Pedestrian and vehicle demand dynamic response traffic signal control system and control method

Abstract

The invention provides a pedestrian and vehicle demand dynamic response traffic signal control system and control method. The system integrates a plurality of components such as a signal lamp unit, a detector unit, a two-dimensional code, a vehicle-mounted unit and a road test unit. The control method comprises the following steps: receiving data uploaded by a pedestrian activation button, various detectors and a two-dimensional code and real-time traffic information provided by the vehicle-mounted unit and the road test unit through the signal processor, and dynamically adjusting a time sequence plan of the signal lamp by using advanced technologies such as deep learning, reinforcement learning and long and short term memory models. Aiming at a single intersection, signal lamp timing is optimized by adopting a reinforcement learning algorithm to reduce pedestrian and vehicle conflicts, and traffic flow is monitored in real time through a vehicle queue model; and in the multi-intersection network, the LSTM model is utilized to predict the traffic flow, and the influence of external factors is considered through multi-modal data fusion, so that the coordination control among the signal lamps is realized, and the traffic efficiency of the whole traffic network is optimized.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent transportation systems, and in particular relates to a traffic signal control system and a control method that dynamically responds to pedestrian and vehicle demands. Background Art

[0002] Pedestrian crossing signals are a specialized subset of traffic signals that are designed specifically to guide pedestrians across the road safely. Pedestrian crossing signals generally include two basic forms: fixed pedestrian crossing signals and countdown pedestrian crossing signals. In addition, there are special types that adapt to specific situations, including voice-activated or touch-activated pedestrian crossing signals, flashing warning lights, etc. Despite the advances in the design of these pedestrian crossing signals, there are still considerable design limitations in pedestrian crossing intelligence, especially in managing complex and fluctuating pedestrian traffic. Existing signal systems often fail to meet the needs of pedestrians. Fixed-timing (timed signal control) and button-activated pedestrian crossing systems are quite rigid and cannot dynamically meet individual pedestrians or provide personalized signal timing solutions. This highlights the need for more adaptive and intelligent systems that can better cope with changing pedestrian traffic conditions.

[0003] At present, the pedestrian crossing systems on existing roads face the following problems: (1) The design of road crosswalk systems often prioritizes vehicle traffic efficiency. Some designs focus mainly on pedestrian safety, but rarely achieve the best balance between efficiency and safety; (2) The data collection methods and data types used by crosswalk systems are limited, resulting in great inaccuracy in information collection; these systems are less intelligent and mainly use timing signals and push-button signals, neither of which can dynamically adjust the green light duration according to pedestrian flow, resulting in low pedestrian green light time utilization; these systems are less intelligent and mainly use timing signals and push-button signals, neither of which can dynamically adjust the green light duration according to pedestrian flow, resulting in low pedestrian green light time utilization.

[0004] Traditional systems have limitations in responding to instantaneous traffic changes and pedestrian needs, resulting in long waiting times, low traffic efficiency and increased safety hazards. Therefore, there is an urgent need for a dynamic traffic light system that can dynamically respond to pedestrian and vehicle crossing needs. By real-time monitoring and analysis of pedestrian and vehicle flow changes, the switching time and sequence of traffic lights can be dynamically adjusted to provide more flexible and efficient traffic management, improve the convenience and safety of pedestrians crossing the street, and optimize the overall traffic flow. Summary of the invention

[0005] The purpose of the embodiments of the present invention is to provide a traffic signal control system that dynamically responds to pedestrian and vehicle demands, thereby improving pedestrian safety and convenience while improving overall traffic efficiency by responding to pedestrian and vehicle crossing demands.

[0006] A second objective of an embodiment of the present invention is to provide a traffic signal control method that dynamically responds to pedestrian and vehicle demands.

[0007] In order to solve the above technical problems, the technical solution adopted by the present invention is a traffic signal control system that dynamically responds to pedestrian and vehicle demands, specifically comprising:

[0008] A signal light unit, the signal light unit comprising a pedestrian activation button, a signal light and a signal processor; the signal light unit is connected to the detector unit by signal; the pedestrian activation button is connected to the signal processor for responding to pedestrian crossing requirements, and the signal processor is used to process received signals and control the state of the signal light according to the signals;

[0009] The detector unit includes a camera, a video processor, an infrared detector, a speed detector and a voice detector; the detector unit is used to feed back the movement information of pedestrians and vehicles to the signal processor in the form of signals;

[0010] A QR code, the encoding of which contains a URL for sending a signal to a signal processor; used to upload pedestrian crossing requirements to the signal processor;

[0011] The on-board unit is mounted on the vehicle; the on-board unit is connected to the signal processor of the signal light unit via the Internet; the on-board unit detects the real-time status and driving information of the vehicle and sends it to the signal light unit, and the on-board unit is also connected to the road test unit by signal;

[0012] The road test unit includes a road test camera and an image recognition module, which is connected to the signal processor signal. The image recognition module of the road test unit matches the vehicle image information collected by the road test camera with the on-board unit, and sends the matching information and the received real-time traffic data to the signal processor.

[0013] Furthermore, the camera is used to collect images of pedestrians and vehicles, the video processor is used to process the images, the infrared detector, speed detector and voice detector collect the pedestrian's position, speed and sound, and upload them to the signal processor, which determines the pedestrian's status information.

[0014] A control method for a traffic signal control system that dynamically responds to pedestrian and vehicle demand, the process is as follows:

[0015] The signal processor receives information from the pedestrian activation button, the detector unit, the QR code, the vehicle-mounted unit, and the road test unit and controls the signal light;

[0016] The on-board unit collects vehicle driving information, the road test unit identifies the vehicle, matches the vehicle information with the on-board unit, and sends the matching result to the signal light unit.

[0017] Furthermore, the signal processor controls the traffic light in the following specific steps: receiving a signal from a pedestrian activation button or a crossing demand signal obtained by a pedestrian scanning a QR code, obtaining the pedestrian crossing speed through data collected by a camera and a speed detector, and calculating the time required for each pedestrian to pass through the intersection, and adjusting the switching time of the traffic light based on the calculation result:

[0018] T p =T s +L / V p

[0019] Among them, T p represents the shortest time required for a single pedestrian to pass through the intersection, V p is the speed limit for pedestrians to cross the intersection, L is the length of the intersection, T s Indicates the initial hesitation time of the pedestrian;

[0020] For users who cannot scan the QR code, use the button to activate.

[0021] Furthermore, the control process of the traffic light by the vehicle-mounted unit and the road test unit is specifically as follows:

[0022] A. The on-board unit collects vehicle information; wherein the on-board unit can monitor the vehicle startup process, and the vehicle information includes the vehicle startup delay time;

[0023] B. The on-board unit and the road test unit match the vehicle status information and send it to the signal processor of the signal light unit for processing:

[0024] When the vehicle is at the signal light that allows vehicles to pass, the vehicle maintains the speed through the intersection, and the road test unit measures the instantaneous speed V of the vehicle at the stop line. j2 , and calculate the time T required to pass the intersection v :

[0025] T v =L / V j2 +T c

[0026] T c =L c / V j2

[0027] Where L is the length of the intersection, T c L is the extra time required for a vehicle to fully pass through the intersection based on its length. c is the vehicle length;

[0028] When the vehicle is not within the allowed passing time, it must stop. The road test unit calculation includes the vehicle start delay time T qThe vehicle passing time T v ':

[0029] T v '=L / V j2 +T c +T q

[0030] C. The road test unit uploads the traffic volume passing through the intersection to the signal processor, together with the number of remaining vehicles in each lane, to recalculate the green light duration of the upcoming signal cycle; the road test unit also uploads the last vehicle information passing through each lane, and the signal processor then fine-tunes the signal light;

[0031] D. After the road test unit detects a special vehicle, it uploads the signal to the signal processor, which controls the traffic light to release the vehicle.

[0032] Furthermore, the signal processor includes a coordination control module, and the coordination control module specifically includes a single-intersection signal light coordination unit and a multi-intersection signal light coordination unit.

[0033] Furthermore, the control method of the single intersection and multi-intersection signal light coordination unit is specifically as follows:

[0034] S1. In each signal cycle, the detector unit and the road test unit are used to collect traffic volume data passing through the intersection; the data is uploaded to the signal processor, including the number of vehicles passing through each lane in the intersection and the number of remaining vehicles;

[0035] S2, using sensor units to collect pedestrian crossing data;

[0036] S3, use historical data to train reinforcement learning models, LSTM models and data fusion models to optimize signal cycles and green light time;

[0037] S4, calculating the optimized signal cycle and travel time;

[0038] S5. Based on the shortest time required for a single pedestrian to pass through the intersection, the vehicle start delay time, the time required for vehicles to pass through the intersection and the real-time traffic conditions, at the end of each signal cycle, the signal cycle and travel time are controlled according to the feedback of pedestrian and vehicle crossing effects and the results of S3 to S4.

[0039] Furthermore, the specific steps of S3 are as follows:

[0040] S301, Adaptive control based on reinforcement learning:

[0041] Define the intersection state:

[0042] S={T i ,Pi ,E i}

[0043] Among them, S represents the state of the intersection, T i represents the real-time traffic flow of the lane, P i represents the current pedestrian crossing demand, E i Indicates external influencing factors;

[0044] Define the control actions for signal cycle and lane green light time: A:

[0045] A={G i ,C i}

[0046] Among them, G i Indicates the green light time of the lane; C i Indicates the signal period;

[0047] Define the reward function R: R = -(α×vehicle waiting time + β×pedestrian waiting time + γ×traffic flow fluctuation)

[0048] Among them, α, β, and γ represent weight parameters;

[0049] Update the policy using Q-learning:

[0050] Among them, η1 is the learning rate; η2 is the discount factor; S' and A' represent the updated state and action of the next step after taking the original control action; η′1 represents the learning rate of the next simulation training after the update; max is the maximum value function;

[0051] S302. Traffic flow prediction model based on LSTM:

[0052] represents the predicted traffic flow at the next time step; H i Represents historical traffic flow; LSTM() represents the output result through the LSTM model;

[0053] S304, define pedestrian priority mechanism: G p,i =max(G min ,min(G max ,k p P i ))

[0054] Among them, G p,i represents the pedestrian green light time of the i-th lane; k p Indicates the ratio coefficient between pedestrian flow and green light time; G max , G min Indicates the upper and lower limits of the green light time.

[0055] Furthermore, the specific steps of S4 are as follows:

[0056] S401. Calculation of optimized signal cycle and travel time:

[0057]

[0058] Among them, λ is the weight coefficient of pedestrian flow; C' represents the optimized signal period; L oss Indicates lost time; Y i It represents the ratio of the saturation flow rate to the actual flow value of the i-th lane;

[0059] S402. In the multi-intersection signal coordination, the time difference ΔT′ between adjacent intersections is:

[0060]

[0061] Among them, δ is the weight coefficient of pedestrian flow change; d represents the distance between adjacent intersections; v represents the traffic flow speed; P i represents the current pedestrian crossing demand, P i-1 Indicates the pedestrian crossing demand at the previous intersection.

[0062] The beneficial effects of the present invention are:

[0063] The present invention meets the real-time needs of pedestrians and vehicles by dynamically adjusting the timing of traffic lights, thereby improving the safety and convenience of pedestrians crossing the street. The system can monitor and analyze pedestrian and vehicle flow data in real time, and optimize the switching time and sequence of traffic lights accordingly, thereby significantly improving the operating efficiency of the entire traffic system. In addition, the present invention also has the ability to self-adjust according to the immediate needs of pedestrians and vehicles, effectively solving the problem that traditional systems cannot flexibly respond to changes in individual needs. DETAILED DESCRIPTION

[0064] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0065] This embodiment provides a traffic signal control system that dynamically responds to pedestrian and vehicle demand. It meets the pedestrian and vehicle crossing demands by adaptively responding to them, and uses the effective green light time to reversely calculate the signal cycle, thereby improving the convenience and safety of pedestrians crossing the street and optimizing the overall traffic flow. Demand response refers to the behavior of the demand side or end consumers changing their short-term consumption methods (consumption time or consumption level) and long-term consumption patterns by responding to market-based price signals, incentives, or direct instructions from system operators. This embodiment introduces a demand response strategy into the traffic signal control system to solve the problem that existing traffic control strategies are difficult to cope with instantaneous traffic changes.

[0066] In some specific embodiments, the traffic signal control system that dynamically responds to pedestrian and vehicle demand includes a traffic light unit; the traffic light unit includes a pedestrian activation button, a traffic light, and a signal processor; after the pedestrian presses the button, the traffic light and the signal processor receive a pedestrian crossing demand signal; the traffic light changes, and the signal processor dynamically adjusts the green light time according to pedestrian demand to ensure that pedestrians cross the street safely; and the pedestrian crossing process is monitored through the detector unit to ensure that the traffic light responds accurately.

[0067] The detector unit includes a camera, a video processor, an infrared detector and a voice detector. The video processor can perform image recognition on the video captured by the camera. The detector module captures the video through the camera, and the video processor performs image recognition and transmits the pedestrian information to the signal light unit. The infrared detector and the voice detector also monitor the movement of pedestrians and vehicles in real time and feed the data back to the signal light unit. The detector module can capture real-time video and perform image recognition to detect the presence of pedestrians and their movement, thereby providing accurate pedestrian crossing data. Through the data collected by devices such as video, infrared position signals, speed, and voice detectors, the signal processor can identify the number, category (such as the elderly, children, disabled people, etc.), and crossing speed of pedestrians. The signal processor uses this data for analysis, calculates the time required for each pedestrian to pass through the intersection, and further combines the vehicle driving information (driving speed, body length, etc.) to adjust the switching time of the signal light. Based on these data, the system continuously optimizes the traffic light control strategy through a reinforcement learning model. It can adaptively adjust the switching time of traffic lights according to actual traffic conditions, and apply a multimodal data fusion model to improve the accuracy of data analysis, so that the system can operate effectively in complex traffic environments, enhance the intelligence level of signal control, and better respond to complex traffic conditions and pedestrian crossing needs.

[0068] The main function of infrared detectors and voice detectors is to dynamically adjust the switching time and sequence of traffic lights according to the real-time needs of pedestrians and vehicles to optimize traffic flow and improve safety. Each device interacts with the traffic light control system through real-time data capture and processing to ensure that the changes in traffic lights can respond to the needs of pedestrians and vehicles in a timely manner.

[0069] The speed detector detects the speed of pedestrians crossing the street and uploads it to the signal processor.

[0070] A QR code, which contains a URL for sending signals to the traffic light unit. When a user scans the QR code with a smart device (such as a smartphone), the application on the device will parse the information in the QR code and send a demand signal (such as the time it takes to cross the street, the number of people crossing the street, etc.) to the traffic light unit via the Internet. After receiving the signal from the server or directly from the user's device, the traffic light unit adjusts the working state of the traffic light according to the signal content.

[0071] This embodiment uses the above device to identify the number and type of pedestrians, introduces QR code scanning, analyzes the user through the signal processor, and calculates the time required for each pedestrian to pass through the intersection. The signal processor interacts with the detector unit and other devices, receives real-time data, and performs calculations and adjustments based on the data to achieve dynamic adjustment of the traffic light.

[0072] In some specific implementations, the signal processor receives a signal from a pedestrian activation button or a crossing demand signal obtained by a pedestrian scanning a QR code; then obtains the pedestrian crossing speed through data collected by a camera and a speed detector, and calculates the time required for each pedestrian to pass through the intersection; and at the same time, adjusts the switching time of the signal light based on the calculation result:

[0073] T p =T s +L / V p

[0074] Among them, T p represents the shortest time required for a single pedestrian to pass through the intersection, V p is the speed at which pedestrians are allowed to cross an intersection, determined based on the average speed observed for that pedestrian at various intersections. This speed is determined using cameras or other roadside speed measurement devices installed. Each time a pedestrian crosses an intersection, these systems record their speed to accurately estimate the time it takes for pedestrians to cross the intersection, thereby minimizing errors in calculating the time attributed to their movement. In addition, the camera monitors the pedestrian's initial hesitation time, T s . L represents the length of the intersection.

[0075] For pedestrians who are unable to scan the QR code, such as elderly people who cannot use smartphones, children without smartphones, or visually impaired people who cannot locate the scanning point, hardware assistance in the form of button-activated signal lights is used, supplemented by detector units for monitoring and protection.

[0076] In some possible implementations, the traffic signal control system that dynamically responds to pedestrian and vehicle demand also includes an on-board unit and a roadside unit; the roadside unit transmits data in real time with the on-board units and signal light units carried by other vehicles through wireless communication technology (such as 5G) and the Internet. The roadside unit includes a road test camera and an image recognition module, which can identify the vehicle's position, length, speed, driving direction, and number; the received real-time traffic data (such as vehicle position, length, speed, driving direction, etc.) is sent to the signal processor for dynamic adjustment of traffic lights to optimize traffic flow, reduce congestion, and improve road safety. This real-time data transmission and processing mechanism enables the signal processor to make timely and intelligent adjustments based on current traffic conditions, thereby better managing urban traffic.

[0077] The detector unit, vehicle-mounted unit, and roadside unit work together. First, the detector unit and road test unit detect pedestrians and vehicles to determine the traffic needs of individual vehicles and form the basis for signal timing information. The road test unit can accurately identify the number, location, and movement trajectory of vehicles; the vehicle-mounted unit detects the real-time status and driving information of the vehicle; and the roadside unit collects and transmits this data to the signal light unit through wireless communication technology. Based on the above information, the signal light unit can intelligently analyze and adjust the switching time of traffic lights to meet the traffic needs of different vehicles and pedestrians, and improve the efficiency and safety of traffic management.

[0078] The on-board unit is connected to the signal processor signal of the signal light.

[0079] In some specific implementations, the control method of the vehicle-mounted unit is:

[0080] A. Analysis of user demand suggestion method: Use the vehicle-mounted unit to collect vehicle driving information (including vehicle body length, vehicle type, vehicle waiting time, vehicle speed, driver crossing needs, etc.).

[0081] B. Design a single-lane green light duration calculation scheme: When a vehicle approaches the stop line, the image recognition module identifies the vehicle, matches the vehicle information with the information provided by the on-board unit, and sends the matching result to the signal light unit to determine the current state of the vehicle:

[0082] State 1: The vehicle is in the green phase of the signal cycle and is allowed to continue driving.

[0083] State 2: The vehicle is not within the permitted passing time and must stop.

[0084] For state 1, when the vehicle is at the signal light that allows vehicles to pass, the vehicle maintains the speed through the intersection, and the road test unit measures the instantaneous speed V of the vehicle at the stop line. j2 , and calculate the time required to cross the intersection:

[0085] T v =L / V j2 +T c (T c =L c / V j2 )

[0086] Where L is the length of the intersection, T c L is the extra time required for a vehicle to fully pass through the intersection based on its length. c is the vehicle length.

[0087] In state 2, the vehicle must stop and wait, so the drive test unit needs to include the vehicle start delay time T in the calculation q The vehicle passing time T v ':

[0088] T v '=L / V j2 +T c +T q

[0089] The on-board unit is able to monitor the vehicle's start process and adopt a vehicle-specific start delay time instead of a fixed start delay time, thereby reducing errors caused by variations in start times for different vehicle types.

[0090] C. Multi-lane green light duration calculation scheme, considering that there are different numbers of vehicles on multiple lanes at the intersection, relies on the image recognition module to identify the last vehicle to pass through each lane. Relying on the road test unit to identify the last vehicle to pass through each lane, the signal is transmitted to the signal processor, and the signal light is fine-tuned (through yellow lights, etc.) to ensure that it passes the intersection safely, thereby clearing the intersection and completing a signal cycle. The traffic volume passing through the intersection is uploaded to the signal processor, and then the signal processor uses this data, together with the remaining number of vehicles in each lane, to recalculate the green light duration of the upcoming signal cycle.

[0091] D. Priority plan for special vehicles. Special vehicles include many categories, including but not limited to the following: emergency rescue vehicles, engineering vehicles, special transport vehicles, special equipment vehicles, and public service vehicles. When the road test unit detects a special vehicle, it transmits the signal to the signal processor, which controls the traffic light to give priority to the special vehicle.

[0092] In some possible implementations, the signal processor also includes a coordination control module, specifically including a single-intersection signal light coordination unit and a multi-intersection (network) signal light coordination unit; wherein the single-intersection signal light coordination unit uses reinforcement learning to dynamically adjust the signal light timing plan to reduce conflicts between pedestrians and vehicles, and monitors traffic dynamics in real time through a vehicle queue model to adjust signal duration; in the multi-intersection (network) signal light coordination unit, traffic flow is predicted through a long short-term memory model (LSTM), and a multimodal data fusion model is applied to optimize the signal cycle and green light time in combination with external factors such as weather and accidents.

[0093] Single intersection signal coordination unit: For a single intersection, through deep learning and reinforcement learning strategies, pedestrian and vehicle needs are integrated into one system to dynamically adjust the signal timing plan, reduce conflicts between pedestrians and vehicles, reduce delays and ease congestion.

[0094] In the control process of the single intersection signal light coordination module, problems such as overlapping vehicle queues and continuous arrival of pedestrians are faced. Solutions include building a vehicle queue model and monitoring traffic dynamics in real time to adjust signal duration. In the future, image recognition technology may improve the accurate management of pedestrian flow.

[0095] Multi-intersection (network) signal light coordination unit. Traffic flow transmission method: When vehicles leave the intersection, road test units (RSUs) are installed on the exit road to capture vehicle information, including speed, number and vehicle type. This data is used to predict the traffic flow at the upcoming intersection, thereby enhancing the interactive capabilities of the entire system by expanding its predictive insights at interconnected intersections.

[0096] Specific control methods for signal coordination units at single intersections and multiple intersections:

[0097] S1. Traffic volume data collection: In each signal cycle, the detector unit and the road test unit are used to collect the traffic volume data passing through the intersection; the data is uploaded to the signal processor, including the number of vehicles passing through each lane and the number of remaining vehicles.

[0098] S2. Pedestrian demand data collection: using pedestrian buttons, QR codes or sensor units to collect pedestrian crossing demand data;

[0099] S3. Use historical data to train reinforcement learning models, LSTM models, and data fusion models to optimize signal cycles and green light times. Combined with actual traffic conditions, the model parameters are continuously adjusted to improve the applicability and accuracy of the model; specifically, single intersection signal light coordination is performed through reinforcement learning; LSTM and data fusion models are applied to coordinate multi-intersection (network) signal lights;

[0100] S301, Adaptive control strategy based on reinforcement learning:

[0101] Node state definition: S = {T i ,P i ,E i}

[0102] Among them, S represents the node (i.e. intersection) state; T i represents the real-time traffic flow of the lane; P i Indicates the current pedestrian crossing demand; E i Indicates external influencing factors, such as weather, accidents, etc.

[0103] Define the control action A of the signal cycle and lane green light time: A={G i ,C i}

[0104] G i Indicates the green light time of the lane; C i Indicates the signal period;

[0105] Define the reward function R: R = -(α×vehicle waiting time + β×pedestrian waiting time + γ×traffic flow fluctuation)

[0106] Among them, α, β, and γ represent weight parameters.

[0107] Update the policy using Q-learning:

[0108] Among them, η1 is the learning rate; η2 is the discount factor; S' and A' represent the updated state and action of the next step after taking the original control action; η1' represents the learning rate of the next simulation training after the update.

[0109] S302, define a multimodal data fusion model D: D = f(T i ,P i ,W i ,AC i )

[0110] Among them, W i Indicates weather conditions; AC i represents the accident situation; f represents the fusion function.

[0111] S303, Traffic flow prediction model based on LSTM:

[0112] represents the predicted traffic flow at the next time step; H i Represents historical traffic flow; LSTM() represents the output result through the LSTM model.

[0113] S304, define pedestrian priority mechanism: G p,i =max(G min ,min(G max ,k p P i ))

[0114] Among them, G p,i represents the pedestrian green light time of the i-th lane; k p Indicates the ratio coefficient between pedestrian flow and green light time; G min , G max Indicates the upper and lower limits of the green light time.

[0115] S4. Calculation of optimized signal cycle and travel time:

[0116] S401, Calculation of optimized signal cycle and green light time

[0117]

[0118] Among them, λ is the weight coefficient of pedestrian flow; C' represents the optimized signal period; L oss Indicates lost time; Y i Represents the ratio of the saturation flow rate to the actual flow value of the i-th lane.

[0119] S402, calculation of the time difference between adjacent intersections. In the multi-intersection signal coordination, the time difference optimization formula between adjacent intersections is:

[0120]

[0121] Among them, δ is the weight coefficient of pedestrian flow change; d represents the distance between adjacent intersections; v represents the traffic flow speed; P i-1 represents the pedestrian crossing demand at the previous intersection, P i Indicates the current pedestrian crossing demand.

[0122] S5. Real-time adjustment and feedback control: Based on the shortest time required for a single pedestrian to pass through an intersection, the vehicle start-up delay time, the time required for vehicles to pass through an intersection, and the real-time traffic conditions, at the end of each signal cycle, the signal cycle and travel time are controlled in real time based on the feedback of pedestrian and vehicle crossing effects and the results of the reinforcement model, data fusion model, and traffic prediction model, so that the signal control system can respond to real-time traffic and pedestrian needs more intelligently and flexibly, thereby improving traffic efficiency and safety.

[0123] The solution of the present invention realizes real-time dynamic adjustment of green light time by introducing adaptive control strategy based on reinforcement learning, multimodal data fusion, traffic flow prediction and pedestrian priority mechanism. By optimizing signal cycle and green light time in real time, the waiting time of vehicles and pedestrians is reduced and the traffic efficiency is improved. The pedestrian priority mechanism and multimodal data fusion are used to ensure the safe passage of pedestrians and vehicles under different traffic conditions. Through reinforcement learning and LSTM model, the system can adaptively adjust to the dynamic changes of traffic flow and pedestrian demand. These innovative measures have significantly improved the efficiency of pedestrian and vehicle crossing.

[0124] Each embodiment in this specification is described in a related manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0125] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims

1. A traffic signal control system that dynamically responds to pedestrian and vehicle demand, characterized in that: Specifically include: A signal light unit, the signal light unit comprising a pedestrian activation button, a signal light and a signal processor; the signal light unit is connected to the detector unit by signal; the pedestrian activation button is connected to the signal processor for responding to pedestrian crossing requirements, and the signal processor is used to process received signals and control the state of the signal light according to the signals; The detector unit includes a camera, a video processor, an infrared detector, a speed detector and a voice detector; the detector unit is used to feed back the movement information of pedestrians and vehicles to the signal processor in the form of signals; A QR code, the encoding of which contains a URL for sending a signal to a signal processor; used to upload pedestrian crossing requirements to the signal processor; The on-board unit is mounted on the vehicle; the on-board unit is connected to the signal processor of the signal light unit via the Internet; the on-board unit detects the real-time status and driving information of the vehicle and sends it to the signal light unit, and the on-board unit is also connected to the road test unit by signal; The road test unit includes a road test camera and an image recognition module, which is connected to the signal processor signal. The image recognition module of the road test unit matches the vehicle image information collected by the road test camera with the on-board unit, and sends the matching information and the received real-time traffic data to the signal processor.

2. A traffic signal control system that dynamically responds to pedestrian and vehicle demand according to claim 1, characterized in that: The camera is used to collect images of pedestrians and vehicles, the video processor is used to process the images, and the infrared detector, speed detector and voice detector collect the position, speed and sound of pedestrians and upload them to the signal processor, which determines the pedestrian status information.

3. The control method of a traffic signal control system that dynamically responds to pedestrian and vehicle demands as claimed in claim 1, characterized in that: The process is as follows: The signal processor receives information from the pedestrian activation button, the detector unit, the QR code, the vehicle-mounted unit, and the road test unit and controls the signal light; The on-board unit collects vehicle driving information, the road test unit identifies the vehicle, matches the vehicle information with the on-board unit, and sends the matching result to the signal light unit.

4. The control method of a traffic signal control system that dynamically responds to pedestrian and vehicle demands according to claim 3 is characterized in that: The signal processor controls the traffic light in the following way: receiving the signal of the pedestrian activation button or the crossing demand signal obtained by the pedestrian scanning the QR code, obtaining the pedestrian crossing speed through the data collected by the camera and the speed detector, and calculating the time required for each pedestrian to pass through the intersection, and adjusting the switching time of the traffic light based on the calculation result: Among them, T p represents the shortest time required for a single pedestrian to pass through the intersection, V p is the speed limit for pedestrians to cross the intersection, L is the length of the intersection, T s Indicates the initial hesitation time of the pedestrian; For users who cannot scan the QR code, use the button to activate.

5. The control method of a traffic signal control system that dynamically responds to pedestrian and vehicle demands according to claim 3 is characterized in that: The specific control process of the traffic light by the vehicle-mounted unit and the road test unit is as follows: A. The on-board unit collects vehicle information; wherein the on-board unit can monitor the vehicle startup process, and the vehicle information includes the vehicle startup delay time; B. The on-board unit and the road test unit match the vehicle status information and send it to the signal processor of the signal light unit for processing: When the vehicle is at the signal light that allows vehicles to pass, the vehicle maintains the speed through the intersection, and the road test unit measures the instantaneous speed V of the vehicle at the stop line. j2 , and calculate the time T required to pass the intersection v : T v =L / V j2 +T c T c =L c / V j2 Where L is the length of the intersection, T c L is the extra time required for a vehicle to fully pass through the intersection based on its length. c is the vehicle length; When the vehicle is not within the allowed passing time, it must stop. The road test unit calculation includes the vehicle start delay time T q The vehicle passing time T v ': T v '=L / V j2 +T c +T q C. The road test unit uploads the traffic volume passing through the intersection to the signal processor, together with the number of remaining vehicles in each lane, to recalculate the green light duration of the upcoming signal cycle; the road test unit also uploads the last vehicle information passing through each lane, and the signal processor then fine-tunes the signal light; D. After the road test unit detects a special vehicle, it uploads the signal to the signal processor, which controls the traffic light to release the vehicle.

6. The control method of a traffic signal control system that dynamically responds to pedestrian and vehicle demands according to claim 2 is characterized in that: The signal processor includes a coordination control module, which specifically includes a single-intersection signal light coordination unit and a multi-intersection signal light coordination unit.

7. A control method for a traffic signal control system according to any one of claims 4 to 6 that dynamically responds to pedestrian and vehicle demands, characterized in that: The control method of the single intersection and multi-intersection signal light coordination unit is specifically as follows: S1. In each signal cycle, the detector unit and the road test unit are used to collect traffic volume data passing through the intersection; the data is uploaded to the signal processor, including the number of vehicles passing through each lane in the intersection and the number of remaining vehicles; S2, using sensor units to collect pedestrian crossing data; S3, use historical data to train reinforcement learning models, LSTM models and data fusion models to optimize signal cycles and green light time; S4, calculating the optimized signal cycle and travel time; S5. Based on the shortest time required for a single pedestrian to pass through the intersection, the vehicle start delay time, the time required for vehicles to pass through the intersection and the real-time traffic conditions, at the end of each signal cycle, the signal cycle and travel time are controlled according to the feedback of pedestrian and vehicle crossing effects and the results of S3 to S4.

8. The control method of a traffic signal control system that dynamically responds to pedestrian and vehicle demands according to claim 7 is characterized in that: The specific steps of S3 are as follows: S301, Adaptive control based on reinforcement learning: Define the intersection state: S={T i ,P i ,E i } Among them, S represents the state of the intersection, T i represents the real-time traffic flow of the lane, P i represents the current pedestrian crossing demand, E i Indicates external influencing factors; Define the control actions for signal cycle and lane green light time: A: A={G i ,C i } Among them, G i Indicates the green light time of the lane; C i Indicates the signal period; Define the reward function R: R = -(α×vehicle waiting time + β×pedestrian waiting time + γ×traffic flow fluctuation) where α, β, and γ represent weight parameters; Update the policy using Q-learning: Among them, η1 is the learning rate; η2 is the discount factor; S' and A' represent the updated state and action of the next step after taking the original control action; η′1 represents the learning rate of the next simulation training after the update; max is the maximum value function; S302. Traffic flow prediction model based on LSTM: represents the predicted traffic flow at the next time step; H i Represents historical traffic flow; LSTM() represents the output result through the LSTM model; S304, define pedestrian priority mechanism: G p,i =max(G min ,min(G max ,k p P i )) Among them, G p,i represents the pedestrian green light time of the i-th lane; k p Indicates the ratio coefficient between pedestrian flow and green light time; G max , G min Indicates the upper and lower limits of the green light time.

9. The control method of a traffic signal control system that dynamically responds to pedestrian and vehicle demands according to claim 7, characterized in that: The specific steps of S4 are as follows: S401. Calculation of optimized signal cycle and travel time: Among them, λ is the weight coefficient of pedestrian flow; C' represents the optimized signal period; L oss Indicates lost time; Y i It represents the ratio of the saturation flow rate to the actual flow value of the i-th lane; S402. In the multi-intersection signal coordination, the time difference ΔT′ between adjacent intersections is: Among them, δ is the weight coefficient of pedestrian flow change; d represents the distance between adjacent intersections; v represents the traffic flow speed; P i represents the current pedestrian crossing demand, P i-1 Indicates the pedestrian crossing demand at the previous intersection.

Images