AI decision-aided sudden traffic event scheduling decision-making method and system

Abstract

The invention provides an AI decision-assisted sudden traffic event scheduling decision-making method and system, and relates to the technical field of traffic decision-making, and the method comprises the steps: connecting a data collection system, and collecting traffic data and environment data as real-time data; analyzing and monitoring abnormal data in the real-time data by using a time sequence, and identifying a traffic event; using the influence range prediction model to predict the influence range of the traffic event; resource scheduling and traffic control are carried out based on the influence range, and a scheduling strategy is generated; and sending the scheduling strategy to an execution unit for strategy execution. According to the invention, the technical problems in the prior art that the response speed of the scheduling decision is slow and the flexibility is insufficient when the emergency traffic event occurs, the timeliness of traffic event processing and the scientificity of the scheduling strategy are influenced, and the low efficiency of urban traffic management and the traffic jam problem are aggravated can be solved; the technical targets of quick response and intelligent scheduling decision making are achieved, and the technical effects of improving urban traffic management efficiency and relieving traffic jam are achieved.

Description

Technical Field

[0001] This application relates to the technical field of traffic decision-making, and particularly to a scheduling decision-making method and system for sudden traffic events assisted by AI decision-making. Background Art

[0002] In the field of urban traffic management, although a variety of technical means have been applied to the scheduling decision-making of sudden traffic events, there are still many deficiencies in the existing technologies.

[0003] Firstly, traditional traffic management mostly relies on manual judgment and fixed-mode management systems, which often cannot make effective scheduling decisions quickly when faced with sudden traffic events. Secondly, the existing traffic information collection and processing methods are lacking in real-time and accuracy, resulting in the timeliness and scientific nature of scheduling decisions being affected.

[0004] In summary, in the existing technology, due to the excessive reliance on manual judgment and fixed-mode management systems, the response speed of scheduling decisions is slow and the flexibility is insufficient when faced with sudden traffic events, further affecting the timeliness of traffic event handling and the scientific nature of scheduling strategies, thus exacerbating the technical problems of low efficiency in urban traffic management and traffic congestion. Summary of the Invention

[0005] The purpose of this application is to provide a scheduling decision-making method and system for sudden traffic events assisted by AI decision-making, so as to solve the technical problems in the existing technology that due to the excessive reliance on manual judgment and fixed-mode management systems, the response speed of scheduling decisions is slow and the flexibility is insufficient when faced with sudden traffic events, further affecting the timeliness of traffic event handling and the scientific nature of scheduling strategies, thus exacerbating the technical problems of low efficiency in urban traffic management and traffic congestion.

[0006] In view of the above problems, this application provides a scheduling decision-making method and system for sudden traffic events assisted by AI decision-making.

[0007] In the first aspect, this application provides a scheduling decision-making method for sudden traffic events assisted by AI decision-making, which is implemented through a scheduling decision-making system for sudden traffic events assisted by AI decision-making, and includes: connecting to a data collection system to collect traffic data and environmental data as real-time data; using time series analysis to monitor abnormal data in the real-time data and identify traffic events; using an impact range prediction model to predict the impact range of the traffic events; performing resource scheduling and traffic control based on the impact range to generate a scheduling strategy; and sending the scheduling strategy to an execution unit for strategy execution.

[0008] In a second aspect, the present application also provides an AI decision - assisted sudden traffic event scheduling decision - making system for implementing the AI decision - assisted sudden traffic event scheduling decision - making method as described in the first aspect, including: a data collection module for connecting to a data collection system to collect traffic data and environmental data as real - time data; an anomaly analysis module for using time - series analysis to monitor abnormal data in the real - time data and identify traffic events; a prediction module for using an impact - range prediction model to predict the impact range of the traffic event; a scheduling module for performing resource scheduling and traffic control based on the impact range and generating a scheduling strategy; and a sending module for sending the scheduling strategy to an execution unit for strategy execution.

[0009] The technical solutions provided in the present application have at least the following technical effects or advantages: By connecting to a data collection system, collecting traffic data and environmental data as real - time data; using time - series analysis to monitor abnormal data in the real - time data and identify traffic events; using an impact - range prediction model to predict the impact range of the traffic event; performing resource scheduling and traffic control based on the impact range and generating a scheduling strategy; and sending the scheduling strategy to an execution unit for strategy execution. That is to say, by achieving the technical goal of rapid response and intelligent scheduling decision - making, the technical effect of improving urban traffic management efficiency and alleviating traffic congestion is achieved.

[0010] The above description is only an overview of the technical solutions of the present application. In order to be able to more clearly understand the technical means of the present application, it can be implemented according to the content of the specification. And in order to make the above - mentioned and other purposes, features and advantages of the present application more obvious and understandable, the following specific embodiments of the present application are specifically given. It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings described below are only exemplary, and for those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0012] Figure 1 It is a flowchart of the AI decision - assisted sudden traffic event scheduling decision - making method of the present application;

[0013] Figure 2 It is a structural diagram of the AI decision - assisted sudden traffic event scheduling decision - making system of the present application.

[0014] Description of reference numerals: data acquisition module 11, anomaly analysis module 12, prediction module 13, scheduling module 14, sending module 15. Specific implementation manners

[0015] By providing a scheduling decision-making method and system for sudden traffic events assisted by AI decision-making, the present application solves the technical problems existing in the prior art that due to over-reliance on manual judgment and a management system with a fixed mode, the response speed of scheduling decisions in the face of sudden traffic events is slow and the flexibility is insufficient, further affecting the timeliness of traffic event handling and the scientific nature of scheduling strategies, thus exacerbating the technical problems of low efficiency of urban traffic management and traffic congestion. The technical goal of achieving rapid response and intelligent scheduling decision-making is realized, and the technical effect of improving the efficiency of urban traffic management and alleviating traffic congestion is achieved.

[0016] Next, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without making creative efforts belong to the scope of protection of the present application. Additionally, it should be noted that for the sake of description, only the parts related to the present application are shown in the accompanying drawings rather than all of them.

[0017] Embodiment 1, please refer to the attached Figure 1 , the present application provides a scheduling decision-making method for sudden traffic events assisted by AI decision-making, which is applied to a scheduling decision-making system for sudden traffic events assisted by AI decision-making, and specifically includes the following steps:

[0018] Step S1: Connect to a data acquisition system to collect traffic data and environmental data as real-time data.

[0019] Specifically, connecting to a data acquisition system means connecting various hardware devices, such as cameras, sensors, GPS devices, etc., to a central processing system through wired or wireless networks to form a network capable of obtaining information in real time. Collecting real-time data can reflect the current road conditions and weather conditions, providing the latest basis for subsequent analysis and decision-making. Among them, traffic data includes video streams, traffic volumes, vehicle speeds, floating cars, vehicle types, and passing time data, which can comprehensively reflect the real-time conditions of roads. Environmental data includes precipitation, visibility, temperature, and icing state data, which can help predict and respond to the impact of weather on traffic.

[0020] Step S2: Use time series analysis to monitor abnormal data in the real-time data and identify traffic events.

[0021] Specifically, traffic data is processed and analyzed. Time series data refers to data points arranged in chronological order, such as the traffic flow per minute or the vehicle speed per hour. The core objective of time series analysis is to discover regular changes and outliers in the data, thereby providing a scientific basis for traffic management. For example, through time series analysis, the traffic management center of a certain city can monitor the traffic flow changes across the city in real time and detect abnormal situations promptly.

[0022] Monitoring abnormal data in real-time data means continuously observing the traffic data collected in real-time based on time series analysis to find data points that do not conform to normal patterns. Abnormal data usually manifests as sudden drops or surges in traffic flow, which may be caused by traffic accidents, road construction, or temporary events. For example, the real-time traffic flow data of a certain road section shows that the traffic flow suddenly drops from 100 vehicles per minute to 20 vehicles per minute, and this abnormal data may indicate a traffic accident on that road section.

[0023] Identifying traffic events means determining the specific types of traffic events corresponding to the abnormal data based on the characteristics of the abnormal data. Traffic events include traffic accidents, road closures, temporary events, etc., and each event has its unique data characteristics. For example, traffic accidents usually lead to a sudden drop in traffic flow and a decrease in vehicle speed, while temporary events may cause a sudden increase in traffic flow and fluctuations in vehicle speed. By analyzing the specific manifestations of abnormal data, the type of traffic event can be accurately identified. For example, the abnormal data of a certain road section shows a sudden drop in traffic flow and the vehicle speed drops below 5 kilometers per hour. Combining with video stream data, it is confirmed that a two-vehicle rear-end collision has occurred on that road section.

[0024] Step S3: Use the impact range prediction model to predict the impact range of the traffic event.

[0025] Specifically, using the impact range prediction model means using a trained model to predict the possible impacts of current or future traffic events. This model can identify the impact patterns of different event types on traffic flow by learning historical traffic event data. For example, the traffic management center of a certain city uses the impact range prediction model to be able to predict the congestion impact of a certain traffic accident on surrounding road sections in real time.

[0026] Predicting the impact range of a traffic event means calculating through the model the spatial area and time duration that the traffic event may affect. The spatial range is usually measured in distance. For example, a certain accident may cause congestion on road sections within a radius of 500 meters around it; the time range is usually measured in minutes or hours. For example, a certain congestion may last for 30 minutes. For example, a certain model predicts that a certain traffic accident will cause congestion on road sections within a radius of 300 meters around it, and the duration is about 25 minutes.

[0027] Step S4: Based on the influence scope, perform resource scheduling and traffic control to generate a scheduling strategy.

[0028] Specifically, according to the size of the influence scope, reasonably allocate resources such as manpower and material resources. Traffic control refers to the management of traffic flow, such as adjusting traffic lights and setting up temporary traffic signs. Generating a scheduling strategy means formulating a set of specific operation plans or schemes to effectively respond to traffic incidents. For example, when a traffic accident occurs at a busy intersection, its influence scope may include several surrounding blocks. According to this influence scope, resource scheduling may include dispatching traffic police and rescue vehicles in the nearby area, while traffic control may involve temporarily closing some lanes or adjusting the traffic light cycle.

[0029] Performing resource scheduling and traffic control based on the influence scope is to reasonably allocate resources and manage traffic flow on the basis of understanding the impact of the incident, so as to generate a scheduling strategy. For example, if the influence scope of the accident is small, perhaps only two rescue vehicles and several traffic police need to be dispatched; while if the influence scope is large, maybe ten rescue vehicles and dozens of traffic police are required, and at the same time, more comprehensive control of the surrounding traffic is needed.

[0030] Step S5: Send the scheduling strategy to the execution unit for strategy execution.

[0031] Specifically, sending the scheduling strategy to the execution unit for strategy execution means transmitting the specific plans or schemes of resource scheduling and traffic control to the designated execution unit. The execution unit refers to the entity or system responsible for implementing the scheduling strategy, which can be a traffic control system, a computer server, or a manual operation team, etc. Performing strategy execution means starting to implement specific operations or measures according to the content of the scheduling strategy. For example, once the scheduling strategy is formulated in the traffic management center, it needs to be sent to the traffic signal control system at the intersection through the computer system, which will adjust the signal timing according to the scheduling strategy.

[0032] The scheduling decision method for sudden traffic incidents assisted by AI decision-making is applied to the scheduling decision system for sudden traffic incidents assisted by AI decision-making, which can achieve the technical goals of rapid response and intelligent scheduling decision-making, and achieve the technical effects of improving urban traffic management efficiency and alleviating traffic congestion.

[0033] Furthermore, this application also includes: The traffic data includes video stream data, vehicle flow data, vehicle speed data, floating car data, vehicle type data, and passing time data, and the environmental data includes precipitation data, visibility data, temperature data, and icing state data.

[0034] Specifically, a connected data acquisition system refers to networking various hardware devices such as sensors, cameras, GPS devices, etc. with a central processing system to form a network capable of obtaining information in real time. The main task of this system is to collect traffic data and environmental data and transmit this data to the analysis platform in real time. Real-time data means that this information is updated instantaneously, capable of reflecting the current road conditions and weather conditions, providing the latest basis for subsequent analysis and decision-making.

[0035] Among them, traffic data includes various types of information. Video stream data is the real-time picture of the road captured by cameras, which can visually display traffic conditions, such as whether there are accidents or congestion. Traffic volume data refers to the number of vehicles passing through a certain section within a unit of time, usually counted per minute or per hour. Vehicle speed data is the driving speed of vehicles on the road, which can help determine whether there is congestion or speeding. Floating car data is the location and speed information uploaded in real time by vehicles equipped with GPS (such as taxis, buses), which can cover a wider road network. Vehicle type data refers to the type of vehicle, such as cars, trucks or buses, and this information helps analyze the impact of different vehicles on traffic. Travel time data refers to the time required for a vehicle to pass through a certain section, which can be used to evaluate the traffic efficiency of the road.

[0036] Next, environmental data is also an important part of traffic management. Precipitation data refers to the current or predicted rainfall, usually in millimeters, which can help determine whether there may be water accumulation or slipperiness on the road. Visibility data refers to the visible distance in the air, usually in meters, and low visibility may affect driving safety. Temperature data refers to the current air temperature, usually in degrees Celsius, and extreme temperatures may cause the road surface to freeze or vehicle failures. Ice state data refers to information on whether the road surface is frozen, usually obtained through road surface sensors, and the ice state will significantly increase the risk of traffic accidents.

[0037] Furthermore, this application also includes: setting a cycle length according to traffic characteristics, where the cycle length includes a daily cycle and a weekly cycle; setting a trend smoothing window based on the product of the cycle length and a preset multiple; using the cycle length and the trend smoothing window as decomposition parameters to decompose the real-time data to obtain a residual sequence; performing anomaly determination on the residual sequence through a threshold method to obtain the abnormal data; and performing event type discrimination on the abnormal data to identify the traffic event.

[0038] Specifically, setting the cycle length according to traffic characteristics means determining a time range to capture the periodic characteristics of traffic data based on the regular changes in traffic flow. The cycle length usually includes a daily cycle and a weekly cycle. The daily cycle refers to a 24-hour cycle, which reflects the daily traffic flow changes, such as the morning and evening rush hours. The weekly cycle refers to a seven-day cycle, which reflects the weekly traffic flow changes, such as the differences between weekdays and weekends. For example, the traffic data of a certain city shows that the peak period for the inbound direction is from 7:00 to 9:00 every morning, and the evening rush hour on Fridays is more congested than usual. These regular changes can be captured by setting the daily cycle and the weekly cycle.

[0039] Setting the trend smoothing window based on the product of the cycle length and a preset multiple means calculating a time window according to the cycle length and a preset multiple to smooth the long-term trend of traffic data. The role of the trend smoothing window is to remove the short-term fluctuations in the data and highlight the long-term change trend. For example, if the daily cycle is 1,440 minutes (24 hours) and the preset multiple is 1.5 times, then the trend smoothing window is 2,160 minutes (36 hours). This window can help analyze the long-term changes in traffic flow, such as whether the traffic volume on a certain road section is increasing year by year.

[0040] Decomposing the real-time data with the cycle length and the trend smoothing window as the decomposition parameters means using these parameters to decompose the traffic data into three parts: seasonal, trend, and residual. The seasonal part reflects the periodic changes, such as the morning and evening rush hours every day. The trend part reflects the long-term trend, such as the annual increase in traffic volume. The residual part is the remaining part after removing the seasonal and trend parts and usually contains the signals of unexpected events. For example, after decomposing the traffic data of a certain road section, the seasonal part shows that the traffic volume during the morning rush hour is 1,000 vehicles per hour every day, the trend part shows that the traffic volume increases by 5% annually, and the residual part may contain the sudden drop in traffic volume caused by a certain accident.

[0041] Judging the anomalies in the residual sequence through the threshold method means setting a threshold to determine whether there are anomalies according to the statistical characteristics of the residual sequence. The threshold method usually uses the standard deviation as the benchmark. For example, a residual value exceeding three times the standard deviation is judged as an anomaly. For example, if the standard deviation of the residual sequence of a certain road section is 50 vehicles per hour, and if a certain residual value exceeds 150 vehicles per hour, it will be judged as an anomaly. This method can effectively capture unexpected events, such as accidents or congestion.

[0042] Determining the event type of abnormal data means further judging which traffic event it belongs to according to the characteristics of the abnormal data. For example, a negative residual value with a long duration may be a sudden drop in traffic flow caused by an accident; a positive residual value with a short duration may be a sudden increase in traffic flow caused by a temporary event. For example, the residual sequence of a certain section shows that an anomaly lasted for twenty minutes and the residual value was negative two hundred vehicles per hour. Combining with the video stream data, it is confirmed that a two-vehicle rear-end collision occurred on this section.

[0043] Furthermore, this application also includes: identifying the mutation direction of the abnormal data, where the mutation direction includes negative residual offset and positive residual offset; timing the duration of the abnormal data; and identifying the traffic event through the mutation direction and the duration in the traffic event library.

[0044] Specifically, identifying the mutation direction of abnormal data means judging whether the abnormal data is offset in the negative direction or the positive direction by analyzing the change trend of the residual sequence. A negative residual offset usually indicates a sudden drop in traffic flow caused by some unexpected situation, such as a traffic accident or road closure; a positive residual offset indicates a sudden increase in traffic flow caused by some unexpected situation, such as a temporary event or vehicle aggregation. For example, the residual sequence of a certain section shows that an abnormal value is negative one hundred vehicles per hour, indicating a sudden drop in traffic flow on this section, which may be due to a traffic accident ahead.

[0045] Timing the duration of abnormal data means the time length from the start to the end of the abnormal data. The length of the duration can help judge the severity and impact range of the traffic event. For example, the duration of a certain abnormal data is thirty minutes, indicating that the event has a relatively lasting impact on traffic and more proactive measures may be needed to relieve congestion. Through the timing function, the impact of traffic events can be evaluated more accurately. For example, the abnormal data of a certain section shows that the sudden drop in traffic flow lasted for twenty minutes. Combining with the mutation direction, it is judged that it may be due to a two-vehicle rear-end collision ahead.

[0046] Identifying the traffic event through the mutation direction and the duration in the traffic event library means comparing the mutation direction and the duration of the abnormal data with a pre-established traffic event library to determine the specific traffic event type. The traffic event library is a data set containing the characteristics of various traffic events. For example, traffic accidents usually show a negative residual offset and a long duration, while temporary events may show a positive residual offset and a short duration. For example, the abnormal data of a certain section shows a negative residual offset and lasts for thirty minutes. After comparison with the traffic event library, it is confirmed that this event is a traffic accident.

[0047] Furthermore, this application also includes: using traffic event records as input features and spatial range records and time range records as output features to train a graph neural network to obtain an impact range prediction model, where the impact range prediction model includes the impact propagation between road network nodes.

[0048] Specifically, using traffic event records as input features means taking the relevant information of traffic events that occurred in history as the input data of the model. Traffic event records include detailed information such as event type, occurrence location, and occurrence time. These data can help the model learn the characteristics and patterns of traffic events. For example, the traffic event records of a certain city show that a traffic accident occurred on the main road during the morning rush hour, causing congestion on the surrounding roads for twenty minutes.

[0049] Using spatial range records and time range records as output features means that the model needs to predict the impact range of traffic events, including spatial range and time range. Spatial range record refers to the area affected by a traffic event. For example, a certain accident caused congestion on the roads within a radius of five hundred meters around it. Time range record refers to the duration of the impact of a traffic event. For example, a certain congestion lasted for thirty minutes. These output features are the targets that the model needs to learn and predict. For example, a certain model can predict that a traffic accident will cause congestion on the roads within a radius of three hundred meters around it and last for about twenty-five minutes by learning historical data.

[0050] Training a graph neural network means using a graph neural network to process traffic event data. Graph neural networks are particularly suitable for processing data with complex relationships, such as nodes and edges in a road network. In a traffic scenario, each intersection of the road network can be regarded as a node, and the road segments can be regarded as edges. The impact of a traffic event will spread between these nodes and edges. By training a graph neural network, the model can learn the impact propagation rules between road network nodes. For example, a certain graph neural network can accurately predict the impact range and duration of a traffic accident on the surrounding road segments by learning historical data.

[0051] Obtaining an impact range prediction model means that by training a graph neural network, a model that can predict the impact range of traffic events is finally obtained. This model can output the prediction results of spatial range and time range based on the input traffic event records. For example, after receiving the record of a certain traffic accident, a certain impact range prediction model predicts that the accident will cause congestion on the roads within a radius of four hundred meters around it and last for about thirty minutes. These prediction results can help traffic management departments take measures in advance to relieve traffic congestion.

[0052] Among them, the impact range prediction model includes the impact propagation between road network nodes, which means that the model can simulate the propagation process of traffic events in the road network. For example, a traffic accident first affects the section where it occurs, and then gradually spreads to the surrounding intersections and sections. By simulating this propagation process, the model can more accurately predict the impact range of traffic events. For example, a certain model predicts that a traffic accident will first affect the section where it occurs, and then spread to two surrounding intersections within ten minutes, ultimately causing congestion on the sections within a radius of 500 meters around.

[0053] Furthermore, this application also includes: extracting the event verification degree and event location based on the impact range, and dispatching rescue vehicle resources and police force resources; extracting continuous events based on the impact range, and performing signal timing; combining the resource dispatching result of the dispatching of rescue vehicle resources and police force resources and the traffic control result of signal timing to generate the dispatching strategy.

[0054] Specifically, extracting the event verification degree and event location based on the impact range means further analyzing the severity and specific occurrence location of traffic events according to the impact range output by the prediction model. The event verification degree refers to the assessment of the severity of the event. For example, a minor accident may only affect a local section, while a major accident may cause large-area congestion; the event location refers to the specific section or intersection where the event occurs. For example, a certain model predicts that the impact range of a traffic accident is within 500 meters around, the event verification degree is "severe", and the event location is the intersection of the main road and the secondary road. These information provide a basis for subsequent resource dispatching.

[0055] Dispatching rescue vehicle resources and police force resources means reasonably allocating rescue vehicle and police force resources according to the verification degree and location of the event. Rescue vehicles include ambulances, fire trucks, and tow trucks, and police force resources include traffic police and patrol cars. The goal of dispatching is to arrive at the event site as soon as possible, reduce traffic congestion and secondary accidents. For example, a traffic accident is evaluated as "severe", and the dispatching center immediately dispatches two ambulances, one fire truck, and three traffic police to the scene, with an estimated arrival time of ten minutes.

[0056] Extracting continuous events based on the impact range and performing signal timing means adjusting the signal timing plan for traffic events with a long duration to relieve congestion. A continuous event refers to an event whose impact time exceeds a certain threshold. For example, a certain congestion is expected to last for more than 30 minutes. Signal timing means adjusting the duration and phase of traffic lights to optimize the traffic flow. For example, the signal timing of a certain section is adjusted from switching once per minute to switching once every 90 seconds to give priority to the traffic flow in the congested direction.

[0057] Generating a scheduling strategy by combining the resource scheduling results of rescue vehicle resources and police force resources and the traffic control results of signal timing means integrating the results of resource scheduling and signal timing to form a complete traffic management plan. The scheduling strategy includes the specific locations and times of resource deployment, as well as the specific plans for signal adjustments. For example, a certain scheduling strategy shows that the rescue vehicle will arrive at the accident scene within ten minutes, and at the same time, the signal timing is adjusted to give priority to the traffic flow in the congested direction, and it is expected that the congestion will be alleviated within twenty minutes.

[0058] Furthermore, this application also includes: detecting congestion at road network nodes. If the detection result is extremely congested, mark to obtain priority control nodes; optimize the signal timing of the priority control nodes according to the timing adjustment rules.

[0059] Specifically, detecting congestion at road network nodes means monitoring each key point in the urban traffic network, such as intersections or overpasses, through technical means to determine whether traffic congestion occurs at these points. Congestion detection usually uses technical means such as cameras, induction coils or floating car data to obtain traffic flow and speed information in real time.

[0060] The detection result being extremely congested means that when the detection system finds that the traffic flow at a certain node significantly exceeds the normal level, or the vehicle driving speed is lower than a certain threshold, it is determined that there is extremely congested situation at this node. In this case, this node will be particularly concerned because it may have a significant impact on the operation efficiency of the entire road network.

[0061] Make a mark on the node where extremely congested is detected and list it as a control object that needs to be prioritized. Such nodes will be the focus of measures taken by the traffic management department to alleviate congestion.

[0062] Optimizing the signal timing of the priority control nodes according to the timing adjustment rules means optimizing the signal cycle of these nodes marked as priority control according to the pre-established traffic signal timing adjustment rules. For example, the green light time may be extended or the red and green light sequences in each direction may be adjusted to improve the traffic efficiency at the intersection.

[0063] For example, assume that the vehicle passing speed at a certain intersection drops below 15 kilometers per hour during the morning rush hour. The congestion detection system identifies it as an extremely congested node. This node is immediately marked as a priority control node. Subsequently, according to the timing adjustment rules, the traffic engineer extends the green light time of this intersection from the original 30 seconds to 40 seconds to alleviate congestion.

[0064] In summary, the AI decision - assisted sudden traffic incident scheduling decision - making method provided by this application has the following technical effects: By connecting to a data acquisition system, traffic data and environmental data are collected as real - time data; time - series analysis is used to monitor abnormal data in the real - time data to identify traffic incidents; an impact - range prediction model is used to predict the impact range of the traffic incident; based on the impact range, resource scheduling and traffic control are performed to generate a scheduling strategy; the scheduling strategy is sent to an execution unit for strategy execution. That is to say, by achieving the technical goals of rapid response and intelligent scheduling decision - making, the technical effects of improving urban traffic management efficiency and alleviating traffic congestion are achieved.

[0065] Embodiment 2: Based on the sudden traffic incident scheduling decision - making method assisted by AI in the foregoing embodiment and with the same inventive concept, this application also provides an AI - assisted sudden traffic incident scheduling decision - making system. Please refer to the attached Figure 2 , including: a data acquisition module 11, which is used to connect to a data acquisition system to collect traffic data and environmental data as real - time data; an anomaly analysis module 12, which is used to monitor abnormal data in the real - time data by using time - series analysis to identify traffic incidents; a prediction module 13, which is used to predict the impact range of the traffic incident by using an impact - range prediction model; a scheduling module 14, which is used to perform resource scheduling and traffic control based on the impact range to generate a scheduling strategy; a sending module 15, which is used to send the scheduling strategy to an execution unit for strategy execution.

[0066] Furthermore, the AI - assisted sudden traffic incident scheduling decision - making system is also used for: The traffic data includes video - stream data, traffic - flow data, vehicle - speed data, floating - car data, vehicle - type data, and passing - time data, and the environmental data includes precipitation data, visibility data, temperature data, and icing - state data.

[0067] Furthermore, the AI - assisted sudden traffic incident scheduling decision - making system is also used for: setting a cycle length according to traffic characteristics, where the cycle length includes a daily cycle and a weekly cycle; setting a trend - smoothing window based on the product of the cycle length and a preset multiple; using the cycle length and the trend - smoothing window as decomposition parameters to decompose the real - time data to obtain a residual sequence; performing anomaly determination on the residual sequence through a threshold method to obtain the abnormal data; and performing event - type discrimination on the abnormal data to identify the traffic incident.

[0068] Further, the AI decision - assisted sudden traffic incident scheduling decision system is also used for: identifying the mutation direction of the abnormal data, where the mutation direction includes residual negative offset and residual positive offset; timing the duration of the abnormal data; and identifying the traffic incident through the mutation direction and the duration in the traffic incident library.

[0069] Further, the AI decision - assisted sudden traffic incident scheduling decision system is also used for: taking traffic incident records as input features and spatial range records and time range records as output features to train a graph neural network to obtain an influence range prediction model, where the influence range prediction model includes the influence propagation between road network nodes.

[0070] Further, the AI decision - assisted sudden traffic incident scheduling decision system is also used for: extracting the event verification degree and the event location based on the influence range, and scheduling rescue vehicle resources and police force resources; extracting continuous events according to the influence range and performing signal timing; and generating the scheduling strategy by combining the resource scheduling result of the scheduling of rescue vehicle resources and police force resources and the traffic control result of signal timing.

[0071] Further, the AI decision - assisted sudden traffic incident scheduling decision system is also used for: detecting congestion of road network nodes, and if the detection result is abnormal congestion, marking to obtain priority regulation nodes; and optimizing the signal timing of the priority regulation nodes according to the timing adjustment rule.

[0072] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The AI decision - assisted sudden traffic incident scheduling decision method and specific examples in the foregoing Embodiment 1 are equally applicable to the AI decision - assisted sudden traffic incident scheduling decision system in this embodiment. Through the foregoing detailed description of the AI decision - assisted sudden traffic incident scheduling decision method, those skilled in the art can clearly know the AI decision - assisted sudden traffic incident scheduling decision system in this embodiment. Therefore, for the sake of brevity of the specification, it will not be elaborated here.

[0073] 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 apparent to those skilled in the art. 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 the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0074] Obviously, those skilled in the art can make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of this application and its equivalent technologies, this application is also intended to include these changes and modifications.

Claims

1. AI decision-making-assisted emergency traffic incident dispatch decision-making method, characterized by: include: Connect to the data collection system to collect traffic data and environmental data as real-time data; Using time series analysis to monitor abnormal data in the real-time data and identify traffic incidents; Predicting the impact range of the traffic incident using an impact range prediction model; Perform resource scheduling and traffic control based on the impact range and generate a scheduling strategy; The scheduling strategy is sent to the execution unit for strategy execution.

2. The AI ​​decision-making-assisted emergency traffic incident dispatch decision-making method according to claim 1, characterized in that: The traffic data includes video stream data, vehicle flow data, vehicle speed data, floating vehicle data, vehicle model data and travel time data, and the environmental data includes precipitation data, visibility data, temperature data and icing status data.

3. The AI ​​decision-making-assisted emergency traffic incident dispatch decision-making method according to claim 1, characterized in that: Use time series analysis to monitor abnormal data in the real-time data and identify traffic events, including: Setting a cycle length according to traffic characteristics, wherein the cycle length includes a daily cycle and a weekly cycle; Setting a trend smoothing window based on the product of the cycle length and a preset multiple; Decomposing the real-time data using the cycle length and the trend smoothing window as decomposition parameters to obtain a residual sequence; Performing abnormality determination on the residual sequence by a threshold method to obtain the abnormal data; The abnormal data is subjected to event type discrimination to identify the traffic event.

4. The AI ​​decision-making-assisted emergency traffic incident dispatch decision-making method as claimed in claim 3, characterized in that: The abnormal data is subjected to event type discrimination to identify the traffic event, including: Identify the mutation direction of the abnormal data, wherein the mutation direction includes a negative residual offset and a positive residual offset; Timing the duration of the abnormal data; The traffic event is obtained by identifying the mutation direction and the duration in a traffic event database.

5. The AI ​​decision-making-assisted emergency traffic incident dispatch decision-making method according to claim 1, characterized in that: Taking traffic event records as input features and spatial range records and time range records as output features, a graph neural network is trained to obtain an impact range prediction model, wherein the impact range prediction model includes the impact propagation between road network nodes.

6. The AI ​​decision-making-assisted emergency traffic incident dispatch decision-making method according to claim 1, characterized in that: Based on the impact range, resource scheduling and traffic control are performed to generate a scheduling strategy, including: Extracting the event verification degree and event location based on the impact range, and dispatching rescue vehicle resources and police resources; Extracting continuous events according to the impact range and performing signal light timing; The dispatch strategy is generated by combining the resource dispatch results of the dispatch of rescue vehicle resources and police resources and the traffic control results of the signal light timing.

7. The AI ​​decision-making-assisted emergency traffic incident dispatch decision-making method according to claim 6, characterized in that: Extract continuous events according to the impact range and perform traffic light timing, including: Conduct congestion detection on the road network nodes. If the detection result is abnormal congestion, mark the node for priority regulation; The signal light timing of the priority control node is optimized according to the timing adjustment rules.

8. AI decision-making-assisted emergency traffic incident dispatch decision system, characterized by: The steps for implementing the AI ​​decision-making-assisted emergency traffic event dispatch decision method according to any one of claims 1 to 7 include: A data acquisition module, wherein the data acquisition module is used to connect to a data acquisition system and collect traffic data and environmental data as real-time data; An abnormality analysis module, the abnormality analysis module is used to monitor abnormal data in the real-time data using time series analysis to identify traffic events; A prediction module, the prediction module is used to predict the impact range of the traffic event using an impact range prediction model; A scheduling module, the scheduling module is used to perform resource scheduling and traffic control based on the impact range and generate a scheduling strategy; A sending module is used to send the scheduling strategy to the execution unit for strategy execution.

Images