Intelligent cycle-based traffic signal phase control method and system, and intelligent phase sequence-based traffic signal phase control method and system

Abstract

The application provides a traffic signal phase control method and system based on intelligent cycle and a traffic signal phase control method and system based on intelligent phase sequence, which comprises the following steps: a visual acquisition and recognition module acquires vehicle video information of multiple lanes at an intersection and recognizes structured vehicle information of each lane; a traffic condition real-time acquisition module aligns real-time vehicle data and signal machine operation time information according to time and obtains a traffic real-time condition model by merging; a phase timing module calls lane real-time vehicle data obtained by the traffic condition real-time acquisition module, performs benchmarking and standardization processing on the lane real-time vehicle data, and applies a phase timing calculation table to time corresponding phases; and a traffic phase execution module sets timing control of each phase according to timing results of each phase calculated by the phase timing module. The application can reasonably allocate time of each phase and adjust phase sequence, so as to realize real-time intelligent traffic signal control.

Description

Technical Field

[0001] This invention relates to the field of traffic signal control technology, specifically to a traffic signal phase control method based on edge-end intelligent cycle and a system for implementing this method, and also to a traffic signal phase control method based on edge-end intelligent phase sequence and a system for implementing this method. Background Technology

[0002] With socio-economic development, the urban population and number of motor vehicles are increasing daily, leading to increasingly severe urban traffic congestion. Currently, urban roads have numerous intersections, many equipped with traffic lights. Motor vehicles and pedestrians follow these signals, but if the signal timings at multiple intersections are not set appropriately, it will exacerbate traffic congestion. Therefore, effective urban traffic signal control can significantly alleviate traffic congestion, while traditional control methods are ill-suited for complex traffic control problems.

[0003] Urban traffic signal control systems have evolved from manual to automatic, from timed to sensor-based, and from point-to-area control. Modern traffic signal control is characterized by randomness, nonlinearity, and time-varying nature, rendering traditional control methods based on precise mathematical models ineffective. With the development of intelligent transportation systems (ITS), signal control systems are increasingly adopting artificial intelligence (AI) technology, and various intelligent algorithms are being applied. ITS integrates advanced information technology, data communication technology, sensing technology, automatic control theory, operations research, artificial intelligence, mechatronics, and computer processing technology, offering advantages such as real-time performance, accuracy, and efficiency. Currently, the most widely used and mature signal timing optimization methods in ITS include: offline signal timing scheme optimization design and online sensor-based signal timing optimization modeling. These methods obtain accurate real-time traffic demand data to quickly and dynamically adjust signal system cycles, phase differences, green light ratios, phase sequences, etc., minimizing overall traffic delays across the entire area.

[0004] However, existing traffic signal sensing control methods are selection systems where multiple signal control schemes are pre-designed based on historical traffic data. Traffic data collected by road network sensors is used to select from these schemes, and timing adjustments are made locally. Overall, the optimization of the system's cycle, green light ratio, and phase difference is predetermined, limiting the degree of optimization of the timing scheme. Because traffic systems are influenced by numerous factors such as intersection location, people's behavior and habits, the distribution of surrounding facilities, and weather, they are typically nonlinear, fuzzy, and uncertain systems, making it extremely difficult to obtain an accurate model. Therefore, existing traffic control systems struggle to precisely control the phase sequence and timing of each phase at traffic intersections. Summary of the Invention

[0005] The first objective of this invention is to provide a traffic signal phase control method based on edge-end intelligent cycles that can accurately control the timing of each phase at a traffic intersection in real time.

[0006] The second objective of this invention is to provide a traffic signal phase control method based on edge-end intelligent phase sequence, which can dynamically adjust the phase sequence of each phase at a traffic intersection.

[0007] A third objective of this invention is to provide an edge-based intelligent cycle traffic signal phase control system that implements the above-described edge-based intelligent cycle traffic signal phase control method.

[0008] A fourth objective of this invention is to provide a traffic signal phase control system based on edge-based intelligent phase sequence that implements the above-described traffic signal phase control method based on edge-based intelligent phase sequence.

[0009] To achieve the aforementioned first objective, the intelligent cycle-based traffic signal phase control method based on the edge end provided by this invention includes: a visual acquisition and recognition module that acquires vehicle video information from multiple lanes at an intersection and uses an AI target recognition model to identify structured vehicle information for each lane; a real-time traffic condition acquisition module that calls the real-time vehicle data of the intersection acquired by the visual acquisition and recognition module and acquires the signal controller's operating information, aligns the real-time vehicle data and the signal controller's operating information according to time, and merges them to obtain a real-time traffic condition model; a phase timing module that calls the lane real-time vehicle data acquired by the real-time traffic condition acquisition module, performs benchmarking and standardization processing on the lane real-time vehicle data, and applies a phase timing calculation table to time the corresponding phases; and a traffic phase execution module that sets the timing control for each phase based on the timing results calculated by the phase timing module and sends control signals to the signal controller.

[0010] As can be seen from the above scheme, this invention utilizes video surveillance cameras at intersections to collect vehicle video information from each lane. It combines this with deep learning-based artificial intelligence recognition algorithms to detect traffic information from one or more lanes, such as vehicle count, speed, vehicle type, road occupancy, and queue length. These parameters are then used to construct two adaptive traffic phase control models: intelligent cycle and intelligent phase sequence. This optimized control strategy uses the shortest average queue length for each intersection and lane as its control objective, automatically allocating the duration of each phase to reduce phase idleness during off-peak hours, thereby improving intersection utilization. Compared to traditional methods, this strategy can simultaneously address multiple control objectives, including saturation balance, unsaturation control, and queue constraints, achieving better results during peak, off-peak, and low-peak times.

[0011] In addition, by benchmarking and standardizing real-time vehicle data of lanes, this invention can accurately calculate the actual traffic demand of each phase and better calculate the timing time required for each phase, making the calculated timing of each phase and the sequential control of each phase more accurate, which helps to reduce traffic congestion.

[0012] A preferred approach is that the phase timing module performs benchmarking and standardization processing on the real-time vehicle data of the lanes, including: first, benchmarking the real-time vehicle data of the lanes by converting various types of vehicles into a number based on cars; then, performing standardization calculations based on the benchmarked results by dividing the number of cars in that phase by the number of lanes in that phase to obtain the standardization result.

[0013] Therefore, converting different types of vehicles, such as cars, trucks, buses, and large trucks, into a quantity based on cars can quantify the actual travel time requirements for each lane, which is beneficial for subsequent standardized calculations.

[0014] A further approach is to perform standardized calculations such that, if a phase has only one entrance, the standardized result is the number of cars in that phase divided by the number of lanes in that phase; if a phase has two or more entrances, the standardized result for each entrance is calculated separately, and the maximum value of the standardized results among the multiple entrances is taken as the standardized result for that phase.

[0015] As can be seen, by using the corresponding calculation methods for both single and multiple entry points, the actual traffic demand for each phase can be accurately calculated. The standardized results are then used for timing, making the timing more accurate.

[0016] A further approach is to apply a phase timing calculation table to time the corresponding phases, including: calculating the phase timing ratio of the reference cycle and the proportion of vehicles corresponding to the phase; and calculating the final phase timing based on the phase timing ratio of the reference cycle and the proportion of vehicles corresponding to the phase. The value of the reference cycle is the sliding weighted average of multiple cycles that are a preset number of cycles before the current cycle.

[0017] Therefore, when determining the final phase timing, considering both the phase timing ratio of the reference cycle and the proportion of vehicles corresponding to each phase makes the final calculated phase timing more reasonable, taking into account the length of the reference cycle and allowing for dynamic adjustment based on real-time traffic conditions.

[0018] A further approach is to calculate the final phase timing based on the phase timing ratio of the reference cycle and the proportion of vehicles corresponding to each phase. This includes: calculating the average of the phase timing ratio of the reference cycle and the proportion of vehicles corresponding to each phase, and multiplying this average by the total duration of each phase cycle to obtain the final phase timing.

[0019] It is evident that using the average of the phase timing ratio and the ratio of the number of vehicles corresponding to the phase in the baseline cycle as the basis for timing calculation can simultaneously take into account the situation of previous cycles and the real-time traffic conditions.

[0020] To achieve the second objective mentioned above, the intelligent phase sequence traffic signal phase control method based on edge computing provided by this invention includes: a visual acquisition and recognition module that acquires vehicle video information from multiple lanes at an intersection and uses an AI target recognition model to identify structured vehicle information for each lane; a real-time traffic condition acquisition module that calls the real-time vehicle data of the intersection acquired by the visual acquisition and recognition module and acquires the signal controller operation information, aligns the real-time vehicle data and the signal controller operation information according to time, and merges them to obtain a real-time traffic condition model; a phase timing module that calls the real-time vehicle data of the lanes acquired by the real-time traffic condition acquisition module, performs benchmarking and standardization processing on the real-time vehicle data of the lanes, and applies a phase timing calculation table to time the corresponding phases; and a traffic phase execution module that executes the traffic signal phases according to the phase... Real-time vehicle data is used to adjust the order of each phase, and the timing control of each phase is set according to the timing results of each phase calculated by the phase timing module. Control signals are then sent to the traffic signal controller. When adjusting the order of each phase based on the real-time vehicle data of each phase: the first proportion of the standardized result of each phase to the standardized result of all phases is calculated; the second proportion of the number of times each phase appears in the completed preset number of cycles is calculated to the number of times all phases appear in the completed preset number of cycles is calculated; the difference between the first proportion and the second proportion of each phase is calculated, and the priority level of each phase is determined based on the difference between the first proportion and the second proportion; the order of each phase is determined based on the priority level of each phase, and if there are no vehicles in a certain phase, that phase is skipped.

[0021] As can be seen from the above scheme, this invention not only calculates the timing of each phase, but also intelligently adjusts the order of each phase, making the sequential operation of each phase more reasonable. Furthermore, for phases without vehicles, this invention can directly skip those phases, thereby avoiding wasted time due to configuring passage times for phases without vehicles.

[0022] A further approach involves calculating the second ratio of the normalized number of occurrences of each phase within a predetermined number of completed cycles to the normalized number of occurrences of all phases within a predetermined number of completed cycles. This includes: obtaining the number of occurrences of each phase within the predetermined number of completed cycles and calculating the normalized number of occurrences of each phase: the normalized number of occurrences of a phase is the number of occurrences of that phase multiplied by 2 raised to a predetermined exponent, where the predetermined exponent is the difference between the phase allocation fairness and the number of completed cycles; the second ratio is the ratio of the normalized number of occurrences of each phase to the normalized number of occurrences of all phases.

[0023] A further approach is to determine the priority level of each phase based on the difference between the first ratio and the second ratio, including: the priority level of a phase is positively correlated with the difference between the first ratio and the second ratio of that phase.

[0024] Therefore, the difference between the first ratio and the second ratio can reflect the current traffic demand of each phase, thereby rationally configuring the priority level of each phase, making the priority level configuration of each phase quantifiable, and achieving reasonable sorting, thus improving the rationality of the phase sequence.

[0025] To achieve the third objective mentioned above, the intelligent cycle-based traffic signal phase control system based on the edge of this invention includes a visual acquisition and recognition module, a real-time traffic condition acquisition module, a phase timing module, and a traffic phase execution module. The visual acquisition and recognition module acquires vehicle video information from multiple lanes at an intersection and uses an AI target recognition model to identify the structured vehicle information for each lane. The real-time traffic condition acquisition module calls the real-time vehicle data of the intersection acquired by the visual acquisition and recognition module and obtains the traffic signal operation information. It aligns the real-time vehicle data and the traffic signal operation information according to time and merges them to obtain a real-time traffic condition model. The phase timing module calls the real-time vehicle data of the lanes acquired by the real-time traffic condition acquisition module, performs benchmarking and standardization processing on the real-time vehicle data of the lanes, and applies a phase timing calculation table to time the corresponding phases. The traffic phase execution module sets the timing control for each phase based on the timing results calculated by the phase timing module and sends control signals to the traffic signal controller.

[0026] To achieve the fourth objective mentioned above, the intelligent phase sequence traffic signal control system based on edge computing provided by this invention includes a visual acquisition and recognition module, a real-time traffic condition acquisition module, a phase timing module, and a traffic phase execution module. The visual acquisition and recognition module acquires vehicle video information from multiple lanes at an intersection and uses an AI target recognition model to identify the structured vehicle information for each lane. The real-time traffic condition acquisition module calls the real-time vehicle data of the intersection acquired by the visual acquisition and recognition module and obtains the signal controller operation information, aligning the real-time vehicle data and the signal controller operation information according to time and merging them to obtain a real-time traffic condition model. The phase timing module calls the real-time vehicle data of the lanes acquired by the real-time traffic condition acquisition module and standardizes and benchmarks the real-time vehicle data of the lanes. The system processes traffic flow data and applies a phase timing calculation table to time the corresponding phases. The traffic phase execution module uses real-time vehicle data for each phase to adjust the phase order. Based on the timing results calculated by the phase timing module, it sets the timing control for each phase and sends control signals to the traffic signal controller. When adjusting the phase order, the system calculates: the first proportion of the standardized result for each phase relative to the standardized results for all phases; the second proportion of the standardized number of times each phase appears in a predetermined number of completed cycles relative to the standardized number of times all phases appear in a predetermined number of completed cycles; the difference between the first and second proportions for each phase; and determines the priority level of each phase based on the difference. Finally, it determines the phase order based on the priority level of each phase. Attached Figure Description

[0027] Figure 1 This is a diagram of an intersection.

[0028] Figure 2 This is a structural block diagram of an embodiment of the traffic signal phase control system based on intelligent cycle and intelligent phase sequence at the edge end of the present invention.

[0029] Figure 3 This is a structural block diagram of the signal transmission of each module in an embodiment of the intelligent cycle and intelligent phase sequence traffic signal phase control system based on the edge end of the present invention.

[0030] Figure 4 This is a structural block diagram of the visual acquisition and recognition module in an embodiment of the intelligent cycle and intelligent phase sequence traffic signal phase control system based on the edge end of the present invention.

[0031] Figure 5 This is a structural block diagram of the real-time traffic condition acquisition module in an embodiment of the intelligent cycle and intelligent phase sequence traffic signal phase control system based on the edge end of the present invention.

[0032] Figure 6This is a structural block diagram of the phase timing module in an embodiment of the traffic signal phase control system based on edge-end intelligent cycle and intelligent phase sequence of the present invention.

[0033] Figure 7 This is a structural block diagram of the traffic phase execution module in an embodiment of the intelligent cycle and intelligent phase sequence traffic signal phase control system based on the edge end of the present invention.

[0034] Figure 8 This is a flowchart of the traffic signal phase control method based on edge-end intelligent cycle and intelligent phase sequence of the present invention.

[0035] The present invention will be further described below with reference to the accompanying drawings and embodiments. Detailed Implementation

[0036] This invention uses cameras installed at intersections to capture real-time images of road users, and then identifies specific users, such as vehicles and pedestrians in lanes, using a deep learning-based target recognition algorithm. Furthermore, it constructs traffic signal phase control and scheduling models at the edge, achieving adaptive control of intelligent traffic. This invention employs two modes for intelligent control: intelligent cycle mode and intelligent phase sequence mode. Through this method, it can effectively improve traffic efficiency at intersections, reduce congestion, and minimize traffic phase idling, solving the problem that existing traffic control systems cannot achieve adaptive and precise control of intersection traffic signals.

[0037] Example of a traffic signal phase control system based on edge-based intelligent cycle and intelligent phase sequence:

[0038] The intelligent periodicity and intelligent phase sequence traffic signal phase control system based on the edge terminal of this invention can intelligently adjust both the periodicity of the phases and the phase sequence of each phase. This system is applied to edge terminal devices. Specifically, the intelligent traffic control system includes a cloud server, edge terminal devices, and terminal devices. One edge terminal device is installed at an intersection, and an intersection includes multiple directions, each direction having multiple lanes, such as... Figure 1 As shown. Typically, different lanes have different traffic directions. For example, if there are four lanes in one direction, the leftmost lane is the left-turn lane, the two middle lanes are straight-ahead lanes, and the rightmost lane is the right-turn lane.

[0039] For each direction, at least one set of traffic lights is required. Multiple sets of traffic lights at an intersection are typically controlled by a single traffic signal controller. Typically, the terminal equipment includes the traffic signal controller and cameras installed at the intersection. This embodiment is implemented using edge devices. These edge devices can communicate with a cloud server to obtain traffic configuration models, and they also communicate with the terminal equipment, such as acquiring runtime data from the traffic signal controller and sending control commands to it.

[0040] See Figure 2 This embodiment includes a traffic flow plan scheduler 100, a video acquisition and recognition module 200, a real-time traffic condition acquisition module 300, a phase timing module 400, and a traffic phase execution module 500. The traffic flow plan dispatcher 100 includes system configuration parameters 101, a traffic flow plan worksheet 102, a traffic flow plan time period table 103, and a dispatch engine 104. The video acquisition and recognition module 200 includes an intersection traffic camera visual acquisition module 201, a visual AI inference module 202, and a structured vehicle data module 203. The real-time traffic condition acquisition module 300 includes an intersection real-time vehicle data collector 301, a traffic signal running information collector 302, and a real-time traffic condition model synthesizer 303. The phase timing module 400 includes a real-time vehicle acquisition device 401, a vehicle data benchmarking module 402, a phase vehicle standardization module 403, and a phase timing calculation table 404. The traffic phase execution module 500 includes a current running phase context 501, a phase ready waiting queue 502, a running phase queue 503, a phase ready waiting queue manager 504, a phase execution command mapper 505, and a traffic signal command executor 506.

[0041] The traffic flow scheduling system 100 is used for unified management of the video acquisition and recognition module 200, the real-time traffic condition acquisition module 300, the phase timing module 400, and the traffic phase execution module 500. (See also...) Figure 3After the traffic flow scheduler 100 starts, it first executes step S1 to load the configuration file and initialize environment variables, thus forming system configuration parameters 101. Then, it executes step S2 to load the traffic flow plan worksheet 102, followed by step S3 to load the traffic flow plan time slot table 103. Finally, it executes step S4 to switch to the current traffic flow plan through the scheduling engine 104 and start each module, sending signals to the video acquisition and recognition module 200, the real-time traffic condition acquisition module 300, the phase timing module 400, and the traffic phase execution module 500. After the real-time traffic condition acquisition module 300 starts running, it communicates with the cloud server 110, for example, obtaining the traffic timing model from the cloud server 110 and returning the real-time traffic condition model to the cloud server 110. The traffic phase execution module 500 sends signals to the traffic signal controller 120 to control the operation of the traffic signal controller 120.

[0042] See Figure 4 The intersection traffic camera visual acquisition module 201 of the video acquisition and recognition module 200 includes multiple cameras, such as cameras 211, 212, 213, and 214. Of course, depending on actual needs, the minimum number of cameras can be one; if there are many lanes, more cameras can be set. Data from multiple cameras is transmitted to the AI ​​vision access configurator 221, configured, and then output to the AI ​​vision acquisition scheduler 222. In this embodiment, the visual AI inference module 202 includes an AI vision inferencer 223, used to infer and identify vehicles based on the data collected by each camera. The AI ​​vision callback processor 224 can retrieve vehicles from the AI ​​vision count.

[0043] See Figure 5 The intersection real-time vehicle data collector 301 of the real-time traffic condition acquisition module 300 can communicate with the vision acquisition and recognition module 200, for example, to obtain structured vehicle data from the vision acquisition and recognition module 200. The signal controller operation information collector 302 acquires real-time data from the signal controller 120. The traffic real-time condition model synthesizer 303 simultaneously acquires the real-time traffic condition data from the intersection real-time vehicle data collector 301 and the real-time information of the signal controller 120 output by the signal controller operation information collector 302, aligns the real-time traffic condition data and the signal controller operation information according to time, and merges them to obtain a traffic real-time condition model.

[0044] See Figure 6The real-time vehicle acquisition unit 401 of the phase timing module 400 communicates with the visual acquisition and recognition module 200 to acquire real-time vehicle data, such as structured vehicle data, and sends the structured vehicle data to the vehicle data benchmarking module 402. The vehicle data benchmarking module 402 performs benchmarking calculations on the vehicle data of the lanes. Then, the phase vehicle standardization module 403 performs standardization calculations on the vehicles in each phase. Finally, the phase timing calculation table 404 calculates the timing duration of each phase based on the standardization results and outputs the timing duration of each phase to the traffic phase execution module 500, which converts it into instructions that the traffic signal controller 120 can recognize and execute.

[0045] See Figure 7 The traffic phase execution module 500 retrieves the phase data to be executed from the phase ready waiting queue 502 in its current running phase context 501, and records the phase data that has been executed to the running phase queue 503. The phase ready waiting queue manager 504 has two modes: intelligent cycle mode and intelligent phase sequence mode. In intelligent cycle mode, the phase sequence is fixed, meaning each phase is executed in a fixed order, but the running cycle and duration of each phase are intelligently adjusted. In intelligent phase sequence mode, the phase sequence is not fixed; the execution order of each phase is dynamically adjusted based on the vehicle data of each phase. For example, if there are few vehicles in a phase, that phase is skipped in one cycle; or if there are many vehicles in a phase, that phase is executed twice consecutively. Preferably, the intelligent phase sequence mode adjusts the initial phase sequence 521 of the traffic plan.

[0046] Therefore, in the intelligent cycle mode, a phase sequence selector 510 and a phase timing unit 520 are provided. The phase sequence selector 510 selects the next phase to be executed based on the currently executed phase and a pre-set fixed phase sequence, while the phase timing unit 520 dynamically calculates the duration of each phase to be executed.

[0047] In intelligent phase sequence mode, a line-plane priority control parameter 511, a priority calculator 512, a fairness calculator 513, a running phase queue 503, and a phase timing unit 520 are set. The priority calculator 512 uses the line-plane priority control parameter 511 and the running phase queue 503 to calculate the priority level of each phase. The fairness calculator 513 is used to adjust the calculated priority level of each phase. At the same time, the phase timing unit 520 calculates the timing of each phase.

[0048] In addition, the current running context 501 is also sent to the phase execution instruction mapper 505, and then sent to the signal 120 via the signal instruction executor 506.

[0049] Example of a traffic signal phase control method based on edge-based intelligent cycle and intelligent phase sequence:

[0050] The following is combined with Figure 8 The workflow of the edge-based intelligent cycle and intelligent phase sequence traffic signal phase control system is described. After the traffic plan scheduler 100 sends a start command to the video acquisition and recognition module 200, the real-time traffic condition acquisition module 300, the phase timing module 400, and the traffic phase execution module 500, the video acquisition and recognition module 200 first executes step S11, using multiple cameras to acquire vehicle video information from multiple lanes at the intersection, and using a pre-set AI target recognition model to identify the structured vehicle information of each lane. Specifically, after acquiring the vehicle video information, the video acquisition and recognition module 200 identifies the vehicles in the video signal. Structured lane recognition identifies the type of vehicle in each lane, such as cars, trucks, buses, or large trucks, and also identifies the number of vehicles in each lane and the length of the vehicle queue.

[0051] Then, the real-time traffic condition acquisition module 300 executes step S12, calling the real-time vehicle data of the intersection acquired by the visual acquisition and recognition module 200, and acquiring the traffic signal operation information. The real-time vehicle data and traffic signal operation information are aligned according to time and merged to obtain the real-time traffic condition model. Specifically, the real-time vehicle data of the intersection has a timestamp, and the traffic signal operation information also has a timestamp. Aligning the timestamps of the real-time vehicle data and the traffic signal operation information achieves time synchronization between them. The synchronized data is the real-time traffic condition model.

[0052] Next, the phase timing module 400 executes step S13, calling the real-time vehicle data of the lanes acquired by the real-time traffic condition acquisition module 300, and performs benchmarking and standardization processing on the real-time vehicle data of the lanes, and applies the phase timing calculation table 404 to perform timing on the corresponding phases. When performing benchmarking and standardization processing on the real-time vehicle data of the lanes, the vehicle data is first benchmarked, and then the standardized vehicles of the lanes are performed using the benchmarked data.

[0053] Specifically, when standardizing vehicle data, the various types of vehicles within a lane are converted into a number based on cars. For example, using cars as the base, trucks, buses, and large trucks are converted into cars; one truck is converted into 1.5 cars, one bus into 2 cars, and one large truck into 2.5 cars. Then, the number of cars in that phase is divided by the number of lanes in that phase to obtain the standardization result.

[0054] When performing standardized calculations, it is necessary to consider whether a phase has multiple inputs. For example... Figure 1 As shown in the intersection model, assuming that for the north phase, if there is only a south entrance, that is, only the case of vehicles traveling from south to north entering the north phase, then the result of the standardized calculation of the north phase is the number of north phases based on cars divided by the number of lanes in the north phase.

[0055] If a phase has more than two entrances, for example, a northern phase has both southbound through traffic and westbound left-turning traffic, then that phase has two distinct entrances: a southern entrance and a western entrance. In this case, the standardized entrance result for each entrance needs to be calculated separately, and the maximum standardized entrance result among all entrances is taken as the standardized result for that phase. The formula is as follows: Standardized Data = max(Number of all benchmarked vehicles corresponding to different entrances / Number of lanes in the phase), where "Number of all benchmarked vehicles corresponding to an entrance / Number of lanes in the phase" represents the standardized entrance result for one entrance.

[0056] Next, step S14 is executed, applying the phase timing calculation table to calculate the timing for the corresponding phases. For example, first, the phase timing ratio of the reference period is calculated, then the proportion of vehicles corresponding to the phase is calculated, and finally, the final phase timing is calculated using the phase timing ratio of the reference period and the proportion of vehicles corresponding to the phase. The reference period is the average value obtained using a moving average model. Specifically, when calculating the phase timing ratio of the reference period, the average duration and benchmarked vehicle count of multiple periods preceding the current period are used to calculate the duration and benchmarked vehicle count of the current period. For example, Ft is the predicted value of the current period to be executed; this predicted value can be the timing time of each phase or the benchmarked vehicle count of that phase.

[0057] The predicted value Ft can be calculated using the following formula:

[0058] Ft = (w1*At-1 + w2*At-2 + w3*At-3 + ... + wn*At-n) / n, where n is the number of periods preceding the current period, At-1 is the actual value of the previous period, At-2 is the actual value of the previous two periods, At-3 is the actual value of the previous three periods, and so on; w1, w2, w3...wn are pre-set weighting coefficients. Therefore, the predicted value Ft is the moving weighted average of the actual values ​​of the previous n periods.

[0059] Since the value of the reference period is obtained by calculating the moving weighted average of the data from the previous n periods, the value of the reference period becomes increasingly accurate. Therefore, the purpose of setting a reference period is to use its value as an anchoring reference parameter in calculating the intelligent period and intelligent phase sequence algorithm, giving the reference period two attributes: reference period time and reference number of vehicles.

[0060] The proportion of vehicles corresponding to a phase is the standardized number of vehicles corresponding to a certain phase divided by the standardized number of vehicles corresponding to all phases, expressed by the following formula: Proportion of vehicles corresponding to a phase = Standardized number of vehicles corresponding to a certain phase / Standardized number of vehicles corresponding to all phases.

[0061] When calculating the final phase timing, it is necessary to calculate the average of the phase timing ratio of the base cycle and the ratio of the number of vehicles corresponding to the phase. Multiplying this average by the total duration of each phase cycle yields the final phase timing, expressed by the following formula: Final phase timing = Total cycle duration × ((Phase timing ratio of the base cycle + Ratio of the number of vehicles corresponding to the phase) / 2). Of course, for ease of control, the calculation result needs to be rounded to ensure that the final phase timing is an integer number of seconds.

[0062] For example, if the total duration of the reference period is 120 seconds, the phase timing calculation table is as follows:

[0063]

[0064] After calculating the final timing of each phase, the traffic phase execution module 500 executes step S15, and sets the timing control of each phase according to the timing results of each phase calculated by the phase timing module 404 in step S14.

[0065] In this embodiment, two operating modes are set. In the intelligent cycle mode, the operation follows a fixed phase sequence, and the timing of each phase is calculated according to the timing duration calculated in step S14. In the intelligent phase sequence mode, the priority level of each phase needs to be calculated, and the execution order of each phase is intelligently adjusted according to the calculation results.

[0066] Therefore, in step S15, the traffic phase execution module 500 also needs to adjust the order of each phase according to the real-time vehicle data of each phase. First, it calculates the first proportion of the standardized result corresponding to each phase to the standardized result corresponding to all phases, which is expressed by the following formula: First proportion = Standardized number of traffic participants corresponding to each phase / Standardized number of traffic participants corresponding to all phases.

[0067] Next, it is necessary to calculate the second ratio of the standardized frequency of each phase in the preset number of completed cycles to the standardized frequency of all phases in the preset number of completed cycles. Specifically, obtain the frequency of each phase in the preset number of completed cycles and calculate the standardized frequency of each phase. The standardized frequency of a phase is the frequency of that phase multiplied by a set exponent of 2, where the exponent is the difference between the phase allocation fairness and the number of completed cycles. The formula is as follows: Standardized frequency of a phase = Frequency of that phase × (2^(Phase allocation fairness - Number of completed cycles)). Finally, calculate the second ratio, which is the ratio of the standardized frequency of each phase to the standardized frequency of all phases. The formula is as follows: Second ratio = Standardized frequency of that phase / Standardized frequency of all phases. The phase allocation fairness can be a preset value or calculated based on multiple factors such as whether the phase corresponds to a main road and the current number of lanes.

[0068] After calculating the first and second ratios, it is necessary to calculate the difference between the first and second ratios for each phase. The priority level of each phase is determined based on this difference; that is, the first ratio of a phase is subtracted from its second ratio to calculate the difference. This difference then determines the priority level and order of each phase. Specifically, the priority level of a phase is directly related to the difference between its first and second ratios. In other words, the larger the difference between the first and second ratios for a phase, the higher its priority. Thus, by ranking the phases according to the differences between their first and second ratios, the passage sequence can be obtained. Preferably, according to the scheme of this embodiment, if a phase has no vehicles, it can be skipped, meaning there will be no passage time for that phase, thus avoiding wasted time. Of course, the priority of each phase also needs to be fine-tuned. For example, assuming that the number of cycles during which the right of way is allowed to be lost is n, if a phase does not gain the right of way in n consecutive cycles, it will definitely gain the right of way at the beginning of the next cycle, so as to reflect the fairness of the passage of each phase.

[0069] Finally, step S16 is executed, the control signal of the signal machine 120 is determined according to the determined phase sequence and the timing of each phase, and the control signal is sent to the signal machine 120 so that the signal machine 120 can control the operation of each light group.

[0070] As can be seen, this invention provides both an intelligent cycle mode and an intelligent phase sequence mode. In the intelligent cycle mode, the phase sequence is fixed, the phase cycle is variable, and the travel time for each phase is adaptive. The travel time for each phase increases with increasing traffic flow and decreases with decreasing traffic flow. This mode is particularly suitable for phase control under tidal traffic flow conditions, and also applicable to controlling traffic flow at entrances where there are no clear patterns, with periods of heavy traffic and periods of light traffic.

[0071] In the intelligent phase sequence mode, the phase sequence is variable. If there are no waiting vehicles in the lane corresponding to a certain phase, the corresponding phase will be skipped, reducing the phase's idle time. The phase passage time is adaptive, increasing with increasing traffic flow and decreasing with decreasing traffic flow. This mode is suitable for intersections where the arrival of vehicles at each entrance is highly random, traffic flow in each direction varies relatively greatly and irregularly, making timed control difficult, or where the volume of motor vehicles, non-motor vehicles, and pedestrians is low on lower-grade roads.

[0072] The method of this invention can respond promptly to random changes in traffic flow, has good control effect, can effectively improve the operating efficiency of intersections and reduce congestion, and can solve the problem that existing traffic control systems cannot perform online real-time precise control of intersection traffic signals.

[0073] Finally, it should be emphasized that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A traffic signal phase control method based on intelligent phase sequence, characterized in that, include: The visual acquisition and recognition module collects vehicle video information from multiple lanes at the intersection and uses an AI target recognition model to identify the structured vehicle information of each lane. The real-time traffic condition acquisition module calls the real-time vehicle data of the intersection obtained by the vision acquisition and recognition module, and obtains the signal controller operation information. The real-time vehicle data and the signal controller operation information are aligned according to time and merged to obtain a real-time traffic condition model. The phase timing module calls the real-time vehicle data of the lane obtained by the real-time traffic condition acquisition module, performs benchmarking and standardization processing on the real-time vehicle data of the lane, and applies the phase timing calculation table to time the corresponding phases. The traffic phase execution module adjusts the order of each phase based on real-time vehicle data of each phase, and sets the timing control of each phase based on the timing results of each phase calculated by the phase timing module, and sends control signals to the traffic signal. The traffic phase execution module adjusts the order of each phase based on real-time vehicle data, including: Calculate the first proportion of the standardized results corresponding to each phase to the standardized results corresponding to all phases; Calculate the second ratio of the normalized number of occurrences of each phase in the preset number of completed cycles to the normalized number of occurrences of all phases in the preset number of completed cycles; Calculate the difference between the first ratio and the second ratio for each phase, and determine the priority level of each phase based on the difference between the first ratio and the second ratio; The order of each phase is determined according to its priority level; If there are no vehicles in a certain phase, that phase is skipped.

2. The traffic signal phase control method based on intelligent phase sequence according to claim 1, characterized in that: The calculation of the second ratio of the normalized number of occurrences of each phase in the preset number of completed cycles to the normalized number of occurrences of all phases in the preset number of completed cycles includes: Obtain the number of times each phase appears in the preset number of completed cycles, and calculate the standardized number of each phase: the standardized number of a phase is the number of times the phase appears multiplied by 2 raised to a predetermined exponent, where the predetermined exponent is the difference between the phase allocation fairness and the number of completed cycles. The second ratio is the ratio of the normalization degree of each phase to the normalization degree of all phases.

3. The traffic signal phase control method based on intelligent phase sequence according to claim 1, characterized in that: The priority level of each phase is determined based on the difference between the first ratio and the second ratio, including: The priority of a phase is directly related to the difference between the first ratio and the second ratio of that phase.

4. The traffic signal phase control method based on intelligent phase sequence according to any one of claims 1 to 3, characterized in that: Applying the phase timing calculation table to time the corresponding phases includes: The phase timing is calculated based on the phase timing ratio of the reference cycle and the vehicle number ratio corresponding to the phase. The final phase timing is then calculated based on the phase timing ratio of the reference cycle and the vehicle number ratio corresponding to the phase. The value of the reference period is the sliding weighted average of a preset number of periods prior to the current period; The calculation of the final phase timing based on the phase timing ratio of the reference cycle and the proportion of vehicles corresponding to the phase includes: calculating the average of the phase timing ratio of the reference cycle and the proportion of vehicles corresponding to the phase, and multiplying the average by the total duration of each phase cycle to obtain the final phase timing.

5. A traffic signal phase control system based on intelligent phase sequence, characterized in that, It includes a visual acquisition and recognition module, a real-time traffic condition acquisition module, a phase timing module, and a traffic phase execution module; The visual acquisition and recognition module acquires vehicle video information from multiple lanes at the intersection and uses an AI target recognition model to identify the structured vehicle information of each lane. The real-time traffic condition acquisition module calls the real-time vehicle data of the intersection obtained by the visual acquisition and recognition module, and obtains the traffic signal operation information. The real-time vehicle data and the traffic signal operation information are aligned according to time and merged to obtain a real-time traffic condition model. The phase timing module calls the real-time vehicle data of the lane obtained by the real-time traffic condition acquisition module, performs benchmarking and standardization processing on the real-time vehicle data of the lane, and applies the phase timing calculation table to time the corresponding phases. The traffic phase execution module adjusts the sequence of each phase based on real-time vehicle data for each phase, and sets the timing control for each phase according to the timing results calculated by the phase timing module. It then sends control signals to the traffic signal controller to adjust the sequence of each phase. Calculate the first proportion of the standardized results corresponding to each phase to the standardized results corresponding to all phases; Calculate the second ratio of the normalized number of occurrences of each phase in the preset number of completed cycles to the normalized number of occurrences of all phases in the preset number of completed cycles; Calculate the difference between the first ratio and the second ratio for each phase, and determine the priority level of each phase based on the difference between the first ratio and the second ratio; The order of each phase is determined based on its priority level.

Images