; the method comprises the following steps: acquiring crossroad traffic operation data, constructing real-time traffic demand information, combining historical traffic data and a current signal control state, establishing a traffic demand prediction model, and predicting traffic demands of each entrance in multiple signal periods in the future; a corresponding relationship between traffic flow and a signal phase is established according to a prediction result, a phase demand index is constructed, future release demands of each signal phase are evaluated, an early intervention control strategy is generated according to an evaluation result, and control parameters such as a signal period, a green light time, a phase switching time and a phase sequence are optimized and adjusted; and feedback correction is carried out according to a traffic operation state after execution. The application can identify future traffic demand changes in advance, improve crossroad operation efficiency, and reduce vehicle delay and queuing pressure.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, and in particular to a method and system for early intervention control of intersection signals based on traffic demand prediction. Background Technology

[0002] The effectiveness of intersection signal control directly impacts the efficiency of urban road traffic. Currently, urban intersection signal control primarily employs fixed-time control or adaptive signal control methods. While adaptive control can adjust signal parameters based on real-time traffic flow conditions, its control is primarily based on the current traffic situation. Adjustments are only made after vehicle queues form or traffic congestion occurs, making it a passive response control method with control lag issues, and thus unable to respond promptly to rapidly changing traffic demands.

[0003] Existing traffic detection methods typically rely on devices such as geomagnetic sensors, coils, and video detection to obtain traffic flow information. However, most methods focus primarily on macroscopic traffic parameters such as total traffic volume, average speed, and queue length at approach lanes. They struggle to accurately capture the dynamic changes in turning traffic demand, such as left turns, straight-ahead traffic, and right turns, within different approach lanes. Consequently, signal control cannot make fine-tuning adjustments for different turning traffic demands during the signal control process.

[0004] On the other hand, existing traffic flow recognition methods based on video analytics are mainly used for vehicle detection, trajectory extraction, and traffic flow statistics. They typically focus on the perception of the current traffic state and lack the ability to predict the trend of traffic demand changes over multiple signal cycles in the future. Although existing signal optimization methods based on traffic prediction can predict future traffic conditions, most of the predictions are for the overall traffic flow of the approach lanes. The prediction results are difficult to directly correspond to the release demand of specific signal phases, and cannot form an effective correlation between traffic demand prediction results and signal control parameters.

[0005] Therefore, existing technologies still lack a method that can obtain steering-level traffic demand based on vehicle trajectory, and combine it with future traffic demand forecasts to make advance judgments on signal phase release demand and dynamically adjust control parameters, so as to realize the transformation of intersection signal control from "real-time response" to "predictive advance intervention". Summary of the Invention

[0006] This invention provides a method and system for early intervention control of intersection signals based on traffic demand forecasting, which solves the shortcomings of existing technologies that can only achieve early intervention through real-time response control and are difficult to achieve in advance prediction. This invention realizes the transformation of intersection signal control from "real-time response" to "predictive early intervention".

[0007] This invention provides a method for early intervention control of intersection signals based on traffic demand forecasting, comprising:

[0008] Step S10: Obtain a traffic scene data model containing standardized video data, spatial mapping relationships, and traffic analysis area information;

[0009] Step S20: Identify motor vehicles in standardized video data and obtain the trajectory information of the motor vehicles. Preprocess the trajectory information to obtain continuous and complete trajectory information. The trajectory information includes vehicle number, spatial location, running time, speed, direction of travel, and lane.

[0010] Step S30: Match the continuous and complete running trajectory with the intersection road topology model. Based on the vehicle's entry and exit positions, lane connection relationships, and vehicle running path, determine the vehicle's corresponding entrance lane, exit lane, and actual travel path. The road topology model includes entrance lane number, exit lane number, lane connection relationships, permitted turning relationships, and stop line location information.

[0011] Step S40: Based on the characteristics of vehicle trajectory changes and road topology constraints, automatically identify vehicle turning behaviors such as left turn, straight, and right turn;

[0012] Step S50: After completing the vehicle turning behavior recognition, the number of vehicles turning at different times in each entrance lane is counted according to the preset time window, and the traffic flow of left turn, straight and right turn in each entrance lane is obtained respectively; and a real-time turning flow spatiotemporal matrix is ​​constructed.

[0013] Step S60: Based on the real-time turning flow spatiotemporal matrix, historical traffic operation data and current signal control status, construct an intersection traffic demand prediction model, and predict different turning traffic flows of each approach lane in multiple future signal cycles based on the prediction model.

[0014] Step S70: Based on the future turning flow prediction results, combined with the intersection lane channelization relationship and the current signal control scheme, establish the mapping relationship between vehicle turning traffic flow and signal phase, and convert the predicted traffic demand for left turns, straight ahead and right turns at each approach lane into the future release demand state of the corresponding signal phase.

[0015] Step S80: Based on the future phase traffic demand status and combined with the real-time operation status of the intersection, construct a phase demand evaluation index system. The evaluation index includes indicators such as predicted turning traffic flow, approach lane queue length, queue growth trend, phase saturation degree, vehicle arrival rate, and historical release efficiency.

[0016] Step S90: Based on the future phase traffic demand status and real-time operation status, construct a phase demand index for predictive control to evaluate the potential traffic pressure of each signal phase within a certain time range in the future.

[0017] Step S100: Output the future turning flow prediction results and phase demand index results to form the decision basis for early intervention and control of traffic signals;

[0018] Step S110: Based on the output of future turning flow prediction results and phase demand index results, evaluate the future traffic demand of each signal phase; and according to the size of the phase demand index, divide each signal phase into different demand levels to obtain the phase demand level. When the future traffic demand of a certain phase exceeds a preset threshold, determine that the phase has potential traffic pressure and trigger early intervention control.

[0019] Step S120: Establish an advance intervention control rule base based on the phase demand level, and generate a traffic signal advance intervention control strategy based on the phase demand index and the advance intervention control rules.

[0020] Step S130: Based on the generated traffic signal advance intervention control strategy, control the traffic signal control equipment to execute, and evaluate the control effect according to the real-time operating status; and when the traffic operating status changes or the control effect does not meet expectations, update the real-time turning flow spatiotemporal matrix, future traffic demand prediction results and phase demand index according to the latest traffic operating status, and correct the advance intervention control strategy.

[0021] According to the intersection signal early intervention control method based on traffic demand forecasting provided by the present invention, step S10 specifically includes the following steps:

[0022] Step S11: Install video acquisition equipment at the target intersection and acquire raw video data through the video acquisition equipment; preprocess the raw video data to obtain a standardized video sequence;

[0023] Step S12: Based on the camera installation parameters and the road space information at the intersection, perform spatial calibration on the video image and establish a mapping relationship between the image coordinate system and the actual road coordinate system;

[0024] Step S13: Based on the intersection road channelization form, lane distribution and traffic organization method, divide the video analysis area to construct the vehicle detection area, entrance lane area, exit lane area, stop line area, queue analysis area and turn recognition area, and dynamically update the area range according to changes in road structure and traffic organization adjustments.

[0025] The traffic scenario data model is formed based on steps S11 to S13.

[0026] According to the intersection signal early intervention control method based on traffic demand forecasting provided by the present invention, step S20 specifically includes the following steps:

[0027] Step S21: Use a multi-target tracking method to perform association matching on motor vehicle targets in continuous video frames. Determine the correspondence of the same vehicle in different time frames based on the changes in vehicle position, direction of movement and target feature information to achieve continuous vehicle tracking.

[0028] Step S22: Based on continuous vehicle tracking, connect the vehicles at different time locations in chronological order to form the vehicle's trajectory;

[0029] Step S23: Smooth and compensate for missing data on the generated vehicle trajectories to reduce trajectory discontinuities caused by vehicle detection errors, occlusion, and video frame drops, thereby obtaining a continuous and complete vehicle trajectory dataset.

[0030] ;

[0031] in: Indicates the first The continuous operating trajectory of the vehicles.

[0032] According to the intersection signal early intervention control method based on traffic demand forecasting provided by the present invention, in step S50, constructing the real-time turning flow spatiotemporal matrix specifically includes the following:

[0033] Let the first Within the first time window, the first Traffic flow at each entrance in different turning directions is as follows: ,in, Indicates the import lane number. Indicates the type of turn, including left turn, straight, and right turn;

[0034] Further construct the spatiotemporal matrix of real-time turning flow at intersections:

[0035] ;

[0036] The matrix elements represent the traffic demand for different turning directions at each entrance within the corresponding time window.

[0037] According to the intersection signal early intervention control method based on traffic demand forecasting provided by the present invention, the traffic demand forecasting model is constructed in step S60 as follows:

[0038] The inputs include historical turning flow patterns, current turning flow status, signal cycles, green light times, phase sequence, and phase differences. The system analyzes the temporal characteristics of traffic flow and the influence of signal control to obtain turning flow predictions for multiple future cycles.

[0039] According to the intersection signal early intervention control method based on traffic demand forecasting provided by the present invention, in step S90, the phase demand index is expressed as:

[0040] ;

[0041] in, Indicates the first The demand index for each signal phase; Indicates the first Standardized results of the evaluation indicators; Indicates the weight of the corresponding indicator; n represents the number of evaluation indicators;

[0042] By calculating the demand index for each signal phase, the future traffic demand for different phases is ranked to obtain the phase priority relationship.

[0043] According to the intersection signal early intervention control method based on traffic demand forecasting provided by the present invention, in step S120, the intervention control rule base includes phase demand priority rule, queue growth constraint rule, maximum waiting time constraint rule, minimum green light time constraint rule, maximum green light time constraint rule, and safe transition constraint rule.

[0044] Based on the current operation status of the intersection and future changes in traffic demand, the corresponding control rules are automatically matched. These rules simultaneously meet the upper and lower limits of the signal cycle, the minimum green light time, the maximum green light time, pedestrian crossing demand, and signal safety transition constraints, ensuring that the signal adjustment process meets the requirements of traffic efficiency and operational safety.

[0045] The traffic signal early intervention and control strategy is expressed as follows:

[0046] ;

[0047] In the formula: This indicates the optimized signal period; This indicates the optimized green light time allocation; This indicates the optimized phase execution order; This represents the optimized phase difference.

[0048] The present invention also provides an intersection signal early intervention control system based on traffic demand forecasting, comprising:

[0049] A video acquisition module is deployed at a road intersection to continuously acquire video data of traffic operations at the intersection and transmit the video data to a data processing module.

[0050] A data processing module, connected to the video acquisition module, is used to preprocess the video data; identify motor vehicles in the video data, obtain vehicle trajectory information, and preprocess the trajectory to obtain a continuous and complete trajectory; match the trajectory with the intersection road topology model to determine the vehicle's approach lane, exit lane, and turning behaviors such as left turn, straight, and right turn; statistically analyze the traffic flow of different turns at each approach lane according to a preset time window and construct a real-time turning flow spatiotemporal matrix; construct a traffic demand prediction model based on the spatiotemporal matrix, historical traffic operation data, and current signal control status to predict the traffic flow of different turns at each approach lane in multiple future signal cycles; convert the predicted traffic demand into the future release demand status of the corresponding signal phase, construct a phase demand index to evaluate the future traffic pressure of each signal phase; and generate a traffic signal early intervention control strategy based on the phase demand index and early intervention control rules.

[0051] The signal control module, connected to the data processing module, is used to receive the advance intervention control strategy and dynamically adjust the intersection signal control parameters under the conditions of meeting the minimum green light time, maximum green light time and safe transition constraints.

[0052] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the intersection signal advance intervention control method based on traffic demand prediction as described above.

[0053] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the intersection signal advance intervention control method based on traffic demand prediction as described above.

[0054] The present invention provides a method and system for early intervention control of intersection signals based on traffic demand forecasting. This method can identify future changes in traffic demand in advance, enabling the intersection signal control to shift from real-time response control to prediction-driven proactive control, thereby improving the operational efficiency of the intersection and reducing vehicle delays and queuing pressure.

[0055] Furthermore, by continuously executing a closed-loop process of "traffic demand forecasting—early intervention decision-making—signal control execution—operational status feedback," traffic signal control can proactively optimize based on future changes in traffic demand, thereby improving intersection operational efficiency. The traffic operation status obtained through feedback is further used to update the real-time turning flow spatiotemporal matrix and future traffic demand forecast results, achieving rolling optimization of traffic demand forecasting, control strategy generation, and control effect feedback. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0057] Figure 1 This is a flowchart illustrating the intersection signal early intervention control method based on traffic demand forecasting provided by the present invention.

[0058] Figure 2 This is a block diagram of the intersection signal early intervention control system based on traffic demand prediction provided by the present invention.

[0059] Figure 3 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0061] In related technologies, intersection traffic signal control methods mainly rely on inductive loops, radar, or video detection equipment to obtain current traffic flow information. It is difficult to accurately obtain vehicle turning trajectories and turning traffic demands. Moreover, existing signal control methods mostly adopt responsive control based on real-time traffic conditions, which cannot effectively predict future changes in traffic demand. This results in problems such as signal control lag, unreasonable allocation of phase resources, and low intersection traffic efficiency.

[0062] To address the aforementioned technical problems, this invention provides a method and system for refined signal early intervention control at intersections based on video trajectory reconstruction and spatiotemporal prediction of turning flow. This method is applicable to multi-intersection coordinated control scenarios, as detailed below.

[0063] like Figure 1As shown, this method uses traffic operation video acquired by intersection video acquisition equipment as the data basis, obtains turning-level traffic demand through vehicle trajectory reconstruction, constructs a traffic demand prediction model to predict turning traffic flow for multiple signal cycles in the future, and then constructs a phase demand index to evaluate the future release demand of each signal phase, generates a traffic signal early intervention control strategy and issues it for execution, and at the same time, it makes feedback corrections based on the traffic operation status after execution, forming a closed-loop optimization control of "traffic demand prediction - early intervention decision - signal control execution - operation status feedback".

[0064] This embodiment provides a method for early intervention control of intersection signals based on traffic demand forecasting, such as... Figure 1 As shown, the process includes steps S10 to S130, and the specific implementation methods for each step are as follows.

[0065] Step S10: Obtain traffic scene data model

[0066] Obtain a traffic scene data model containing standardized video data, spatial mapping relationships, and traffic analysis area information. Step S10 specifically includes steps S11 to S13.

[0067] Step S11: Install video acquisition equipment at the target intersection and acquire raw video data through the video acquisition equipment; preprocess the raw video data to obtain a standardized video sequence.

[0068] Specifically, video capture equipment is deployed at locations within the intersection that can completely cover the vehicle operating area. This video capture equipment includes high-definition network cameras, intelligent traffic cameras, or other devices with video capture capabilities. The video capture equipment can be installed on electronic police poles, monitoring poles, or high points on surrounding buildings above the intersection's approach lanes, at a height of, for example, 6 to 10 meters, with a viewing angle that completely covers the vehicle operating area from the stop line of each approach lane to the exit lane. The video capture frame rate is, for example, no less than 25 frames per second, and the resolution is, for example, no less than 1920×1080 pixels, to ensure the accuracy of vehicle target detection and tracking.

[0069] The video capture device continuously acquires traffic operation videos of the intersection and transmits the acquired video data to a traffic analysis server via wired or wireless network. Let the acquired video sequence be represented as:

[0070] ;

[0071] Where: It represents the t-th video frame; T represents the total number of video frames.

[0072] To reduce the impact of factors such as camera installation angle, lens distortion, lighting changes, weather conditions, and equipment vibration on traffic analysis results, the collected video data is preprocessed to obtain standardized video sequences that meet the requirements of vehicle detection and trajectory analysis. The preprocessing includes image correction, distortion correction, image enhancement, and time synchronization. Distortion correction uses the Zhang Zhengyou calibration method to obtain camera intrinsic parameters and distortion coefficients, and then corrects the distortion of video frames. Image enhancement uses histogram equalization, Gaussian filtering, and other methods to improve image quality. When multiple video acquisition devices are deployed at an intersection, time synchronization (e.g., based on Network Time Protocol (NTP)) ensures that the time reference of each video frame is consistent.

[0073] Step S12: Based on the camera installation parameters and the road space information at the intersection, perform spatial calibration on the video image and establish a mapping relationship between the image coordinate system and the actual road coordinate system.

[0074] Specifically, at least four non-collinear calibration points (such as lane line intersections, stop line endpoints, and other feature points) are selected on the road plane. The coordinates of each calibration point in the actual road coordinate system are measured, and its pixel coordinates in the video image are obtained. Based on the image coordinates and actual coordinates of each calibration point, the homography transformation matrix is ​​solved, and the homography mapping relationship between the image coordinate system and the actual road coordinate system is established. This realizes the transformation of vehicle position from image space to road space, providing a unified spatial reference for subsequent vehicle trajectory reconstruction, speed calculation, and steering behavior recognition.

[0075] Step S13: Based on the intersection road channelization form, lane distribution and traffic organization method, divide the video analysis area to construct the vehicle detection area, entrance lane area, exit lane area, stop line area, queue analysis area and turn recognition area, and dynamically update the area range according to changes in road structure and traffic organization adjustments.

[0076] Specifically, based on the positions of road markings such as lane lines, stop lines, and directional arrows in the actual road coordinate system, the video analysis area is divided into several functional areas: the vehicle detection area covers the range of vehicles entering the video frame; the approach lane area and exit lane area correspond to the lane ranges in each approach and exit direction of the intersection, respectively; the stop line area corresponds to the location of the stop lines in each approach lane; the queuing analysis area covers the queuing space upstream of the stop lines in the approach lanes; and the turning recognition area covers the vehicle turning space inside the intersection. Each functional area is stored in the traffic scene data model as a polygonal vector area, and the range of each functional area is updated accordingly when the road structure changes or the traffic organization is adjusted.

[0077] Based on steps S11 to S13, a traffic scene data model is formed, which includes standardized video data, spatial mapping relationships, and traffic analysis area information, and serves as the data input for subsequent vehicle detection and trajectory reconstruction.

[0078] Step S20: Vehicle Identification and Trajectory Acquisition

[0079] Motor vehicles are identified in standardized video data, and their trajectory information is obtained. This trajectory information is then preprocessed to obtain continuous and complete trajectory information. The trajectory information includes vehicle number, spatial location, travel time, speed, direction of travel, and lane. Step S20 specifically includes steps S21 to S23.

[0080] Step S21: Use a multi-target tracking method to perform association matching on motor vehicle targets in continuous video frames. Determine the correspondence of the same vehicle in different time frames based on the changes in the vehicle's position, direction of movement, and target feature information to achieve continuous vehicle tracking.

[0081] Specifically, a standardized video sequence is input into the vehicle detection module. A deep learning-based target detection model (such as the YOLO series target detection model) is used to identify motor vehicle targets in the video frames, obtaining the detection box position, target category, and time information of the vehicle targets. Based on this, a multi-target tracking method (such as DeepSORT, ByteTrack, and other multi-target tracking algorithms) is used to perform association matching on vehicle targets in consecutive video frames: Kalman filtering is used to predict the position of the vehicle target in the next frame, and the matching cost between the detection box and the existing tracking trajectory is calculated by combining the vehicle position change, movement direction, and appearance feature information. The data association is completed through the Hungarian algorithm to determine the correspondence of the same vehicle in different time frames, and a unique vehicle number is assigned to the same vehicle to achieve continuous vehicle tracking.

[0082] Step S22: Based on continuous vehicle tracking, connect the vehicles at different time locations in chronological order to form the vehicle's running trajectory.

[0083] Specifically, for each frame detection location corresponding to the same vehicle number, the spatial mapping relationship established in step S12 is converted into actual road coordinates, and these coordinates are connected in chronological order to form the vehicle's trajectory. The trajectory information includes the vehicle number, spatial location, running time, speed, direction of travel, and lane: the speed is calculated based on the spatial position difference between adjacent frames and the frame interval; the direction of travel is determined based on the azimuth of the vehicle's displacement vector; and the lane is determined based on the relative relationship between the vehicle's spatial location and the entrance lane area and lane range in the traffic analysis area information.

[0084] Step S23: Smooth and compensate for missing data on the generated vehicle trajectories to reduce trajectory discontinuities caused by vehicle detection errors, occlusion, and video frame drops, thereby obtaining a continuous and complete vehicle trajectory dataset.

[0085] ;

[0086] in: Let represent the continuous trajectory of the i-th vehicle, and n represent the number of vehicles.

[0087] Specifically, the position sequence in the vehicle trajectory is smoothed using Kalman filtering or moving average to suppress detection jitter. For trajectory interruptions caused by vehicle occlusion or video frame drops, if the interruption duration is less than a preset duration threshold, linear interpolation or spline interpolation is used to compensate for the missing information based on the position and speed information before and after the interruption point, thus completing the trajectory. Invalid trajectories with a length less than a preset length threshold are discarded. This results in a continuous and complete vehicle trajectory dataset, which describes the operational characteristics of vehicles throughout the entire process of passing through intersections, providing a data foundation for subsequent turning behavior recognition based on trajectory topology.

[0088] Step S30: Matching the running trajectory with the road topology model

[0089] The continuous and complete running trajectory is matched with the intersection road topology model. Based on the vehicle's entry and exit positions, lane connections, and running path, the corresponding entrance lane, exit lane, and actual travel path are determined. The road topology model includes entrance lane number, exit lane number, lane connection relationship, permitted turning relationship, and stop line position information.

[0090] Specifically, a road topology model of the intersection is pre-established. This model includes: the numbers of each approach lane and the corresponding lane divisions, the numbers of each exit lane, the lane connection relationships between approach and exit lanes, the permissible turning relationships of each approach lane, and the location information of each approach lane's stop line. For each continuous and complete trajectory, its initial trajectory point is spatially matched with each approach lane area to determine the approach lane and lane the vehicle enters the intersection; its final trajectory point is spatially matched with each exit lane area to determine the exit lane the vehicle leaves the intersection; and combined with the lane connection relationships and the actual trajectory's running path within the intersection, the actual travel path of the vehicle is determined.

[0091] Step S40: Automatic recognition of steering behavior

[0092] Based on the characteristics of vehicle trajectory changes and road topology constraints, the system automatically identifies vehicle turning behaviors such as left turn, straight-ahead, and right turn.

[0093] Specifically, the characteristics of the vehicle's trajectory change before and after entering the intersection are extracted, and the difference in heading angle between the vehicle's approach heading angle before entering the intersection and its exit heading angle after leaving the intersection is calculated. Simultaneously, based on the vehicle's approach and exit lane combination relationship determined in step S30, a joint judgment is made using the permissible turning relationship corresponding to that approach lane in the road topology model: when the approach and exit lane combination relationship corresponds to a unique turn, the turning type is directly determined; when multiple possible turns exist, left turn, straight turn, or right turn is determined based on the comparison result of the heading angle difference and a preset angle threshold. By jointly judging the trajectory change characteristics and road topology constraints, the turning recognition error caused by vehicle lane changes, avoidance maneuvers, and trajectory deviations is reduced.

[0094] Step S50: Steering Flow Statistics and Real-Time Steering Flow Spatiotemporal Matrix Construction

[0095] After completing the vehicle turning behavior recognition, the number of vehicles turning at different times in each entrance lane is counted according to the preset time window, and the traffic flow of left turn, straight and right turn in each entrance lane is obtained respectively; and a real-time turning flow spatiotemporal matrix is ​​constructed.

[0096] Specifically, the preset time window can be set according to the length of the intersection signal cycle, such as one signal cycle or a fixed duration of 1 to 5 minutes. Within each time window, based on the approach lane number and turning recognition results of each vehicle, the number of vehicles in different turning directions of each approach lane is cumulatively counted to obtain the traffic flow for left turns, straight ahead, and right turns of each approach lane.

[0097] Let the first Within the first time window, the first Traffic flow at each entrance in different turning directions is as follows: ,in, Indicates the import lane number. Indicates the type of turn, including left turn, straight, and right turn.

[0098] Further construct the spatiotemporal matrix of real-time turning flow at intersections:

[0099]

[0100] In this matrix, each element represents the traffic demand for different turning directions at each approach lane within the corresponding time window. The number of matrix rows corresponds to the number of approach lanes, and the three columns of each matrix row correspond to the traffic flow for left turns, straight ahead, and right turns at that approach lane, respectively. This real-time turning flow spatiotemporal matrix is ​​updated continuously with the time window, forming a time series of turning traffic demand for each approach lane, providing data input for traffic demand forecasting.

[0101] Step S60: Construct a traffic demand forecasting model

[0102] Based on the real-time turning flow spatiotemporal matrix, historical traffic operation data, and current signal control status, an intersection traffic demand prediction model is constructed. Based on the prediction model, different turning traffic flows at each approach lane in multiple future signal cycles are predicted.

[0103] Specifically, the inputs to the traffic demand forecasting model include historical turning flow patterns, current turning flow status, and signal control information such as signal cycle, green light time, phase sequence, and phase difference. Historical turning flow patterns are represented by the real-time turning flow spatiotemporal matrix sequence of each historical time window; the current turning flow status is represented by the real-time turning flow spatiotemporal matrix of the current time window; and the current signal control status includes the currently executing signal cycle, green light time for each phase, phase sequence, and phase difference. The traffic demand forecasting model can employ recurrent neural network time series forecasting models such as Long Short-Term Memory (LSTM) networks and Gated Recurrent Units (GRUs), or it can use time series analysis or regression forecasting models. Taking the LSTM model as an example, the turning flow spatiotemporal matrix sequence of several historical time windows and the corresponding signal control parameters are used as model inputs. By analyzing the temporal variation characteristics of traffic flow and the influence of signal control, the evolution law of turning traffic flow under signal control status is learned, and the prediction results of different turning traffic flows at each approach lane in multiple future signal cycles are output. Let the current time be... ,future The steering flow prediction results for each signal cycle are expressed as follows:

[0104] ;

[0105] in, Indicates the future number Each signal cycle corresponds to the predicted traffic demand for different turning directions at each approach lane. N is the preset number of prediction cycles, for example, 2 to 5 signal cycles.

[0106] Step S70: Mapping of Turning Traffic Flow and Signal Phase

[0107] Based on the future turning flow forecast results, combined with the intersection lane channelization relationship and the current signal control scheme, a mapping relationship between vehicle turning traffic flow and signal phase is established, and the predicted traffic demand for left turns, straight ahead and right turns at each approach lane is converted into the future release demand state of the corresponding signal phase.

[0108] Specifically, based on the lane channelization relationship at the intersection and the phase settings in the current signal control scheme, a mapping table of "approach lane number - lane connection relationship - permitted turning direction - corresponding release phase" is established. For each turn of each approach lane, the predicted traffic demand for that turn in future signal cycles is assigned to the corresponding signal phase according to the signal phase on which its release depends. The predicted traffic demand for each turn released in the same signal phase is summarized to obtain the future release demand status of each signal phase in future signal cycles, thereby realizing the conversion of traffic demand prediction results into signal control demand.

[0109] Step S80: Construct a phase demand evaluation index system

[0110] Based on the future phase traffic demand status and combined with the real-time operation status of the intersection, a phase demand evaluation index system is constructed. The evaluation index includes indicators such as predicted turning traffic flow, approach lane queue length, queue growth trend, phase saturation degree, vehicle arrival rate, and historical release efficiency.

[0111] Specifically, each evaluation indicator is obtained as follows: the predicted turning traffic flow is determined by the future release demand status of each signal phase obtained in step S70; the queuing length of the approach lane is determined based on the road length occupied by continuous queues of vehicles with speeds below a preset speed threshold or stationary vehicles within the queuing analysis area; the queuing growth trend is characterized by the rate of change of the queuing length of the approach lane in adjacent time windows; the phase saturation degree is characterized by the ratio of the actual traffic flow of the phase to the release capacity of that phase; the vehicle arrival rate is characterized by the number of vehicles passing through the stop line area of ​​the approach lane per unit time; and the historical release efficiency is characterized by the vehicle release utilization rate of the phase green light time within the historical signal cycle.

[0112] Step S90: Construct the phase demand index

[0113] Based on future phase traffic demand status and real-time operational status, a Phase Demand Index (PDI) for predictive control is constructed to evaluate the potential traffic pressure of each signal phase within a certain future time range.

[0114] Specifically, each evaluation index in step S80 is standardized, for example, by using the range normalization method to map each index value to the [0,1] interval, thus obtaining the standardized result of each index; the phase demand index is expressed as:

[0115] ;

[0116] in, Indicates the first The demand index for each signal phase; Indicates the first Standardized results of the evaluation indicators; denoted by , where n represents the number of evaluation indicators. The weights of each indicator can be determined using the analytic hierarchy process (AHP) or the entropy weight method, with future predicted turning traffic flow serving as the primary evaluation criterion and thus having a relatively large weight. By calculating the demand index for each signal phase, the future traffic demand for different phases is ranked to obtain the phase priority relationship, thereby achieving a quantitative evaluation of the future release pressure for different signal phases.

[0117] Step S100: Output decision basis

[0118] The system outputs future turning flow predictions and phase demand index results, forming the basis for decision-making regarding proactive traffic signal intervention and control. The outputs include predicted traffic demand for left turns, straight-ahead traffic, and right turns at each approach lane over multiple future signal cycles, the phase demand index corresponding to each signal phase, and the future release priority determined based on the phase demand index. These results serve as inputs for generating subsequent proactive intervention and control strategies, enabling a shift from real-time traffic condition perception to future traffic demand prediction and proactive signal control.

[0119] Step S110: Phase Demand Assessment and Grading

[0120] Based on the output of future turning flow prediction results and phase demand index results, the future traffic demand of each signal phase is evaluated; and according to the magnitude of the phase demand index, each signal phase is divided into different demand levels to obtain the phase demand level; when the future traffic demand of a certain phase exceeds a preset threshold, it is determined that there is potential traffic pressure in that phase and early intervention control is triggered.

[0121] Specifically, suppose there are m signal phases at the intersection, and the set of phase demand indices is represented as follows:

[0122] ;

[0123] in, Indicates the first The demand index for each signal phase. Based on the magnitude of the phase demand index, each signal phase is divided into different demand levels. For example, according to the preset first threshold (high demand threshold) / second threshold (medium demand threshold), each phase is divided into three levels: high demand, medium demand, and low demand. When the phase demand index of a certain phase exceeds the preset high demand threshold, it is determined that there is potential traffic pressure in the future time range, and early intervention control for that phase is triggered before the end of the current control cycle.

[0124] Understandably, unlike traditional methods that adjust signals based on current traffic conditions, this embodiment utilizes traffic demand forecasts for the next N signal cycles to identify potential traffic pressure in advance and generate optimized control schemes for the next control cycle or multiple future cycles, thereby achieving a shift from real-time response control to prediction-driven active control.

[0125] Step S120: Establish a rule base and generate early intervention control strategies.

[0126] A rule base for advance intervention control is established based on the phase demand level, and a traffic signal advance intervention control strategy is generated based on the phase demand index and the advance intervention control rules.

[0127] Specifically, the intervention control rule base includes phase demand priority rules, queue growth constraint rules, maximum waiting time constraint rules, minimum green light time constraint rules, maximum green light time constraint rules, and safe transition constraint rules: the phase demand priority rule means that phases with high phase demand indices are given priority in obtaining release resources; the queue growth constraint rule means that when the queue growth trend of a certain approach lane exceeds the preset growth rate, the release time of the corresponding phase is increased first; the maximum waiting time constraint rule means that the red light waiting time of each phase does not exceed the preset maximum waiting time; the minimum green light time constraint rule means that the green light time of each phase is not less than the preset minimum green light time to ensure pedestrian crossing needs and vehicle release safety; the maximum green light time constraint rule means that the green light time of each phase is not greater than the preset maximum green light time; and the safe transition constraint rule means that the transition time for yellow light, all red light, etc., is set according to specifications when switching phases.

[0128] Based on the current intersection operation status and future traffic demand changes, corresponding control rules are automatically matched. These rules simultaneously satisfy upper and lower limits of signal cycle, minimum green light time, maximum green light time, pedestrian crossing demand, and signal safety transition constraints, ensuring that the signal adjustment process meets traffic efficiency and operational safety requirements. Under the above constraints, with the optimization objective of minimizing total vehicle delay or maximizing traffic efficiency at the intersection, the signal control parameters for future control cycles are optimized (e.g., based on the Webster timing method or by allocating the green light ratio using the phase demand index), generating a traffic signal early intervention control strategy.

[0129] Based on the phase demand index ranking results, signal control parameters are collaboratively optimized. Signal cycles, green light time allocation, phase switching timing, and phase execution sequence are dynamically adjusted in conjunction with future traffic demand trends to prioritize high-demand traffic flows for passage resources. For adjacent intersections with coordinated control conditions, the coordinated phase difference can be further adjusted based on predicted traffic demand changes to improve regional traffic efficiency. The traffic signal early intervention control strategy is expressed as follows:

[0130] ;

[0131] in: This indicates the optimized signal period; This indicates the optimized green light time allocation; This indicates the optimized phase execution order; This represents the optimized phase difference.

[0132] Furthermore, the generated traffic signal early intervention control strategy is output and sent to the traffic signal control equipment for execution. The control strategy includes parameters such as optimized signal cycle, green light time for each phase, phase switching timing, phase sequence, and phase difference. This early intervention control strategy is dynamically generated based on future traffic demand predictions, realizing a shift in intersection signal control from "real-time detection – immediate adjustment" to "demand prediction – early optimization."

[0133] Step S130: Strategy Execution and Feedback Correction

[0134] Based on the generated traffic signal early intervention control strategy, the traffic signal control equipment is controlled to execute, and the control effect is evaluated according to the real-time operating status. When the traffic operating status changes or the control effect does not meet expectations, the real-time turning flow spatiotemporal matrix, future traffic demand prediction results, and phase demand index are updated according to the latest traffic operating status, and the early intervention control strategy is modified.

[0135] Specifically, the generated traffic signal early intervention control strategy is sent to the traffic signal control equipment (such as a road traffic signal controller). After receiving the control strategy, the traffic signal control equipment executes the optimized signal control scheme under the conditions of meeting the minimum green light time, maximum green light time and safe transition constraints. It dynamically adjusts control parameters such as intersection signal cycle, green light time of each phase, phase switching timing, phase execution sequence and phase difference to match the signal release status with future traffic demand.

[0136] The control strategy includes control parameters such as signal cycle, green light time, phase switching timing, phase execution sequence, and phase difference. After receiving the control strategy, the traffic signal control equipment executes the optimized signal control scheme under the conditions of meeting the minimum green light time, maximum green light time, and safe transition constraints, so that the signal release status matches future traffic demand.

[0137] During the execution of the signal control scheme, traffic operation status information at the intersection is continuously collected, including vehicle trajectory, traffic flow, queue length, vehicle delay, vehicle speed, and phase saturation. The control effect is evaluated based on the real-time operation status. When the traffic operation status changes or the control effect fails to meet the expected target, the real-time turning flow spatiotemporal matrix, future traffic demand forecast results, and phase demand index are updated based on the latest traffic operation status to further revise the early intervention control strategy.

[0138] The above embodiments enable traffic signal control to proactively optimize based on future changes in traffic demand by continuously executing a closed-loop process of "traffic demand forecasting - early intervention decision-making - signal control execution - operational status feedback," thereby achieving rolling optimization of traffic demand forecasting, control strategy generation, and control effect feedback.

[0139] This application also provides an intersection signal early intervention control system based on traffic demand forecasting, used to execute the aforementioned early intervention control method. For example... Figure 2 As shown, the system includes a video acquisition module, a data processing module, and a signal control module.

[0140] A video acquisition module, deployed at road intersections, continuously acquires video data of traffic operations at the intersections and transmits the video data to a data processing module. The video acquisition module includes high-definition network cameras, intelligent traffic cameras, or other devices with video acquisition capabilities, and is deployed in locations that can completely cover the area where vehicles are operating.

[0141] A data processing module, connected to the video acquisition module, is used to preprocess the video data; identify motor vehicles in the video data, obtain vehicle trajectory information, and preprocess the trajectory to obtain a continuous and complete trajectory; match the trajectory with the intersection road topology model to determine the vehicle's approach lane, exit lane, and turning behaviors such as left turn, straight, and right turn; statistically analyze the traffic flow of different turns at each approach lane according to a preset time window and construct a real-time turning flow spatiotemporal matrix; construct a traffic demand prediction model based on the spatiotemporal matrix, historical traffic operation data, and current signal control status to predict the traffic flow of different turns at each approach lane in multiple future signal cycles; convert the predicted traffic demand into the future release demand status of the corresponding signal phase, construct a phase demand index to evaluate the future traffic pressure of each signal phase; and generate a traffic signal early intervention control strategy based on the phase demand index and early intervention control rules. The data processing module can be deployed on a traffic analysis server and configured to execute the processing procedures described in steps S10 to S130 of the above embodiments.

[0142] A signal control module, connected to the data processing module, receives the advance intervention control strategy and dynamically adjusts the intersection signal control parameters under the conditions of minimum green light time, maximum green light time, and safe transition constraints. The signal control module includes a road traffic signal controller connected to the signal light groups in each direction of the intersection, and drives the signal light groups to execute the optimized signal control scheme according to the received control strategy.

[0143] Figure 3 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 3As shown, the electronic device may include a processor 310, a communication interface 320, a memory 330, and a communication bus 340, wherein the processor 310, the communication interface 320, and the memory 330 communicate with each other via the communication bus 340. The processor 310 can call logical instructions in the memory 330 to execute a traffic demand prediction-based intersection signal early intervention control method. This method includes: acquiring a traffic scene data model containing standardized video data, spatial mapping relationships, and traffic analysis area information; identifying motor vehicles in the standardized video data and obtaining the vehicle's trajectory information; preprocessing the trajectory information to obtain continuous and complete trajectory information; the trajectory information includes vehicle number, spatial location, travel time, travel speed, travel direction, and lane; and mapping the continuous and complete trajectory with the intersection road... The topology model is matched to determine the corresponding entrance lane, exit lane, and actual travel path of a vehicle based on its entry and exit positions at the intersection, lane connections, and vehicle travel paths. The road topology model includes entrance lane numbers, exit lane numbers, lane connections, permitted turning relationships, and stop line location information. Left turns, straight ahead, and right turns are automatically identified based on vehicle trajectory change characteristics and road topology constraints. After identifying vehicle turning behaviors, the number of vehicles making different turns at each entrance lane is counted according to a preset time window to obtain the traffic flow for left turns, straight ahead, and right turns at each entrance lane. Construct a real-time turning flow spatiotemporal matrix; based on the real-time turning flow spatiotemporal matrix, historical traffic operation data, and current signal control status, construct an intersection traffic demand prediction model, and predict different turning traffic flows for each approach lane within multiple future signal cycles based on the prediction model; based on the future turning flow prediction results, combined with the intersection lane channelization relationship and the current signal control scheme, establish a mapping relationship between vehicle turning traffic flow and signal phases, converting the predicted traffic demand for left turns, straight ahead, and right turns for each approach lane into the future release demand status of the corresponding signal phases; based on the future phase traffic demand status and combined with the real-time operation status of the intersection, construct... A phase demand evaluation index system is established, comprising indicators such as predicted turning traffic flow, approach lane queue length, queue growth trend, phase saturation, vehicle arrival rate, and historical release efficiency. Based on future phase traffic demand status and real-time operational status, a phase demand index for predictive control is constructed to evaluate the potential traffic pressure of each signal phase within a certain future timeframe. The system outputs future turning flow prediction results and phase demand index results, forming a decision-making basis for early intervention control of traffic signals. Based on the output of future turning flow prediction results and phase demand index results, the system evaluates the future traffic demand of each signal phase.Based on the phase demand index, each signal phase is divided into different demand levels, resulting in phase demand levels. When the future traffic demand for a certain phase exceeds a preset threshold, it is determined that the phase has potential traffic pressure, and early intervention control is triggered. An early intervention control rule base is established based on the phase demand levels, and a traffic signal early intervention control strategy is generated based on the phase demand index and the early intervention control rules. Based on the generated traffic signal early intervention control strategy, the traffic signal control equipment is controlled to execute, and the control effect is evaluated based on the real-time operating status. Furthermore, when the traffic operating status changes or the control effect does not meet expectations, the real-time turning flow spatiotemporal matrix, future traffic demand prediction results, and phase demand index are updated according to the latest traffic operating status, and the early intervention control strategy is corrected.

[0144] Furthermore, the logical instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0145] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the intersection signal early intervention control method based on traffic demand prediction provided by the above methods. The method includes: acquiring a traffic scene data model containing standardized video data, spatial mapping relationships, and traffic analysis area information; identifying motor vehicles in the standardized video data and obtaining the trajectory information of the motor vehicles; preprocessing the trajectory information to obtain continuous and complete trajectory information; the trajectory information includes vehicle number, spatial location, running time, and driving speed. The system identifies the vehicle's direction of travel and lane affiliation; it matches the continuous and complete travel trajectory with the intersection road topology model, determining the vehicle's entrance lane, exit lane, and actual travel path based on the vehicle's entry and exit positions, lane connections, and travel path; the road topology model includes entrance lane number, exit lane number, lane connections, permitted turning relationships, and stop line location information; it automatically identifies vehicle turning behaviors such as left turns, straight ahead, and right turns based on vehicle trajectory change characteristics and road topology constraints; after identifying vehicle turning behaviors, it counts the number of vehicles turning at different times in each entrance lane according to a preset time window, obtaining the left-turn, straight-ahead, and right-turn traffic flow for each entrance lane; and it constructs a real-time turning system. A traffic flow spatiotemporal matrix is ​​constructed. Based on the real-time turning flow spatiotemporal matrix, historical traffic operation data, and current signal control status, a traffic demand prediction model for the intersection is built. This model predicts the different turning traffic flows at each approach lane within multiple future signal cycles. Based on the future turning flow prediction results, combined with the intersection's lane channelization relationship and the current signal control scheme, a mapping relationship between vehicle turning traffic flow and signal phases is established, converting the predicted traffic demand for left turns, straight ahead, and right turns at each approach lane into the future release demand state for the corresponding signal phase. Based on the future phase traffic demand state and the real-time operation status of the intersection, a phase demand evaluation index system is constructed. The evaluation indexes include predicted turning traffic flow, approach lane queue length, queue growth trend, and phase... Indicators such as traffic saturation, vehicle arrival rate, and historical release efficiency are used. Based on future phase traffic demand status and real-time operation status, a phase demand index for predictive control is constructed to evaluate the potential traffic pressure of each signal phase within a certain time range in the future. The future turning flow prediction results and phase demand index results are output to form a decision basis for traffic signal early intervention control. Based on the output future turning flow prediction results and phase demand index results, the future traffic demand of each signal phase is evaluated. According to the size of the phase demand index, each signal phase is divided into different demand levels to obtain the phase demand level. When the future traffic demand of a certain phase exceeds a preset threshold, it is determined that there is potential traffic pressure in that phase and early intervention control is triggered.An advance intervention control rule base is established based on phase demand levels. Based on the phase demand index and advance intervention control rules, a traffic signal advance intervention control strategy is generated. Based on the generated traffic signal advance intervention control strategy, traffic signal control equipment is controlled to execute, and the control effect is evaluated according to real-time operating status. Furthermore, when traffic operating status changes or the control effect does not meet expectations, the real-time turning flow spatiotemporal matrix, future traffic demand prediction results, and phase demand index are updated according to the latest traffic operating status, and the advance intervention control strategy is revised.

[0146] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0147] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0148] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for advance intervention control of intersection signals based on traffic demand forecasting, characterized in that, include: Step S10: Obtain a traffic scene data model containing standardized video data, spatial mapping relationships, and traffic analysis area information; Step S20: Identify motor vehicles in standardized video data and obtain the trajectory information of the motor vehicles. Preprocess the trajectory information to obtain continuous and complete trajectory information. The trajectory information includes vehicle number, spatial location, running time, driving speed, running direction and lane. Step S30: Match the continuous and complete running trajectory with the intersection road topology model. Based on the vehicle's entry and exit positions, lane connection relationships, and vehicle running path, determine the vehicle's corresponding entrance lane, exit lane, and actual travel path. The road topology model includes entrance lane number, exit lane number, lane connection relationships, permitted turning relationships, and stop line location information. Step S40: Based on the characteristics of vehicle trajectory changes and road topology constraints, automatically identify vehicle turning behaviors such as left turn, straight, and right turn; Step S50: After completing the vehicle turning behavior recognition, the number of vehicles turning at different times in each entrance lane is counted according to the preset time window, and the traffic flow of left turn, straight and right turn in each entrance lane is obtained respectively; a real-time turning flow spatiotemporal matrix is ​​constructed. Step S60: Based on the real-time turning flow spatiotemporal matrix, historical traffic operation data and current signal control status, construct an intersection traffic demand prediction model, and predict different turning traffic flows of each approach lane in multiple future signal cycles based on the prediction model. Step S70: Based on the future turning flow prediction results, combined with the intersection lane channelization relationship and the current signal control scheme, establish the mapping relationship between vehicle turning traffic flow and signal phase, and convert the predicted traffic demand for left turns, straight ahead and right turns at each approach lane into the future release demand state of the corresponding signal phase. Step S80: Based on the future phase traffic demand status and combined with the real-time operation status of the intersection, construct a phase demand evaluation index system. The evaluation index includes indicators such as predicted turning traffic flow, approach lane queue length, queue growth trend, phase saturation degree, vehicle arrival rate, and historical release efficiency. Step S90: Based on the future phase traffic demand status and real-time operation status, construct a phase demand index for predictive control to evaluate the potential traffic pressure of each signal phase within a certain time range in the future. Step S100: Output the future turning flow prediction results and phase demand index results to form the decision basis for early intervention and control of traffic signals; Step S110: Based on the output of future turning flow prediction results and phase demand index results, evaluate the future traffic demand of each signal phase; and according to the size of the phase demand index, divide each signal phase into different demand levels to obtain the phase demand level. When the phase demand index of a certain phase exceeds a preset threshold, determine that the phase has potential traffic pressure and trigger early intervention control. Step S120: Establish an advance intervention control rule base based on the phase demand level, and generate a traffic signal advance intervention control strategy based on the phase demand index and the advance intervention control rules. Step S130: Based on the generated traffic signal early intervention control strategy, control the traffic signal control equipment to execute, and evaluate the control effect according to the real-time operating status; and when the traffic operating status changes or the control effect does not meet expectations, update the real-time turning flow spatiotemporal matrix, future traffic demand prediction results and phase demand index according to the latest traffic operating status, and revise the early intervention control strategy.

2. The method for advance intervention control of intersection signals based on traffic demand forecasting according to claim 1, characterized in that, Step S10 specifically includes the following steps: Step S11: Install video acquisition equipment at the target intersection and acquire raw video data through the video acquisition equipment; preprocess the raw video data to obtain a standardized video sequence; Step S12: Based on the camera installation parameters and the road space information at the intersection, perform spatial calibration on the video image and establish a mapping relationship between the image coordinate system and the actual road coordinate system; Step S13: Based on the intersection road channelization form, lane distribution and traffic organization method, divide the video analysis area to construct the vehicle detection area, entrance lane area, exit lane area, stop line area, queue analysis area and turn recognition area, and dynamically update the area range according to changes in road structure and traffic organization adjustments. The traffic scenario data model is formed based on steps S11 to S13.

3. The method for advance intervention control of intersection signals based on traffic demand forecasting according to claim 1, characterized in that, Step S20 specifically includes the following steps: Step S21: Use a multi-target tracking method to perform association matching on motor vehicle targets in continuous video frames. Determine the correspondence of the same vehicle in different time frames based on the changes in vehicle position, direction of movement and target feature information to achieve continuous vehicle tracking. Step S22: Based on continuous vehicle tracking, connect the vehicles at different time locations in chronological order to form the vehicle's trajectory; Step S23: Smooth and compensate for missing data on the generated vehicle trajectories to reduce trajectory discontinuities caused by vehicle detection errors, occlusion, and video frame drops, thereby obtaining a continuous and complete vehicle trajectory dataset. ； in: Indicates the first The continuous operating trajectory of the vehicles.

4. The method for advance intervention control of intersection signals based on traffic demand forecasting according to claim 1, characterized in that, In step S50, constructing the real-time steering flow spatiotemporal matrix specifically includes the following: Let the first Within the first time window, the first Traffic flow at each entrance in different turning directions is as follows: ,in, Indicates the import lane number. Indicates the type of turn, including left turn, straight, and right turn; Further construct the spatiotemporal matrix of real-time turning flow at intersections: ； The matrix elements represent the traffic demand for different turning directions at each entrance within the corresponding time window.

5. The method for advance intervention control of intersection signals based on traffic demand forecasting according to claim 1, characterized in that, In step S60, the traffic demand forecasting model is constructed as follows: The inputs include historical turning flow patterns, current turning flow status, signal cycles, green light times, phase sequence, and phase differences. The system analyzes the temporal characteristics of traffic flow and the influence of signal control to obtain turning flow predictions for multiple future cycles.

6. The method for advance intervention control of intersection signals based on traffic demand forecasting according to claim 1, characterized in that, In step S90, the phase demand index is expressed as: ； in, Indicates the first The demand index for each signal phase; This represents the standardized result of the j-th evaluation index for the i-th signal phase; Indicates the weight of the corresponding indicator; n represents the number of evaluation indicators; By calculating the demand index for each signal phase, the future traffic demand for different phases is ranked to obtain the phase priority relationship.

7. The method for advance intervention control of intersection signals based on traffic demand forecasting according to claim 1, characterized in that, In step S120, the intervention control rule base includes phase demand priority rules, queue growth constraint rules, maximum waiting time constraint rules, minimum green light time constraint rules, maximum green light time constraint rules, and safe transition constraint rules. Based on the current operation status of the intersection and future changes in traffic demand, the corresponding control rules are automatically matched. These rules simultaneously meet the upper and lower limits of the signal cycle, the minimum green light time, the maximum green light time, pedestrian crossing demand, and signal safety transition constraints, ensuring that the signal adjustment process meets the requirements of traffic efficiency and operational safety. The traffic signal early intervention and control strategy is expressed as follows: ； In the formula: This indicates the optimized signal period; This indicates the optimized green light time allocation; This indicates the optimized phase execution order; This represents the optimized phase difference.

8. A traffic demand forecast-based intersection signal early intervention control system, characterized in that, include: A video acquisition module is deployed at a road intersection to continuously acquire video data of traffic operations at the intersection and transmit the video data to a data processing module. A data processing module, connected to the video acquisition module, is used for preprocessing the video data; The system identifies motor vehicles in the video data, obtains vehicle trajectory information, and preprocesses the trajectory to obtain a continuous and complete trajectory. The trajectory is then matched with the intersection road topology model to determine the vehicle's entrance lane, exit lane, and turning behaviors such as left turn, straight, and right turn. Traffic flow at different turns at each approach lane is statistically analyzed according to a preset time window, and a real-time turning flow spatiotemporal matrix is ​​constructed. Based on the spatiotemporal matrix, historical traffic operation data, and current signal control status, a traffic demand prediction model is constructed to predict the traffic flow at different turns at each approach lane in multiple future signal cycles. The predicted traffic demand is converted into the future release demand status of the corresponding signal phase, and a phase demand index is constructed to evaluate the future traffic pressure of each signal phase. A traffic signal early intervention control strategy is generated based on the phase demand index and early intervention control rules. The signal control module, connected to the data processing module, is used to receive the advance intervention control strategy and dynamically adjust the intersection signal control parameters under the conditions of meeting the minimum green light time, maximum green light time, and safe transition constraints; and can update the real-time turning flow spatiotemporal matrix, future traffic demand prediction results, and phase demand index based on the operating status fed back by the signal control module, and correct the advance intervention control strategy.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the intersection signal early intervention control method based on traffic demand forecasting as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the intersection signal early intervention control method based on traffic demand forecasting as described in any one of claims 1 to 7.

Images