Perception intelligent driving multi-mode traffic signal optimization method and system

Abstract

The invention discloses a sensing intelligent driving multi-mode traffic signal optimization method and system, and relates to the technical field of traffic signal optimization, and the method comprises the steps: collecting traffic flow data; constructing a multi-mode traffic flow model; connecting a plurality of traffic lights in the target area to obtain a street lamp control signal; based on the traffic flow data and the street lamp control signal, using the multi-mode traffic flow model to predict a traffic flow change trend; and configuring a timing optimization scheme of a plurality of traffic lights in the target area according to the traffic flow change trend prediction result. According to the method and the device, the technical problem of low traffic intersection passing efficiency caused by the fact that a traffic signal lamp timing scheme cannot be dynamically adjusted accurately according to real-time traffic conditions and multi-mode traffic behavior interaction relations in the prior art is solved, and the purposes of optimizing the traffic signal lamp timing of the target area according to the prediction result and improving the traffic signal lamp timing efficiency are achieved. And the traffic efficiency of the traffic intersection is effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic signal optimization, and particularly to a multi-mode traffic signal optimization method and system driven by perception intelligence. Background Art

[0002] With the rapid development of urbanization, the urban population and the number of motor vehicles have increased sharply, and the road traffic flow has become extremely complex. The traditional traffic signal timing system has many defects. It is usually set according to fixed time intervals or simple traffic flow statistics, and cannot adapt to complex and changeable traffic conditions. It cannot reasonably allocate road resources during peak and off-peak hours, ignores the traffic needs of non-motor vehicles and pedestrians, does not consider the interactive effects between different traffic modes, and the existing traffic flow prediction methods have poor accuracy and cannot provide reliable support for signal timing optimization.

[0003] The existing technology has the technical problem that the traffic signal timing scheme cannot be accurately adjusted dynamically according to the real-time traffic conditions and the interactive relationship of multi-mode traffic behaviors, resulting in low traffic efficiency at intersections. Summary of the Invention

[0004] This application provides a multi-mode traffic signal optimization method and system driven by perception intelligence, which is used to solve the technical problem that the traffic signal timing scheme in the existing technology cannot be accurately adjusted dynamically according to the real-time traffic conditions and the interactive relationship of multi-mode traffic behaviors, resulting in low traffic efficiency at intersections.

[0005] In view of the above problems, this application provides a multi-mode traffic signal optimization method and system driven by perception intelligence.

[0006] In the first aspect of this application, a multi-mode traffic signal optimization method driven by perception intelligence is provided. The method includes:

[0007] Collect traffic flow data, where the traffic flow data includes motor vehicle flow, non-motor vehicle flow, and pedestrian flow; construct a multi-mode traffic flow model, which is used to simulate the interactive behavior relationship between motor vehicle behavior characteristics, non-motor vehicle behavior characteristics, and pedestrian behavior characteristics; connect multiple traffic signals in the target area to obtain a street lamp control signal; based on the traffic flow data and the street lamp control signal, use the multi-mode traffic flow model to predict the traffic flow change trend; according to the prediction result of the traffic flow change trend, configure the timing optimization scheme of multiple traffic signals in the target area.

[0008] In the second aspect of this application, a multi-mode traffic signal optimization system driven by perception intelligence is provided. The system includes:

[0009] A traffic flow data collection module for collecting traffic flow data, where the traffic flow data includes motor vehicle flow, non-motor vehicle flow, and pedestrian flow; a multi-mode traffic flow model construction module for constructing a multi-mode traffic flow model, which is used to simulate the interaction behavior relationship among motor vehicle behavior characteristics, non-motor vehicle behavior characteristics, and pedestrian behavior characteristics; a street lamp control signal acquisition module for connecting multiple traffic lights in the target area to obtain street lamp control signals; a traffic flow change trend prediction module for predicting the traffic flow change trend based on the traffic flow data and the street lamp control signals using the multi-mode traffic flow model; a timing optimization scheme configuration module for configuring the timing optimization scheme of multiple traffic lights in the target area according to the traffic flow change trend prediction result.

[0010] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0011] Collect traffic flow data, where the traffic flow data includes motor vehicle flow, non-motor vehicle flow, and pedestrian flow; construct a multi-mode traffic flow model, which is used to simulate the interaction behavior relationship among motor vehicle behavior characteristics, non-motor vehicle behavior characteristics, and pedestrian behavior characteristics; connect multiple traffic lights in the target area to obtain street lamp control signals; predict the traffic flow change trend based on the traffic flow data and the street lamp control signals using the multi-mode traffic flow model; configure the timing optimization scheme of multiple traffic lights in the target area according to the traffic flow change trend prediction result. It achieves the technical effect of optimizing the timing of traffic lights in the target area according to the prediction result, effectively improving the traffic efficiency at intersections. Description of the Drawings

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0013] Figure 1 It is a schematic flowchart of a multi-mode traffic signal optimization method driven by perception intelligence provided by an embodiment of this application;

[0014] Figure 2 It is a schematic structural diagram of a multi-mode traffic signal optimization system driven by perception intelligence provided by an embodiment of this application.

[0015] Explanation of the reference numerals: traffic flow data collection module 10 , multi-mode traffic flow model construction module 20 , street light control signal acquisition module 30 , traffic flow change trend prediction module 40 , timing optimization scheme configuration module 50 . DETAILED DESCRIPTION

[0016] The present application provides a multi-modal traffic signal optimization method and system driven by perception intelligence, which is used to solve the technical problem in the prior art that the traffic signal timing scheme cannot be accurately adjusted dynamically according to the real-time traffic conditions and the interactive relationship between multi-modal traffic behaviors, resulting in low traffic efficiency at traffic intersections.

[0017] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0018] Embodiment 1, as Figure 1 As shown, the present application provides a multi-mode traffic signal optimization method driven by perception intelligence, the method comprising:

[0019] Step S100: Collecting traffic flow data, where the traffic flow data includes motor vehicle flow, non-motor vehicle flow, and pedestrian flow.

[0020] Specifically, the traffic flow data is collected first, using a variety of sensor devices, such as geomagnetic sensors, video detectors, and radar sensors, which are deployed at traffic intersections and surrounding roads. Geomagnetic sensors can accurately sense the passage of motor vehicles, thereby calculating the motor vehicle flow; video detectors use image recognition technology to not only identify motor vehicles, but also effectively monitor non-motor vehicles and pedestrians and count their flow data; radar sensors can further supplement the detection of traffic participants' speed and other information. The motor vehicle flow, non-motor vehicle flow, and pedestrian flow data collected by these devices are transmitted to the server of the traffic signal control center in real time and stably, providing a solid data foundation for subsequent analysis and processing.

[0021] Step S200: constructing a multi-modal traffic flow model, wherein the multi-modal traffic flow model is used to simulate the interactive behavior relationship between motor vehicle behavior characteristics, non-motor vehicle behavior characteristics, and pedestrian behavior characteristics.

[0022] Specifically, when constructing a multi-modal traffic flow model for simulating the interaction relationship of the behavior characteristics of motor vehicles, non-motor vehicles and pedestrians, a neural network model is adopted. First, comprehensively collect the historical traffic flow data of the target area, covering different time periods, weather, seasons, etc., including the number, speed, direction, etc. of motor vehicles, non-motor vehicles and pedestrians. Then, perform data preprocessing, clean the noise and error records, unify the data scale through normalization, and extract meaningful features such as flow change rate, density, etc. Subsequently, select a suitable neural network model according to the requirements, such as the recurrent neural network (RNN) that can process sequential data and its variants the long short-term memory network (LSTM), gated recurrent unit (GRU). When designing the model architecture, the number of neurons in the input layer depends on the number of extracted features, and the features of motor vehicles, non-motor vehicles and pedestrians are respectively input into different branches; the hidden layer learns complex relationships through multiple non-linear transformations, and interaction nodes are set between different branches to capture interaction behaviors; the output layer outputs the simulated traffic flow state. Then, use the preprocessed historical data to train the model, set appropriate loss functions and optimization algorithms, and continuously adjust the weights and biases during training to make the model fit the data and learn the interaction rules of the multi-modal traffic flow, and finally construct a multi-modal traffic flow model that meets the requirements.

[0023] Step S300: Connect multiple traffic lights in the target area to obtain a street lamp control signal.

[0024] Specifically, connect multiple traffic lights in the target area to obtain a street lamp control signal. In actual operation, through wired or wireless communication technology, the server of the traffic signal control center is stably connected to each traffic light in the target area. Adopt a standardized communication protocol, such as the NTCIP protocol, to ensure smooth and accurate data transmission and instruction interaction between the server and the traffic lights. Once the connection is successful, the server can obtain various control signals of the traffic lights in real time, including the current state of the traffic lights (red light, green light or yellow light), the duration setting of the signal cycle, and the duration allocation of each phase (such as straight phase, left turn phase, etc.). These street lamp control signals are an important basis for understanding the current operation of traffic signals and play an indispensable role in subsequent analysis of traffic flow changes and optimization of signal timing.

[0025] Step S400: Based on the traffic flow data and the street lamp control signal, use the multi-modal traffic flow model to predict the traffic flow change trend.

[0026] Specifically, based on the previously collected traffic flow data and the obtained street lamp control signals, the constructed multi-mode traffic flow model is used to predict the changing trend of traffic flow. On the computing platform of the traffic signal control center, the real-time traffic flow data and the latest obtained street lamp control signals are used as input parameters and input into the multi-mode traffic flow model. The model will, based on various traffic behavior patterns and interaction relationships learned during the training process, and in combination with the current traffic conditions, simulate and predict the changing trend of traffic flow in the future for a certain period of time. For example, predicting the changes in the motor vehicle flow in different directions at a certain intersection within the next 15 minutes, whether congestion will occur, the possible locations of congestion, and the degree of congestion; the aggregation of non-motor vehicles and pedestrians at different time periods and their impact on motor vehicle traffic, etc. To improve the accuracy and reliability of the prediction, time series analysis methods and advanced machine learning prediction algorithms such as Long Short-Term Memory Network (LSTM) and Support Vector Regression (SVR) are used. These algorithms can deeply mine the time series characteristics and potential laws in traffic data, take into account the uncertainty of traffic demand and the impact of emergencies (such as traffic accidents, road construction, etc.) on traffic flow, and through means such as setting random variables and scenario analysis, comprehensively consider various factors for prediction, so that the prediction results are more in line with the actual traffic conditions. The prediction results will be presented to traffic management personnel in an intuitive visual form, such as showing the traffic flow change curves in different time periods and different directions through charts, and marking the areas where congestion may occur using maps, etc., to facilitate the management personnel to timely understand the development trend of traffic flow and provide strong support for subsequent decision-making.

[0027] Step S500: Configure the timing optimization scheme for multiple traffic signals within the target area according to the prediction results of the traffic flow change trend.

[0028] Specifically, according to the prediction results of the traffic flow change trend, a timing optimization plan for multiple traffic lights in the target area is configured. At the traffic signal control center, first, edge computing nodes are deployed at multiple traffic lights in the target area. These edge computing nodes have certain computing and data processing capabilities and can perform preliminary analysis and processing of traffic data locally. Based on the local traffic data collected by the edge computing nodes and combined with the prediction information of the traffic flow change trend, a local traffic light timing plan is generated. The local timing plan generated by each edge computing node will synchronize data with the central node through the network and upload its own plan information to the central node. At the central node, the motor vehicle passing efficiency, non-motor vehicle passing efficiency, and pedestrian passing efficiency are jointly used as the joint optimization objectives. According to the real-time traffic flow data, dynamic weight allocation is performed for these three optimization objectives. For example, during the morning and evening rush hours, the motor vehicle flow is large, and at this time, the weight of the motor vehicle passing efficiency is appropriately increased; during the school dismissal period, the pedestrian flow around increases, so the weight of the pedestrian passing efficiency is increased. By dynamically adjusting the weights, multiple different timing adaptation plans are obtained. To screen out the optimal timing plan, first, an important factor of pedestrian crossing demand is introduced. Based on the pedestrian flow data and the actual demand for pedestrian crossing, the multiple timing adaptation plans are sorted by priority, and the first screening condition is set. For example, if a certain plan can ensure that pedestrians cross the road quickly under the premise of safety and reduce the waiting time of pedestrians, then this plan will have a higher priority in the sorting. At the same time, to further optimize the screening process, the high-flow areas and high-conflict areas in the target area are marked. Through in-depth analysis of traffic data and on-site observation, the key traffic flow nodes are identified. These key nodes usually include locations such as road intersections, crosswalks, and non-motor vehicle lane intersections where traffic congestion and conflicts are likely to occur. For the key traffic flow nodes in different areas, their traffic flow characteristics are analyzed separately. At the key traffic flow nodes in high-flow areas, the potential conflicts between different traffic modes, such as the passing conflicts between motor vehicles and non-motor vehicles, pedestrians, are mainly analyzed, and the corresponding first feature vector is configured; at the key nodes in high-conflict areas, indicators such as the queue length, waiting time, and queue dissipation speed of vehicles and pedestrians are analyzed in detail, and the second feature vector is configured. Based on the information contained in these two feature vectors, the multiple timing adaptation plans are screened for the second time, and the second screening condition is set. After these two rounds of screening, the most suitable timing optimization plan for multiple traffic lights in the target area is finally determined. This optimization plan will be transmitted to the control systems of each traffic light through the network to automatically update the timing parameters of the traffic lights and achieve real-time optimization control of traffic signals. Moreover, during the operation of traffic signals, the actual operation of traffic flow is continuously monitored, feedback data is collected, and the timing plan is dynamically adjusted according to the new traffic conditions to ensure that traffic signals can always adapt to the changing traffic demands, improve the overall passing efficiency of traffic intersections, relieve traffic congestion, and ensure traffic safety.

[0029] In a possible implementation manner, step S200 further includes:

[0030] Step S210: Obtain historical traffic flow data within the target area.

[0031] Step S220: Based on the historical traffic flow data, decompose the historical flow of motor vehicles, non-motor vehicles, and pedestrians, and set a training data set.

[0032] Step S230: Based on the training data set, train the multi-mode traffic flow model, and use a sliding window mechanism to dynamically update the prediction result of the traffic flow change trend.

[0033] Specifically, to obtain the historical traffic flow data within the target area, data is extracted from the database of the traffic management department and the storage units of various traffic monitoring devices distributed in the target area. These data cover a long time period, including information in rich scenarios such as weekdays and non-weekdays, different seasons, morning and evening rush hours, and off-peak hours, and include traffic flow data of motor vehicles, non-motor vehicles, and pedestrians, as well as associated information such as corresponding timestamps, weather conditions, and road construction, providing a comprehensive basis for subsequent in-depth analysis.

[0034] Based on the obtained historical traffic flow data, decompose the historical flow of motor vehicles, non-motor vehicles, and pedestrians and set a training data set. Using data processing algorithms, split the traffic flow data of different traffic participants according to dimensions such as time series, road segments, and traffic conditions. For example, take every 5 minutes as a time interval, and extract the traffic flow data of motor vehicles, non-motor vehicles, and pedestrians on each road segment respectively. Then, according to different traffic scene labels, such as "rush-hour congested road segments" and "off-peak smooth road segments", classify and organize the split data, and select representative data samples to form a training data set. This processing can ensure that the training data covers various typical traffic conditions and improve the comprehensiveness and accuracy of model training.

[0035] The multi-modal traffic flow neural network model will be trained based on the prepared training data set, and the sliding window mechanism will be used to dynamically update the prediction results of the traffic flow change trend. First, the training data set is input into the neural network model. The input layer of the model receives various feature data of motor vehicles, non-motor vehicles, and pedestrians. After complex non-linear transformations by multiple neurons in the hidden layer, it learns the interaction relationships between the behavior characteristics of different traffic subjects. During the training process, a suitable loss function (such as the mean squared error loss function) is used to measure the difference between the model prediction value and the actual value, and the stochastic gradient descent algorithm is used to continuously adjust the weight and bias parameters of the model to minimize the loss function value and improve the prediction accuracy of the model. At the same time, in order to enable the model to adapt to the changing traffic conditions, a sliding window mechanism is introduced. With a fixed time interval as the window length, as new traffic flow data is continuously generated, the new data is added to the window, and at the same time, the earliest data in the window is removed to form the updated window data. The updated window data is used to perform incremental training or fine-tuning on the model, enabling the model to learn the latest traffic flow characteristics and rules in real time, thereby dynamically updating the prediction results of the traffic flow change trend, ensuring the timeliness and accuracy of the prediction, and providing a reliable basis for the subsequent optimization of traffic signal timing.

[0036] In a possible implementation manner, step S500 further includes:

[0037] Step S510: Deploy edge computing nodes according to multiple traffic lights in the target area.

[0038] Step S520: Generate a local traffic light timing plan based on the edge computing nodes and synchronize data with the central node.

[0039] Step S530: Determine the timing optimization plan for multiple traffic lights in the target area through multiple local traffic light timing plans uploaded to the central node.

[0040] Specifically, edge computing nodes are deployed according to multiple traffic lights in the target area. Conduct a detailed survey of the traffic layout in the target area, comprehensively considering factors such as the distribution location of traffic lights, the busyness of surrounding roads, and network coverage. For areas with large traffic flow and complex intersections, edge computing nodes with stronger performance will be key deployed; while in areas with relatively less vehicle flow and simpler road conditions, lighter nodes will be deployed. These edge computing nodes are connected to the traffic lights through wired or wireless communication technologies and can obtain the status information of the traffic lights and the surrounding traffic data in real time.

[0041] Generate a local traffic signal timing plan based on the deployed edge computing nodes and synchronize data with the central node. With its own data collection function, the edge computing node collects in real time the detailed traffic data around the traffic signals in the area, covering information such as the flow, speed and direction of motor vehicles, non-motor vehicles and pedestrians. Then, using the genetic algorithm preset in the node, combined with the historical traffic data of the area, the current real-time traffic conditions and specific traffic rules, these data are deeply analyzed and processed to generate a traffic signal timing plan suitable for this local area. This plan aims to maximize the traffic passing efficiency of the local area and reduce the waiting time of vehicles and pedestrians. After generating the plan, the edge computing node uploads the local timing plan and related real-time traffic data to the central node through a stable and reliable communication network. At the same time, it receives instructions from the central node, traffic information of other areas and the overall traffic planning strategy, etc., to complete data synchronization, so as to ensure that the local timing plan is consistent and coordinated with the traffic management objectives of the entire target area.

[0042] Determine the timing optimization plan for multiple traffic signals in the target area through multiple local traffic signal timing plans uploaded to the central node. After receiving the local timing plans uploaded by each edge computing node, the central node will conduct a comprehensive analysis of these plans. The factors considered include the traffic relevance between different regions, the balance of traffic flow at each intersection, and the improvement effect of the overall traffic congestion situation, etc. The central node uses a more advanced optimization algorithm to integrate and adjust multiple local plans, weighing the interests and needs of all aspects, and finally determines a timing optimization plan that can not only meet the traffic needs of each local area but also maximize the overall traffic benefits of the target area, and then distributes this plan to each traffic signal to achieve the optimized control of traffic signals.

[0043] In a possible implementation manner, step S530 further includes:

[0044] Step S531: Take the passing efficiency of motor vehicles, non-motor vehicles and pedestrians as the joint optimization target.

[0045] Step S532: Dynamically allocate weights to the joint optimization target according to the real-time traffic flow data to obtain multiple timing adaptation plans.

[0046] Step S533: Screen the multiple timing adaptation plans to determine the timing optimization plan for multiple traffic signals in the target area.

[0047] Specifically, the traffic efficiency of motor vehicles, non-motor vehicles, and pedestrians is set as the joint optimization goal. This means that when formulating the timing plan, instead of solely focusing on a certain type of traffic mode, the needs of all parties are comprehensively considered, aiming to achieve a more reasonable allocation of road resources. The traffic efficiency of motor vehicles is reflected in the number of motor vehicles passing through the intersection per unit time and their average delay time; the traffic efficiency of non-motor vehicles can be measured by indicators such as the waiting time of non-motor vehicles at the intersection and the change in riding speed; the traffic efficiency of pedestrians focuses on the time required for pedestrians to cross the street and the waiting duration. Taking these three as the joint optimization goal can comprehensively improve the overall traffic capacity of the intersection and ensure the travel experience of different traffic participants.

[0048] According to the real-time traffic flow data, dynamic weight allocation is carried out for the joint optimization goal, and then multiple timing adaptation plans are obtained. The edge computing node continuously collects the real-time traffic flow data of the intersection, including the flow changes of motor vehicles, non-motor vehicles, and pedestrians at different times and in different directions. Based on this data, using the weight allocation algorithm, the weights of the three optimization goals are flexibly adjusted according to the real-time changes in traffic flow. For example, during the morning rush hour on weekdays, the traffic flow of motor vehicles increases significantly, and at this time, the weight of the traffic efficiency of motor vehicles is appropriately increased; while during the school dismissal period, the pedestrian flow around the school surges, and the weight of the traffic efficiency of pedestrians is correspondingly increased. By dynamically adjusting the weights and combining various possible combinations of signal timing, multiple timing adaptation plans that meet different traffic conditions are generated to adapt to the complex and changeable traffic scenarios.

[0049] Multiple timing adaptation plans are screened to determine the timing optimization plan for multiple traffic lights in the target area. The screening process comprehensively considers various factors, such as the pedestrian crossing demand. Based on the pedestrian flow data and the actual demand for pedestrians to cross the street, such as whether there are a large number of relatively slow-moving people such as students and the elderly, multiple timing adaptation plans are initially screened and prioritized. At the same time, high-flow areas and high-conflict areas are marked, and key traffic flow nodes are identified, such as road intersections, crosswalks, and the convergence points of non-motor vehicle lanes. For these key nodes, the potential conflicts between different traffic modes, as well as the characteristics of the queue length, waiting time, and queue dissipation speed of vehicles and pedestrians are analyzed. According to the results of these detailed analyses, the timing adaptation plans are screened again to remove those plans that may lead to increased traffic congestion and conflicts. After multiple rounds of screening, the timing optimization plan that can maximize the overall traffic efficiency of the intersection and ensure traffic safety under the current traffic conditions is selected from multiple timing adaptation plans to achieve precise control of traffic lights and optimize the urban traffic order.

[0050] In a possible implementation manner, step S533 further includes:

[0051] Step S5331: Introduce the pedestrian crossing demand.

[0052] Step S5332: Based on the pedestrian flow data and the pedestrian crossing needs, prioritize the multiple timing adaptation schemes and set the first screening condition.

[0053] Specifically, when determining multiple traffic signal timing optimization schemes in the target area, the pedestrian crossing needs are introduced. As an important part of traffic participants, the pedestrian crossing needs cannot be ignored. This not only involves the safety of pedestrians but also affects the operation efficiency of the entire traffic system. In actual traffic scenarios, the pedestrian crossing needs are affected by various factors, such as the distribution of functional areas around the intersection (whether it is close to schools, shopping malls, residential areas, etc.), the pedestrian aggregation situation during specific time periods (such as morning and evening rush hours, school arrival and dismissal times, etc.). Through on-site investigations, historical data statistics, and real-time monitoring, etc., collect pedestrian crossing need information, including the desired crossing time interval for pedestrians, the acceptable maximum waiting time, etc.

[0054] Based on the pedestrian flow data and the pedestrian crossing needs, prioritize the multiple timing adaptation schemes and set the first screening condition. Combine the real-time collected pedestrian flow data with the previously collected pedestrian crossing need information to evaluate each timing adaptation scheme. If a certain timing scheme can provide sufficient green light time for pedestrians when the pedestrian flow is large, ensuring that pedestrians can cross the road safely and smoothly, and at the same time minimize the impact on the traffic of motor vehicles and non-motor vehicles, then this scheme will be in a higher position in the priority ranking. For example, at intersections near schools, during the school dismissal period, the pedestrian flow is large and mostly students. For those timing schemes that can ensure that students have sufficient time to cross the road safely and do not cause long-term congestion of motor vehicles, they are given priority. In this way, prioritize the multiple timing adaptation schemes according to pedestrian-related factors, set it as the first screening condition, and initially screen out the schemes that better meet the pedestrian needs, laying a foundation for further determining the final timing optimization scheme, reflecting the protection of pedestrian rights and interests and the emphasis on the humanized management of the traffic system.

[0055] In a possible implementation manner, step S533 further includes:

[0056] Step S5333: Mark the high-flow areas and high-conflict areas.

[0057] Step S5334: Based on the high-flow areas and high-conflict areas, identify the key traffic flow nodes and determine the traffic flow characteristics corresponding to the key traffic flow nodes.

[0058] Step S5335: According to the traffic flow characteristics corresponding to the key traffic flow nodes, conduct a secondary screening of the multiple timing adaptation schemes and set the second screening condition.

[0059] Specifically, mark the high - traffic areas and high - conflict areas. With the help of traffic flow monitoring devices, such as geomagnetic sensors, high - definition cameras, etc., continuously collect the real - time traffic flow data of each road section in the target area. Through data analysis algorithms, compare the traffic flow data of each area with the historical average flow. When the traffic flow in a certain area is significantly higher than the average level, mark it as a high - traffic area; at the same time, combined with the traffic conflict monitoring system, evaluate the frequency and severity of conflicts between different traffic modes (motor vehicles, non - motor vehicles, pedestrians), and mark the areas where conflicts occur frequently as high - conflict areas. For example, during the morning and evening rush hours, the intersections and surrounding road sections of urban arterial roads are prone to congestion and conflicts due to the sharp increase in traffic volume and the interweaving of vehicles in different directions, and are usually marked as high - traffic and high - conflict areas.

[0060] Based on the marked high - traffic areas and high - conflict areas, identify the key nodes of the traffic flow and determine their corresponding traffic flow characteristics. In these areas, positions such as road intersections, crosswalks, and the convergence points of non - motor vehicle lanes are identified as key nodes of the traffic flow because the traffic flow converges and diverges here, and the traffic conditions are complex. For these key nodes, analyze the traffic flow characteristics from different dimensions. At the key nodes in high - traffic areas, focus on analyzing the potential conflicts between different traffic modes, such as the possibility of conflicts between motor vehicle turning and non - motor vehicle going straight and pedestrians, and configure the first feature vector corresponding to the key nodes of the traffic flow accordingly; at the key nodes in high - conflict areas, focus on indicators such as the queue length of vehicles and pedestrians, waiting time, and queue dissipation speed, and configure the second feature vector corresponding to the key nodes of the traffic flow based on this. Combine the first and second feature vectors to comprehensively determine the traffic flow characteristics corresponding to the key nodes of the traffic flow, providing an accurate basis for subsequent screening.

[0061] According to the traffic flow characteristics corresponding to the key nodes of the traffic flow, conduct a secondary screening of multiple timing adaptation schemes and set the second screening conditions. Substitute each timing adaptation scheme into the actual scenario of the key nodes of the traffic flow for simulation evaluation. Give higher priority to those schemes that can effectively relieve the traffic pressure in high - traffic areas and reduce the frequency of conflicts in high - conflict areas. For example, if a certain timing scheme can significantly shorten the queue length of vehicles and reasonably reduce the waiting time of pedestrians at the key nodes in high - conflict areas, then this scheme is more likely to be retained; on the contrary, if a scheme leads to increased traffic congestion and more conflicts at the key nodes, it will be screened out. Through this secondary screening process based on the characteristics of the key nodes of the traffic flow, set more stringent second screening conditions, and further screen out better schemes from multiple timing adaptation schemes, so that the finally determined traffic signal timing optimization scheme better meets the actual traffic needs, can effectively improve traffic operation efficiency, and ensure traffic safety.

[0062] In a possible implementation manner, step S5334 further includes:

[0063] Identify the key nodes of traffic flow, where the key nodes of traffic flow include road intersections, crosswalks, and intersections of non-motorized lanes; at the key nodes of traffic flow in the high-flow area, analyze the potential conflicts between different traffic modes, and configure the first feature vector corresponding to the key nodes of traffic flow; at the key nodes of traffic flow in the high-conflict area, analyze the queue lengths, waiting times, and queue dissipation speeds of vehicles and pedestrians, and configure the second feature vector corresponding to the key nodes of traffic flow; based on the first feature vector and the second feature vector, determine the traffic flow characteristics corresponding to the key nodes of traffic flow.

[0064] Specifically, to accurately grasp the traffic conditions and reasonably optimize the traffic signal timing, it is necessary to identify the key nodes of traffic flow and analyze their characteristics. First, through a comprehensive survey and data analysis of the traffic network in the target area, it is clear that road intersections, crosswalks, and intersections of non-motorized lanes are the key nodes of traffic flow. These locations are areas where different traffic modes converge and transform, with complex and variable traffic conditions, which have a significant impact on the overall traffic operation efficiency.

[0065] At the key nodes of traffic flow in the high-flow area, deeply analyze the potential conflicts between different traffic modes. For example, at road intersections, there are conflicts between motor vehicles turning right and non-motorized vehicles and pedestrians going straight; at crosswalks, there may also be conflicts between pedestrians crossing the street and motor vehicles turning right. In response to these situations, collect relevant data, such as the flow, speed, and driving direction of different traffic modes, and configure the first feature vector corresponding to the key nodes of traffic flow based on this. This vector can quantify the possibility and severity of potential conflicts between different traffic modes, providing key information for evaluating traffic conditions.

[0066] At the key nodes of traffic flow in the high-conflict area, focus on analyzing the queue lengths, waiting times, and queue dissipation speeds of vehicles and pedestrians. Use traffic monitoring devices, such as cameras and sensors, to obtain this data in real time. Long queues and waiting times not only reduce traffic efficiency but may also trigger more conflicts; while the queue dissipation speed reflects the ability to relieve traffic congestion. By analyzing these indicators, configure the second feature vector corresponding to the key nodes of traffic flow, which can intuitively reflect the degree of traffic congestion and the ability to restore smooth traffic in the high-conflict area.

[0067] For the potential conflict situations among different transportation modes at the key nodes of the traffic flow in the high-flow regions reflected by the first eigenvector, and the congestion conditions of the key nodes of the traffic flow in the high-conflict regions reflected by the second eigenvector, calculate the numerical values of their respective indicators. Then, according to different traffic scenarios and actual requirements, assign corresponding weights to the first eigenvector and the second eigenvector. For example, during the morning and evening rush hours on weekdays, since traffic congestion has a greater impact on the overall traffic, a higher weight can be assigned to the second eigenvector that reflects the congestion situation; while when the traffic flow is relatively stable at ordinary times, the weights of the two vectors can be appropriately balanced. Then, add the product of each indicator value in each eigenvector and its corresponding weight to obtain the weighted scores of the first eigenvector and the second eigenvector respectively. After that, perform a weighted summation on these two weighted scores to obtain a preliminary comprehensive score. In order to make this comprehensive score more accurately reflect the traffic flow characteristics, the comprehensive score can be further corrected and adjusted by combining historical traffic data and expert experience. For example, if a certain type of traffic conflict or congestion often occurs at a key node of the traffic flow in history, the influence of relevant factors can be appropriately increased in the comprehensive score. After such processing, the final obtained numerical value and the information it represents are the traffic flow characteristics corresponding to the key nodes of the traffic flow, which comprehensively consider potential conflicts and congestion conditions and can provide a reliable decision-making basis for work such as traffic signal timing optimization.

[0068] Embodiment 2, based on the same inventive concept as the multi-mode traffic signal optimization method driven by perception intelligence in the foregoing embodiment, as Figure 2 shown, the present application provides a multi-mode traffic signal optimization system driven by perception intelligence. The system in the embodiments of the present application and the method embodiments are based on the same inventive concept. Among them, the system includes:

[0069] A traffic flow data collection module 10, configured to collect traffic flow data, where the traffic flow data includes motor vehicle flow, non-motor vehicle flow, and pedestrian flow.

[0070] A multi-mode traffic flow model construction module 20, configured to construct a multi-mode traffic flow model, where the multi-mode traffic flow model is used to simulate the interaction behavior relationship among the behavior characteristics of motor vehicles, non-motor vehicles, and pedestrians.

[0071] A street lamp control signal acquisition module 30, configured to connect to a plurality of traffic signals in a target area to obtain a street lamp control signal.

[0072] A traffic flow change trend prediction module 40, configured to predict the traffic flow change trend based on the traffic flow data and the street lamp control signal by using the multi-mode traffic flow model.

[0073] The timing optimization scheme configuration module 50 is used to configure the timing optimization schemes for multiple traffic lights in the target area according to the prediction results of the traffic flow change trend.

[0074] Further, the system is also used to implement the following functions:

[0075] Obtain the historical traffic flow data in the target area; based on the historical traffic flow data, decompose the historical motor vehicle flow, historical non-motor vehicle flow, and historical pedestrian flow, and set the training data set; based on the training data set, train the multi-mode traffic flow model, and use the sliding window mechanism to dynamically update the prediction results of the traffic flow change trend.

[0076] Further, the system is also used to implement the following functions:

[0077] Deploy edge computing nodes according to multiple traffic lights in the target area; based on the edge computing nodes, generate local traffic light timing schemes and synchronize data with the central node; determine the timing optimization schemes for multiple traffic lights in the target area through the multiple local traffic light timing schemes uploaded to the central node.

[0078] Further, the system is also used to implement the following functions:

[0079] Take the motor vehicle passing efficiency, non-motor vehicle passing efficiency, and pedestrian passing efficiency as the joint optimization goal; according to the real-time traffic flow data, perform dynamic weight allocation on the joint optimization goal to obtain multiple timing adaptation schemes; screen the multiple timing adaptation schemes to determine the timing optimization schemes for multiple traffic lights in the target area.

[0080] Further, the system is also used to implement the following functions:

[0081] Introduce the pedestrian crossing demand; based on the pedestrian flow data and pedestrian crossing demand, rank the multiple timing adaptation schemes by priority and set the first screening condition.

[0082] Further, the system is also used to implement the following functions:

[0083] Mark the high-flow areas and high-conflict areas; based on the high-flow areas and high-conflict areas, identify the key traffic flow nodes and determine the traffic flow characteristics corresponding to the key traffic flow nodes; according to the traffic flow characteristics corresponding to the key traffic flow nodes, perform secondary screening on the multiple timing adaptation schemes and set the second screening condition.

[0084] Further, the system is also used to implement the following functions:

[0085] Identify key traffic flow nodes, where the key traffic flow nodes include road intersections, crosswalks, and non-motor vehicle lane intersections; at the key traffic flow nodes in the high-flow area, analyze potential conflicts between different traffic modes and configure the first feature vector corresponding to the key traffic flow nodes; at the key traffic flow nodes in the high-conflict area, analyze the queue lengths, waiting times, and queue dissipation speeds of vehicles and pedestrians, and configure the second feature vector corresponding to the key traffic flow nodes; based on the first feature vector and the second feature vector, determine the traffic flow characteristics corresponding to the key traffic flow nodes.

[0086] It should be noted that the above sequence of embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above specific embodiments of this specification have been described. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0087] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included within the protection scope of the present application.

[0088] This specification and the drawings are only exemplary descriptions of the present application and are considered to have covered any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. A multi-modal traffic signal optimization method driven by perception intelligence, characterized in that: The method comprises: Collecting traffic flow data, the traffic flow data including motor vehicle flow, non-motor vehicle flow, and pedestrian flow; Constructing a multi-modal traffic flow model, wherein the multi-modal traffic flow model is used to simulate the interactive behavior relationship between motor vehicle behavior characteristics, non-motor vehicle behavior characteristics, and pedestrian behavior characteristics; Connect multiple traffic lights in the target area to obtain street light control signals; Based on the traffic flow data and the street lamp control signal, using the multi-mode traffic flow model, predicting the traffic flow change trend; According to the prediction results of traffic flow change trend, the timing optimization plan of multiple traffic lights in the target area is configured.

2. The method according to claim 1, characterized in that Constructing a multi-modal traffic flow model, the method comprising: Obtain historical traffic flow data in the target area; Based on the historical traffic flow data, the historical traffic flow of motor vehicles, the historical traffic flow of non-motor vehicles, and the historical traffic flow of pedestrians are decomposed to set a training data set; Based on the training data set, the multi-mode traffic flow model is trained, and a sliding window mechanism is used to dynamically update the traffic flow change trend prediction result.

3. The method according to claim 2, characterized in that A timing optimization scheme for configuring a plurality of traffic lights in a target area, the method comprising: Deploy edge computing nodes based on multiple traffic lights in the target area; Based on the edge computing node, a local traffic light timing plan is generated, and data is synchronized with the central node; By uploading multiple local traffic signal light timing plans to the central node, a timing optimization plan for multiple traffic signal lights in a target area is determined.

4. The method according to claim 3, characterized in that Determining a timing optimization scheme for a plurality of traffic lights in a target area, the method comprising: Take the traffic efficiency of motor vehicles, non-motor vehicles and pedestrians as joint optimization goals; According to the real-time traffic flow data, dynamically weight the joint optimization objectives to obtain multiple timing adaptation schemes; The multiple timing adaptation schemes are screened to determine a timing optimization scheme for multiple traffic lights in a target area.

5. The method according to claim 4, characterized in that The plurality of timing adaptation schemes are screened to determine a timing optimization scheme for a plurality of traffic lights in a target area. The method includes: Introducing pedestrian crossing requirements; Based on pedestrian flow data and pedestrian crossing requirements, the multiple timing adaptation schemes are prioritized and a first filtering condition is set.

6. The method according to claim 5, characterized in that The method comprises: Mark high-traffic and high-conflict areas; Based on the high-flow area and the high-conflict area, identifying key traffic flow nodes, and determining traffic flow characteristics corresponding to the key traffic flow nodes; According to the traffic flow characteristics corresponding to the key traffic flow nodes, the multiple timing adaptation schemes are screened for the second time and a second screening condition is set.

7. The method according to claim 6, characterized in that Determining traffic flow characteristics corresponding to key traffic flow nodes, the method further includes: Determine key traffic flow nodes, wherein the key traffic flow nodes include road intersections, pedestrian crossings, and intersections of non-motorized vehicle lanes; At the key traffic flow nodes in the high-traffic area, analyzing potential conflicts between different traffic modes, and configuring first feature vectors corresponding to the key traffic flow nodes; At the key traffic flow nodes in the high-conflict area, analyzing the queue length, waiting time and queue dissipation speed of vehicles and pedestrians, and configuring a second eigenvector corresponding to the key traffic flow node; Based on the first feature vector and the second feature vector, traffic flow characteristics corresponding to key traffic flow nodes are determined.

8. A multi-mode traffic signal optimization system driven by perception intelligence, characterized in that: The system is used to implement the multi-mode traffic signal optimization method driven by perception intelligence according to any one of claims 1 to 7, and the system comprises: Traffic flow data collection module, used to collect traffic flow data, the traffic flow data includes motor vehicle flow, non-motor vehicle flow, and pedestrian flow; A multi-mode traffic flow model construction module is used to construct a multi-mode traffic flow model, wherein the multi-mode traffic flow model is used to simulate the interactive behavior relationship between motor vehicle behavior characteristics, non-motor vehicle behavior characteristics, and pedestrian behavior characteristics; A street light control signal acquisition module is used to connect multiple traffic lights in the target area to obtain street light control signals; A traffic flow change trend prediction module, used to predict the traffic flow change trend based on the traffic flow data and the street lamp control signal using the multi-mode traffic flow model; The timing optimization scheme configuration module is used to configure the timing optimization scheme of multiple traffic lights in the target area according to the prediction results of traffic flow change trend.

Images