Traffic light intelligent control system and method based on AI decision

Abstract

The invention provides an AI decision-making-based traffic light intelligent control system and method, and relates to the technical field of intelligent traffic, and the system comprises a network construction module which is used for constructing a bionic traffic neural network; the road network topology construction module is used for constructing a road network topology structure and mapping the road network topology structure to the bionic traffic neural network; the abnormal node identification module is used for collecting traffic monitoring data and projecting the traffic monitoring data to the bionic traffic neural network for pulse transmission, and identifying abnormal nodes; and the control parameter adjustment module is used for performing pulse transmission fitting through the bionic traffic neural network to obtain adjustment control parameters of the traffic lights at the intersections. The technical problem that traffic light regulation and control are difficult to adapt to complex traffic flow changes and emergencies in real time due to the lack of deep analysis on complex dynamics of a traffic system in the prior art is solved, the flexibility and dynamic adaptability of traffic light control are improved, and then the road traffic efficiency is improved.

Description

Technical Field

[0001] This application relates to the technical field of intelligent transportation, and particularly to an intelligent traffic light control system and method based on AI decision-making. Background Art

[0002] Traffic light control is an important part of urban traffic management. Its main purpose is to ensure smooth and safe road traffic by reasonably regulating the timing of traffic lights. Currently, intelligent traffic light control mainly relies on adaptive control and intelligent control based on optimization algorithms. The adaptive control method detects traffic flow based on induction coils or cameras and dynamically adjusts the signal duration. However, it usually only optimizes at a single intersection or in a small area, making it difficult to coordinate the traffic flow distribution of the entire traffic network. The intelligent control methods based on optimization algorithms (such as reinforcement learning, fuzzy control, etc.) mostly rely on isolated intersection traffic flow models or simple linear models, lacking a deep description of the complex dynamics of the entire traffic network. These methods all ignore the mutual influence between intersections, such as the spread of traffic congestion between adjacent intersections and the coordinated changes of vehicle flows in different directions. When dealing with emergencies (such as accidents, construction, or large-scale traffic flow fluctuations), there are problems of response lag and untimely adjustment, thus affecting the improvement of road traffic efficiency. Summary of the Invention

[0003] This application provides an intelligent traffic light control system and method based on AI decision-making, which solves the technical problem that in the prior art, due to the lack of in-depth analysis of the complex dynamics of the traffic system, the traffic light regulation is limited to local optimization and it is difficult to adapt to complex traffic flow changes and emergencies in real time, and achieves the technical effects of improving the flexibility and dynamic adaptability of traffic light control, optimizing the overall traffic flow distribution, and improving road traffic efficiency.

[0004] In view of the above problems, on the one hand, this application provides an intelligent traffic light control system based on AI decision-making. The system includes: a network construction module for constructing a bionic traffic neural network; a road network topology construction module for obtaining the road network distribution within the control area, constructing a road network topology structure, and mapping the road network topology structure to the bionic traffic neural network, where each intersection corresponds to a neuron and the topological connection relationship corresponds to the synaptic connection of neurons; an abnormal node identification module for collecting traffic monitoring data through monitoring devices, projecting it to the bionic traffic neural network according to the location of the traffic monitoring data for pulse transmission, and identifying abnormal nodes; a control parameter adjustment module for obtaining the adjustment control parameters of traffic lights at each intersection through pulse transmission fitting by the bionic traffic neural network based on the abnormal nodes.

[0005] On the other hand, the present application also provides an intelligent traffic light control method based on AI decision-making. The method includes: constructing a bionic traffic neural network; obtaining the road network distribution within the control area, constructing a road network topology structure, and mapping the road network topology structure to the bionic traffic neural network, where each intersection corresponds to a neuron and the topological connection relationship corresponds to the synaptic connection of the neuron; collecting traffic monitoring data through monitoring devices, projecting it according to the positioning of the traffic monitoring data into the bionic traffic neural network for pulse transmission, and identifying abnormal nodes; based on the abnormal nodes, performing pulse transmission fitting through the bionic traffic neural network to obtain the adjustment control parameters of the traffic lights at each intersection.

[0006] One or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0007] The bionic traffic neural network is constructed by the network construction module, simulating the neuron and synaptic connection relationship, and mapping the traffic flow to the pulse transmission mechanism of the neural network, providing the global optimization ability based on the neural network for subsequent traffic light control, enabling the system to learn and adapt to traffic dynamic changes from an overall perspective. The road network topology construction module obtains the road network distribution information within the control area and maps it to the bionic traffic neural network, making each intersection correspond to a neuron and the topological connection relationship correspond to the synaptic connection of the neuron, so as to be able to simulate the vehicle flow conduction relationship in the traffic network and capture the dynamic interaction effects between intersections. The abnormal node identification module collects real-time traffic monitoring data and uses the bionic traffic neural network for pulse transmission analysis to identify abnormal traffic nodes, so that traffic anomalies can be quickly discovered and transmitted as input signals to the neural network, enabling the system to have the real-time perception ability for emergencies. The control parameter adjustment module performs pulse transmission fitting using the bionic traffic neural network based on the abnormal node information to dynamically adjust the control parameters of the traffic lights at each intersection. Through this optimization method, the system can dynamically and adaptively adjust the signal duration within the global range, realizing the intelligent regulation of traffic lights, thereby improving the balance of traffic flow and the traffic efficiency.

[0008] In summary, the present application constructs a traffic flow model through a bionic traffic neural network, maps intersections to neurons, and topological connections to synaptic connections, realizing the dynamic modeling of the overall traffic state of the road network. Based on the pulse transmission mechanism, the system can collect traffic monitoring data in real time, identify abnormal traffic nodes, and adjust the traffic light signal parameters through global optimization, making the traffic light signal control more flexible and adaptable, capable of coordinating the signal timings of multiple intersections, optimizing the overall traffic flow distribution, improving the road traffic efficiency, and enhancing the adaptive ability to sudden traffic events.

[0009] The above description is only an overview of the technical solution of this application. In order to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of this application more obvious and understandable, the following specific embodiments of this application are given. Brief Description of the Drawings

[0010] Figure 1 It is a schematic structural diagram of an intelligent traffic light control system based on AI decision-making provided by an embodiment of this application.

[0011] Figure 2 It is a schematic flow diagram of an intelligent traffic light control method based on AI decision-making provided by an embodiment of this application.

[0012] Description of the reference numerals: Network construction module 10, road network topology construction module 20, abnormal node identification module 30, control parameter adjustment module 40. Detailed Embodiments

[0013] The embodiment of this application provides an intelligent traffic light control system and method based on AI decision-making, which solves the technical problem that in the prior art, due to the lack of in-depth analysis of the complex dynamics of the traffic system, the traffic light regulation is limited to local optimization and it is difficult to adapt to the complex traffic flow changes and emergencies in real time. Through the pulse transmission mechanism of the bionic traffic neural network, the technical effects of improving the flexibility and dynamic adaptability of traffic light control, optimizing the overall traffic flow distribution, and improving the road traffic efficiency are achieved.

[0014] Embodiment 1, as Figure 1 shown, the embodiment of this application provides an intelligent traffic light control system based on AI decision-making, and the system includes:

[0015] A network construction module 10, which is used to construct a bionic traffic neural network.

[0016] Specifically, the bionic traffic neural network is an artificial intelligence model inspired by the biological neural network. In this model, the traffic road network is abstracted as a neural network, where intersections correspond to neurons, road connections correspond to synapses, and the change in traffic flow is similar to the transmission of nerve impulses. The network construction module 10 uses computer programming tools, such as relevant neural network libraries in Python (such as PyTorch or TensorFlow), to create a neural network model by defining classes and functions, enabling it to represent the operating state of the traffic system and obtain the bionic traffic neural network. Specifically, first, neurons are defined, and traffic intersections are used as neurons in the neural network. For example, in the traffic network of an urban area, intersections A (crossroads), B (T-junctions), C (roundabouts), etc. can all exist as neurons. Then, synaptic connections are defined, and the connections between neurons are established using road network data. For example, assuming that intersection A is connected to B and C, then A→B and A→C correspond to the synaptic connections of the neural network. Next, the neural network parameters are initialized, and the connection weights between neurons are set. These weights can represent the traffic capacity of the road, traffic density, signal light status, etc. For example, if the traffic flow on the road A→B is greater than that on A→C, then the connection weight of A→B will be higher. Finally, the pulse transmission mechanism is simulated, and an information propagation is carried out using a spiking neural network to simulate the flow of traffic in the road network. For example, if the traffic flow at intersection A increases, its corresponding neuron will "trigger" a pulse, affecting the signal light adjustment at adjacent intersections B and C.

[0017] Constructing the bionic traffic neural network provides a basic analysis and decision-making framework for the entire intelligent traffic signal control system, enabling it to process complex relationships in the traffic system in a way similar to the biological neural network, making the entire intelligent traffic signal control system more intelligent and adaptable.

[0018] The road network topology construction module 20 is used to obtain the road network distribution within the control area, construct the road network topology structure, and map the road network topology structure to the bionic traffic neural network, where each intersection corresponds to a neuron, and the topological connection relationship corresponds to the neuron synaptic connection.

[0019] Specifically, obtain the road network distribution data within the control area from a GIS (Geographic Information System) or urban traffic management system, such as road structure, intersection locations, number of lanes, signal light positions, etc. By analyzing the connecting roads between these intersections, use graph theory methods (such as adjacency matrix or adjacency list) to represent the road network structure, where nodes represent intersections and edges represent roads, thereby generating the road network topology. During the construction process, specialized map drawing software or traffic planning software can be used to assist in completion, and these software can visually display the road network topology. Then map the road network topology within the control area to the previously constructed bionic traffic neural network, making neurons and synapses correspond to the actual intersection and road connection relationships. Among them, each intersection corresponds to a neuron, and the topological connection relationship corresponds to the neuron synapse connection.

[0020] By establishing the topological structure and mapping it to the bionic traffic neural network, the overall structure of the traffic system within the control area is accurately expressed, enabling the actual traffic flow within the control area to be simulated and analyzed in the bionic traffic neural network, facilitating global optimization and control of traffic lights based on the dynamic interaction of the entire network.

[0021] The abnormal node recognition module 30 is used to collect traffic monitoring data through monitoring devices, project it to the bionic traffic neural network according to the positioning of the traffic monitoring data for pulse transmission, and identify abnormal nodes.

[0022] Specifically, use monitoring devices such as traffic cameras, radars, induction coils, and intelligent traffic management systems to obtain real-time traffic monitoring data, including traffic flow, average vehicle speed, queue length, sudden accidents, etc. According to the location of the intersection, map the monitored traffic state data to the corresponding neurons in the bionic traffic neural network for pulse transmission simulation. In the bionic traffic neural network, traffic data can be used as a stimulus signal to trigger changes in the state of neurons, and the state changes of neurons are transmitted to other neurons (intersections) through pulse signals. The frequency, intensity, and timing of the pulses carry information about the traffic state, such as traffic flow and congestion level. When an abnormal situation (such as traffic congestion or an accident) occurs at a certain intersection, its corresponding neuron will emit pulse signals with different frequencies or intensities. By analyzing the results of pulse transmission, such as setting a certain threshold or algorithm, identify the nodes that are significantly different from the normal traffic pattern, and these nodes are the abnormal nodes.

[0023] Through abnormal node recognition, abnormal situations in the traffic system can be detected in a timely manner, such as the source of traffic congestion or potential locations of traffic accidents. By analyzing traffic monitoring data through pulse transmission in the bionic traffic neural network, the identification of abnormal nodes is made more accurate and efficient, providing key target nodes for subsequent control parameter adjustment, and helping to take measures in advance to relieve traffic pressure or avoid the spread of traffic congestion.

[0024] The control parameter adjustment module 40 is used to perform pulse transfer fitting through the bionic traffic neural network based on the abnormal node, and obtain the adjusted control parameters of the traffic lights at each intersection.

[0025] Specifically, based on the previously identified abnormal node, pulse transfer fitting is performed in the bionic traffic neural network. The change of the traffic state is simulated by the pulse signal in the bionic traffic neural network, and the parameters of the neurons (intersections) are adjusted to make the control mode of the traffic lights adapt to the current traffic conditions. For example, if an abnormal node represents a congested intersection, then in the neural network, starting from this node, the diffusion and influence of the traffic flow are simulated through pulse transfer, and it is calculated how the traffic lights at each relevant intersection should be adjusted to relieve this congestion situation. According to this simulation result, the adjusted control parameters of the traffic lights at each intersection are obtained, such as the duration of the traffic lights, the phase ratio, the priority, etc.

[0026] Based on the pulse calculation of the bionic neural network, it can dynamically adjust the traffic light control parameters according to the actual traffic anomalies in the management and control area, enabling the traffic light control to have global optimization ability and dynamic adaptability, and being able to quickly adjust the signals in case of emergencies, improving the road traffic efficiency.

[0027] Furthermore, the abnormal node recognition module 30 further includes:

[0028] The signal conversion module is used to convert the traffic monitoring data into pulse signals.

[0029] The pulse simulation fitting module is used to project the pulse signal converted from the traffic monitoring information into the corresponding bionic traffic neural network according to the monitoring and positioning information of the traffic monitoring data, and perform pulse simulation fitting according to the synaptic connection relationship of the neurons.

[0030] The matching and recognition module is used to perform matching in the bionic traffic neural network according to the pre-trained abnormal pulse parameters to identify the abnormal node.

[0031] Specifically, the signal conversion module obtains traffic monitoring data, which contains various types of traffic information, such as traffic flow, vehicle speed, queue length, abnormal events, etc. Then these data are converted into pulse signals according to the pre-set conversion rules. In the neural network model, the pulse signal is a data expression method for simulating neuron activities. The bionic traffic neural network adopts a spiking neural network and uses discrete pulse sequences to represent the change of the traffic state. Converting the traffic monitoring data into pulse signals enables these data to adapt to the information processing mechanism of the bionic traffic neural network, providing a suitable data input form for subsequent pulse simulation fitting and abnormal node recognition in the bionic traffic neural network.

[0032] The pulse simulation fitting module first locates the corresponding neuron positions in the bionic traffic neural network according to the monitoring location information of the traffic monitoring data (such as the location coordinates or topological ID of the intersection), and projects the converted pulse signals onto the corresponding neurons in the bionic traffic neural network. Then, according to the synaptic connection relationships between neurons, the pulse signals are propagated and interacted in the network, enabling the bionic traffic neural network to simulate the evolution of real traffic flow. For example, if an intersection (neuron) receives a pulse signal indicating high traffic flow, it will transmit this signal to the neurons of adjacent intersections according to certain rules (such as allocating signal intensity according to the importance of the connected roads) based on the connection relationships with adjacent intersections (adjacent neurons). Through pulse simulation fitting, it is possible to simulate the propagation and mutual influence of the traffic conditions represented by the traffic monitoring data in the bionic traffic neural network, which helps to understand more deeply the relationships between various intersections in the traffic system.

[0033] The abnormal pulse parameters refer to the pulse parameter standards representing traffic anomalies obtained through learning a large amount of abnormal traffic condition data. For example, high-frequency pulses (with frequencies much higher than the normal level) indicate severe traffic congestion; sudden pulse peaks (sharp increases within a short time) indicate traffic accidents or emergencies. The matching and recognition module performs matching operations in the bionic traffic neural network using the pre-trained abnormal pulse parameters. After the pulse simulation fitting module completes the propagation and interaction of pulse signals in the network, the matching and recognition module checks the pulse parameters at each neuron (intersection). For example, if the pre-trained pulse intensity range at a certain intersection in the abnormal situation is greater than or equal to 0.5, and the currently monitored pulse intensity at this intersection is 0.8, it indicates that there is an abnormal situation at this intersection. By traversing the neurons in the entire bionic traffic neural network for matching checks, the neurons whose pulse parameters meet the abnormal pulse parameter standards are identified, and the intersections corresponding to these neurons are the abnormal nodes. Compared with the traditional fixed-threshold detection method, performing matching and recognition in the bionic traffic neural network based on pre-trained abnormal pulse parameters can adaptively learn abnormal patterns, accurately find the nodes that may have traffic anomalies, improve the accuracy and efficiency of abnormal node recognition, and provide key target nodes for subsequent traffic control parameter adjustment.

[0034] Furthermore, the signal conversion module is also used to perform the following steps:

[0035] Identify the vehicle flow, congestion level, and monitoring time information of the traffic monitoring data; convert the vehicle flow, congestion level, and monitoring time information into pulse frequency, intensity, and timing.

[0036] Specifically, the signal conversion module first accurately identifies vehicle flow, congestion level, and monitoring time information from traffic monitoring data. The vehicle flow information can be directly obtained from the count data acquired by sensors (such as induction coils); the identification of the congestion level requires integrating various data, for example, calculated based on traffic flow and vehicle speed (such as congestion level = traffic flow / vehicle speed, this is just a simple example); the monitoring time information is directly obtained from the time record of the monitoring device.

[0037] In the bionic traffic neural network, the state of neurons is controlled by pulse signals. Therefore, it is necessary to map the vehicle flow, congestion level, and monitoring time information to the pulse frequency, intensity, and timing of neurons. Among them, the vehicle flow corresponds to the pulse frequency. The greater the vehicle flow, the higher the pulse frequency; the congestion level corresponds to the pulse intensity. The higher the congestion level, the stronger the amplitude of the neuron pulse; the pulse timing is dynamically adjusted in combination with the monitoring time information, which refers to the generation situation of the pulse signal at a specific time point or time period, reflecting the characteristics of traffic monitoring data in the time dimension. Exemplarily, first, a conversion relationship is set to convert the vehicle flow into the pulse frequency. For example, a benchmark vehicle flow value is determined based on historical traffic data. When the actual vehicle flow is higher than this benchmark value, for every certain number of additional vehicles, the pulse frequency increases by a certain value accordingly. For example, if the benchmark vehicle flow is 100 vehicles per hour, and for every additional 20 vehicles in the actual vehicle flow, the pulse frequency increases by 1 Hz. Then, the pulse intensity is determined according to the classification of the congestion level. The congestion level is divided into three levels: low, medium, and high, corresponding to pulse intensities of 0.2, 0.5, and 0.8 respectively. Finally, the monitoring time information is converted into the pulse timing. If the monitoring time is during the traffic peak period (such as 7:00 - 9:00 in the morning and 5:00 - 7:00 in the evening), then it is set that the probability of generating a pulse signal during this time period is higher or some characteristics of the pulse signal (such as pulse frequency or intensity) will change. For example, during the peak period, the pulse frequency increases by 20% on the original basis to represent the particularity of traffic during this time period.

[0038] The signal conversion module accurately expresses the key information in traffic monitoring data in the form of the frequency, intensity, and timing of pulse signals by performing the above steps, enabling a more comprehensive and detailed simulation and analysis of traffic conditions in the bionic traffic neural network, thereby improving the detection and response capabilities of the traffic signal intelligent control system to traffic anomalies.

[0039] Furthermore, the abnormal node recognition module 30 further includes:

[0040] A preset feature acquisition module for obtaining preset traffic abnormal features.

[0041] A training module for learning the pulse signal transmission mode and pulse parameters based on the historical case data of the preset traffic anomaly features to obtain pre-trained anomaly pulse parameters.

[0042] Specifically, the preset feature acquisition module obtains the preset traffic anomaly features by analyzing historical data, traffic management experience, etc. These preset anomaly features are some pre-defined characteristics that can represent traffic anomaly situations. For example, a sudden increase in vehicle flow, a sharp drop in vehicle flow, abnormal vehicle speed (such as a sudden deceleration of vehicles on a highway), an overly long green light duration but low traffic efficiency (such as a traffic light control imbalance), etc. Exemplarily, data mining techniques can be used to analyze a large amount of historical traffic data, such as using clustering algorithms to classify traffic data into normal and abnormal situations to discover potential preset traffic anomaly features. At the same time, the business rules and expert opinions of the traffic management department will also be referred to to ensure the acquisition of comprehensive and accurate preset traffic anomaly features.

[0043] The training module first collects the historical case data related to the preset traffic anomaly features. These historical case data include traffic monitoring data such as vehicle flow, vehicle speed, road occupancy, etc. when various preset traffic anomaly features occur, as well as the corresponding actual traffic conditions (such as whether there is congestion, an accident, etc.). According to the structure of the bionic traffic neural network, these historical case data are converted into the form of pulse signals and the pulse signals are transmitted in the network according to the synaptic connection relationship of the neurons. During the pulse signal transmission process, the training module observes and records the changes in the pulse signal transmission mode and pulse parameters. Among them, the pulse signal transmission mode is the way and rule of the pulse signal transmitted from one neuron to another in the bionic traffic neural network, including the propagation path of the signal, the attenuation or enhancement situation on different synaptic connections, etc. The pulse parameters include pulse frequency, intensity, timing, etc. Exemplarily, when there is a sudden increase in vehicle flow (a preset traffic anomaly feature), observe how the frequency and intensity of the pulse signal change in the bionic traffic neural network and how the signal propagates in the network. By learning a large amount of historical case data, gradually summarize the rules of the pulse signal transmission mode and pulse parameters under different preset traffic anomaly features, so as to obtain pre-trained anomaly pulse parameters. The pre-trained anomaly pulse parameters reflect the typical pulse signal characteristics in the case of traffic anomalies, providing a reliable judgment basis for the matching and recognition module, thus improving the performance and efficiency of the entire anomaly node recognition module 30.

[0044] Furthermore, the control parameter adjustment module 40 includes:

[0045] A target setting module for setting the adjustment control target of the anomaly node.

[0046] A parameter conversion module, which is used to convert the adjustment control target into pulse transfer parameters.

[0047] An optimization adjustment module, which takes the pulse transfer parameters as the target, minimizes the adjustment intersection range as the constraint condition, and performs optimization adjustment based on the bionic traffic neural network to obtain the adjustment control parameters of the traffic lights at each intersection.

[0048] Specifically, the main task of the target setting module is to set the adjustment control target for abnormal nodes, that is, to clarify the optimization direction of traffic light adjustment. The adjustment control target is usually set according to the principles of alleviating congestion, balancing traffic flow, and reducing global impact. When setting the target, it is necessary to analyze according to the specific situation of the abnormal node. For example, when the highway is congested, it is necessary to improve the traffic efficiency and reduce the green light interval; when an accident occupies the road, it is necessary to increase the green light time of the bypass to guide vehicles to detour; when the traffic flow is unbalanced, the green light of the lane with less traffic flow should be appropriately shortened.

[0049] The parameter conversion module converts the adjustment control target set by the target setting module into pulse transfer parameters. Pulse transfer parameters are related parameters used to describe the pulse signal transmission process in the bionic traffic neural network, and these parameters are associated with the adjustment control target of the traffic lights. For example, the frequency, intensity, propagation direction of the pulse, etc. Exemplarily, if the adjustment control target is to increase the green light time, it is necessary to increase the pulse duration; if the adjustment control target is to reduce the red light time, it is necessary to reduce the pulse interval; if the adjustment control target is to allow more vehicles to pass, it is necessary to increase the pulse frequency; if the adjustment control target is to control the influence range of a certain intersection, it is necessary to limit the pulse propagation range.

[0050] The optimization adjustment module first obtains the pulse transfer parameters converted by the parameter conversion module, and sets the minimization of the adjusted intersection range as a constraint condition. That is, when adjusting the traffic light control parameters to solve the traffic problems of abnormal nodes, the number of intersections involved in the adjustment should be minimized as much as possible to reduce the impact range on the surrounding traffic. In the bionic traffic neural network, optimization algorithms (such as genetic algorithms and particle swarm optimization algorithms) are used to simulate different combinations of traffic light adjustment control parameters within the adjusted intersection range. The adjusted parameters include the extended green light time, the shortened red light time, etc. According to the set pulse transfer parameter target, the adjustment effect under each parameter combination is calculated, such as the change in traffic flow, the increase in average vehicle speed, and the decrease in road occupancy. During the optimization process, the traffic lights at the intersections closest to the abnormal node are adjusted first. When the parameter adjustment at the intersections near the abnormal node cannot meet the adjustment control target, the adjustment control target range is expanded outward. For example, for different combinations of the traffic light duration and phase at adjacent intersections, observe the transmission of pulse signals in the bionic traffic neural network and the corresponding improvement effect of the traffic condition. By comparing the performance indicators of different parameter combinations, select the parameter combination that best meets the pulse transfer parameter target and has the smallest adjusted intersection range as the adjustment control parameters for the traffic lights at each intersection, minimize the unnecessary impact on the surrounding traffic as much as possible, improve the efficiency and accuracy of traffic management, and achieve more intelligent traffic light control.

[0051] Furthermore, the optimization adjustment module includes:

[0052] The neuron positioning module is used to locate the excitatory neurons and correlation neurons according to the abnormal node. The excitatory neurons are the neurons corresponding to the intersections with abnormal traffic characteristics, and the correlation neurons are the neurons corresponding to the intersections adjacent to the excitatory neurons.

[0053] The target analysis module is used to analyze the adjustment target according to the excitatory neurons to obtain the required adjustment control target.

[0054] The pulse parameter adjustment module is used to send the required adjustment control target to the correlation neurons, and adjust the pulse parameters of the correlation neurons according to the required adjustment control target to obtain the adjustment control parameters for the traffic lights at the intersections of the excitatory neurons and correlation neurons.

[0055] Specifically, the neuron positioning module first finds the corresponding neurons in the bionic traffic neural network through the identifiers of abnormal nodes (such as intersection numbers or topology IDs), and marks them as excitatory neurons. Then, according to the preset network topology structure, it determines the neurons adjacent to these excitatory neurons and marks them as relevant neurons. By accurately positioning the excitatory neurons and relevant neurons, it provides clear operation objects for subsequent adjustment target analysis and pulse parameter adjustment, enabling subsequent operations to focus on the neurons corresponding to the intersections directly related to the abnormal nodes, and improving the pertinence and efficiency of the adjustment process.

[0056] The target analysis module obtains the traffic monitoring data of the abnormal nodes corresponding to the excitatory neurons, and analyzes the specific required adjustment control objectives, that is, the specific objectives that need to adjust the relevant intersections (the intersections corresponding to the excitatory neurons and relevant neurons), according to these traffic monitoring data and the pulse transmission parameters corresponding to the preset adjustment control objectives. If the traffic flow at the intersection corresponding to the excitatory neuron exceeds the normal capacity and the vehicle speed is extremely low, the required adjustment control objective is analyzed as reducing the traffic flow entering this intersection by 30%, and at the same time improving the vehicle evacuation ability of the adjacent intersections. The target analysis module obtains the required adjustment control objective through the adjustment target analysis of the excitatory neurons, making the traffic control adjustment more specific and targeted, helping to more accurately solve the traffic problems of abnormal nodes, improving the effectiveness of traffic management, and providing a clear adjustment direction for the pulse parameter adjustment module.

[0057] The pulse parameter adjustment module first sends the required adjustment control objective obtained by the target analysis module to the relevant neurons. Then it adjusts the pulse parameters of the relevant neurons according to the required adjustment control objective. For example, if the required adjustment control objective is to reduce the traffic flow entering the intersection corresponding to the excitatory neuron (the intersection with abnormal traffic characteristics), then the pulse parameter adjustment module will reduce the pulse frequency of the relevant neurons adjacent to this excitatory neuron (the neurons corresponding to the adjacent intersections). At the same time, it adjusts parameters such as pulse intensity or timing to optimize the vehicle evacuation method of the adjacent intersections.

[0058] By adjusting the pulse parameters of the relevant neurons, the adjustment control parameters of the traffic lights at the intersections of the excitatory neurons and relevant neurons are obtained, which can effectively control and guide the traffic flow according to the required adjustment control objective, realize the intelligent management of the traffic system under the framework of the bionic traffic neural network, improve the ability to cope with abnormal traffic conditions, and optimize the traffic coordination between intersections.

[0059] Furthermore, the control parameter adjustment module 40 further includes:

[0060] A node marking module, configured to identify and mark traffic parameter holders and their traffic equilibrium weights based on the abnormal nodes.

[0061] A function construction module, configured to construct traffic control benefit functions for each traffic parameter holder.

[0062] An equilibrium optimization module, configured to perform equilibrium optimization search through a game theory optimization model according to the traffic equilibrium weights and the traffic control benefit functions, so as to obtain the adjustment control parameters of the traffic lights at each intersection.

[0063] Specifically, the node marking module identifies, through monitoring devices and historical data analysis, the vehicle groups or traffic flow directions that have the greatest impact on the current traffic condition, marks them as traffic parameter holders, and simultaneously generates the traffic equilibrium weights corresponding to each traffic parameter holder. The traffic equilibrium weights reflect the importance or influence of different traffic parameter holders in the traffic equilibrium state. According to historical data, if a certain type of vehicle has a relatively large average traffic volume at this intersection and has a greater impact on the overall traffic fluency, a higher traffic equilibrium weight is assigned. For example, at a certain intersection, buses carry more passengers and have a greater impact on the overall traffic efficiency and public interests, so they have a higher traffic equilibrium weight. By identifying and marking traffic parameter holders and their traffic equilibrium weights, a basis is provided for constructing traffic control benefit functions and subsequent equilibrium optimization, so that the subsequent optimization process can consider the needs and influences of different traffic participants, thereby formulating more fair and reasonable traffic control strategies.

[0064] The traffic control benefit function is a mathematical function that describes the benefits obtained by each traffic parameter holder under traffic control. This function is usually related to traffic parameters (such as traffic volume, vehicle speed, waiting time, etc.) and traffic control measures (such as traffic light duration, lane allocation, etc.). The function construction module constructs traffic control benefit functions for each marked traffic parameter holder. First, traffic parameters and traffic control variables related to each traffic parameter holder are determined. Then, the function is constructed according to traffic engineering principles, traffic behavior analysis, and empirical data. For example, for the private car traffic parameter holder, its traffic control benefit function can be set as a function of travel time, and the travel time is related to factors such as the traffic light duration and traffic volume at the intersection. The traffic control benefit function is constructed as follows: where T is the travel time, Q is the traffic volume, V is the vehicle speed, r is the red light duration, and g is the green light duration.

[0065] The equilibrium optimization module obtains the traffic equilibrium weights marked by the node marking module and the traffic control benefit function constructed by the function construction module. Then, this information is substituted into the game theory optimization model. In the game theory optimization model, according to the interaction relationships among different traffic parameters, equilibrium optimization search is carried out through methods such as iterative calculation. Exemplarily, there are three traffic parameters: private cars, buses, and pedestrians. The game theory optimization model considers the different interest demands of the three at intersections (represented by the traffic control benefit function) and their respective traffic equilibrium weights, and continuously adjusts traffic control parameters such as the duration of traffic lights at each intersection until an equilibrium state is reached, that is, no traffic parameter can obtain more benefits by unilaterally changing its strategy (such as running a red light, illegal lane change, etc.). Finally, the adjustment control parameters of traffic lights at each intersection are determined. In the actual implementation process, game theory analysis software such as Gambit can be used, or general mathematical modeling and optimization software such as the optimization toolbox in MATLAB can be used to implement the equilibrium optimization process. This game theory-based method considers the interest equilibrium of different traffic parameters, makes the traffic control strategy more scientific and reasonable, can improve the overall traffic efficiency, reduce traffic conflicts, and meet the needs of different traffic participants.

[0066] In summary, the intelligent traffic light control system based on AI decision provided by the embodiments of the present application has the following technical effects:

[0067] In the embodiments of the present application, by constructing a bionic traffic neural network, the overall modeling and dynamic perception of the road network traffic state are realized. The road network topology construction module 20 maps the actual road network structure into the neural network, enabling the system to accurately represent the complexity of the traffic network. The abnormal node recognition module 30 monitors traffic data in real time, discovers and processes abnormal situations in a timely manner, and improves the response ability of the system. The control parameter adjustment module 40 dynamically adjusts the traffic light control parameters based on the recognition results of abnormal nodes to optimize the traffic flow distribution. Overall, this technical solution significantly improves the flexibility and dynamic adaptability of traffic light control, can adapt to complex traffic flow changes and emergencies in real time, effectively optimize the overall traffic flow distribution, and thus improve the road traffic efficiency, providing an innovative and effective solution for modern urban traffic management.

[0068] Embodiment 2, as Figure 2 shown, based on the same inventive concept as in the foregoing Embodiment 1, the embodiments of the present application provide an intelligent traffic light control method based on AI decision, and the method includes:

[0069] Step S1: Construct a bionic traffic neural network.

[0070] Step S2: Obtain the road network distribution within the control area, construct a road network topology structure, and map the road network topology structure to the bionic traffic neural network, where each intersection corresponds to a neuron, and the topological connection relationship corresponds to the synaptic connection of neurons.

[0071] Step S3: Collect traffic monitoring data through monitoring devices, project the traffic monitoring data according to its positioning into the bionic traffic neural network for pulse transmission, and identify abnormal nodes.

[0072] Step S4: Based on the abnormal nodes, perform pulse transmission fitting through the bionic traffic neural network to obtain the adjustment control parameters of the traffic lights at each intersection.

[0073] Further, step S3 of the embodiment of the present application includes:

[0074] Convert the traffic monitoring data into pulse signals; project the pulse signals after converting the traffic monitoring information into the corresponding bionic traffic neural network according to the monitoring positioning information of the traffic monitoring data, and perform pulse simulation fitting according to the synaptic connection relationship of neurons; match in the bionic traffic neural network according to the abnormal pulse parameters obtained through pre-training to identify the abnormal nodes.

[0075] Further, converting the traffic monitoring data into pulse signals includes:

[0076] Identify the vehicle flow, congestion level, and monitoring time information of the traffic monitoring data; convert the vehicle flow, congestion level, and monitoring time information into pulse frequency, intensity, and timing.

[0077] Further, before matching in the bionic traffic neural network according to the abnormal pulse parameters obtained through pre-training, it includes:

[0078] Obtain preset traffic abnormal features; perform learning on the pulse signal transmission mode and pulse parameters according to the historical case data of the preset traffic abnormal features to obtain pre-trained abnormal pulse parameters.

[0079] Further, step S4 of the embodiment of the present application includes:

[0080] Set the adjustment control target for the abnormal nodes; convert the adjustment control target into pulse transmission parameters; with the pulse transmission parameters as the target and minimizing the adjustment intersection range as the constraint condition, perform optimization adjustment based on the bionic traffic neural network to obtain the adjustment control parameters of the traffic lights at each intersection.

[0081] Further, obtaining the adjustment control parameters of the traffic lights at each intersection further includes:

[0082] Based on the abnormal node, locate the excitatory neurons and the correlation neurons. The excitatory neurons are the neurons corresponding to the intersections with abnormal traffic characteristics, and the correlation neurons are the neurons corresponding to the intersections adjacent to the excitatory neurons; perform adjustment target analysis according to the excitatory neurons to obtain the required adjustment control target; send the required adjustment control target to the correlation neurons, and adjust the pulse parameters of the correlation neurons according to the required adjustment control target to obtain the adjustment control parameters of the traffic lights at each intersection of the excitatory neurons and the correlation neurons.

[0083] Further, obtaining the adjustment control parameters of the traffic lights at each intersection further includes:

[0084] Based on the abnormal node, identify the traffic parameter markers and their traffic equilibrium weights; construct the traffic control benefit functions of each traffic parameter marker; according to the traffic equilibrium weights and the traffic control benefit functions, perform equilibrium optimization search through a game theory optimization model to obtain the adjustment control parameters of the traffic lights at each intersection.

[0085] Through the foregoing detailed description of a traffic light intelligent control system based on AI decision-making in this specification, those skilled in the art can clearly know a traffic light intelligent control method in this embodiment. For the method disclosed in Embodiment 2, since it corresponds to the system disclosed in Embodiment 1, it has corresponding execution steps and beneficial effects. For the relevant parts, refer to the description of the system part.

[0086] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An intelligent traffic light control system based on AI decision-making, characterized in that: include: Network building module, used to build bionic traffic neural network; A road network topology construction module is used to obtain the road network distribution in the control area, construct a road network topology structure, and map the road network topology structure to the bionic traffic neural network, wherein each intersection corresponds to a neuron, and the topological connection relationship corresponds to the neuron synaptic connection; An abnormal node identification module is used to collect traffic monitoring data through monitoring equipment, project the traffic monitoring data into the bionic traffic neural network for pulse transmission according to the location of the traffic monitoring data, and identify abnormal nodes; The control parameter adjustment module is used to perform pulse transfer fitting based on the abnormal node through the bionic traffic neural network to obtain the adjustment control parameters of the traffic lights at each intersection.

2. The traffic light intelligent control system based on AI decision-making according to claim 1 is characterized in that: The abnormal node identification module also includes: A signal conversion module, used for converting the traffic monitoring data into a pulse signal; A pulse simulation fitting module is used to project the pulse signal converted from the traffic monitoring information into the corresponding bionic traffic neural network according to the monitoring positioning information of the traffic monitoring data, and perform pulse simulation fitting according to the synaptic connection relationship of the neurons; The matching and identification module is used to match the abnormal pulse parameters obtained by pre-training in the bionic traffic neural network and identify the abnormal nodes.

3. The traffic light intelligent control system based on AI decision-making according to claim 2 is characterized in that: The signal conversion module is also used to perform the following steps: Identify vehicle flow, congestion level, and monitoring time information of the traffic monitoring data; The vehicle flow, congestion level, and monitoring time information are converted into pulse frequency, intensity, and timing.

4. The traffic light intelligent control system based on AI decision-making according to claim 3 is characterized in that: The abnormal node identification module also includes: A preset feature acquisition module, used to obtain preset traffic anomaly features; The training module is used to learn the pulse signal transmission mode and pulse parameters according to the historical case data of the preset traffic abnormality characteristics to obtain pre-trained abnormal pulse parameters.

5. The traffic light intelligent control system based on AI decision-making according to claim 1 is characterized in that: The control parameter adjustment module comprises: A target setting module is used to set the adjustment control target of abnormal nodes; A parameter conversion module, used for converting the adjustment control target into a pulse transfer parameter; The optimization adjustment module is used to take the pulse transfer parameter as the target, minimize the adjustment intersection range as the constraint condition, perform optimization adjustment based on the bionic traffic neural network, and obtain the adjustment control parameters of the traffic lights at each intersection.

6. The traffic light intelligent control system based on AI decision-making according to claim 5 is characterized in that: The optimization adjustment module comprises: A neuron positioning module, used to locate excitatory neurons and correlation neurons according to the abnormal node, wherein the excitatory neurons are neurons corresponding to intersections where abnormal traffic characteristics occur, and the correlation neurons are neurons corresponding to intersections adjacent to the excitatory neurons; A target parsing module, used for performing adjustment target parsing according to the excitatory neurons to obtain a demand adjustment control target; The pulse parameter adjustment module is used to send the demand adjustment control target to the correlation neuron, adjust the pulse parameters of the correlation neuron according to the demand adjustment control target, and obtain the adjustment control parameters of the traffic lights at each intersection of the excitatory neurons and the correlation neurons.

7. The traffic light intelligent control system based on AI decision-making according to claim 1 is characterized in that: The control parameter adjustment module also includes: A node marking module, for identifying and marking traffic parameters and their traffic balancing weights based on the abnormal nodes; Function construction module, used to construct traffic control interest function of each traffic parameter owner; The equilibrium optimization module is used to perform equilibrium optimization search based on the traffic equilibrium weight and the traffic control benefit function through a game theory optimization model to obtain the adjustment control parameters of the traffic lights at each intersection.

8. A traffic light intelligent control method based on AI decision-making, characterized in that: The method is performed by a traffic light intelligent control system based on AI decision-making according to any one of claims 1 to 7, comprising: Constructing a bionic traffic neural network; Obtaining the road network distribution within the control area, constructing a road network topology structure, and mapping the road network topology structure to the bionic traffic neural network, wherein each intersection corresponds to a neuron, and the topological connection relationship corresponds to the neuron synaptic connection; Collect traffic monitoring data through monitoring equipment, project the traffic monitoring data into the bionic traffic neural network according to the location of the traffic monitoring data for pulse transmission, and identify abnormal nodes; Based on the abnormal nodes, pulse transfer fitting is performed through the bionic traffic neural network to obtain adjustment control parameters of traffic lights at various intersections.

Images