Traffic state monitoring and dynamic early warning method based on group interaction model

Abstract

The invention discloses a traffic state monitoring and dynamic early warning method based on a group interaction model, and relates to the technical field of traffic monitoring, and the method comprises the steps: carrying out the preprocessing of obtained comprehensive traffic information, and extracting corresponding travel data features and hub operation data features; constructing a group interaction model according to the extracted travel data features, and judging a traffic state grade by taking a time-space correlation threshold value as an inspection index; establishing a traffic object evaluation index system, an emergency event fuzzy evaluation model, a transport capacity operation situation evaluation model and a hub traffic operation situation evaluation model; and inputting the extracted hub operation data characteristics into each evaluation model to evaluate the traffic situation of the hub, matching an early warning mode to give an alarm, and outputting an auxiliary traffic scheduling scheme. According to the invention, the traffic state in the range of the road network and the hub can be analyzed and effectively evaluated in time, and early warning processing is carried out, so that evacuation and traffic flow scheduling of passengers are facilitated, and the travel comfort and safety of the passengers are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic monitoring, and in particular to a traffic status monitoring and dynamic early warning method based on a group interaction model. Background Art

[0002] Comprehensive transportation hubs are the intersection of various modes of transportation, and they play a pivotal role in the transportation system. As a key link in the comprehensive transportation system and a place for the main passenger flow distribution and transfer in the city, they are responsible for the rapid conversion between the city's external and internal transportation, as well as the rapid connection between various modes of urban public transportation. In view of the complex passenger and vehicle flow conditions within the scope of comprehensive transportation hubs, by analyzing group travel information and evaluating the situation of transportation hubs, it can help urban transportation management departments better respond to traffic incidents and emergencies.

[0003] The existing comprehensive transportation hub monitoring methods have the problems of scattered monitoring information, strong pertinence, lack of comprehensive data processing and analysis capabilities, and cannot timely analyze and effectively evaluate the traffic status within the road network and hub, and issue early warnings for traffic dynamics. The problem of road network congestion caused by the mismatch between traffic flow and capacity is becoming more and more serious. If the road network congestion situation is not effectively evaluated and handled, it will cause road traffic congestion and even large-scale passenger detention and other undesirable phenomena, which will have a great impact on society and cause property losses. Summary of the invention

[0004] In view of this, the present invention provides a traffic status monitoring and dynamic early warning method based on a group interaction model. By fusing and analyzing the collected data, and constructing a group interaction model and a comprehensive traffic indicator monitoring system, comprehensive data processing and analysis are achieved, which solves the technical problems that the existing comprehensive transportation hub monitoring information is scattered and highly targeted, lacks the ability to comprehensively process and analyze data, and cannot timely analyze and effectively evaluate the traffic status within the road network and hub, and cannot warn of traffic dynamics.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The present invention provides a traffic status monitoring and dynamic early warning method based on a group interaction model, comprising the following steps:

[0007] Obtaining comprehensive traffic information within the scope of the traffic hub, preprocessing the comprehensive traffic information, and extracting corresponding travel data features;

[0008] Extracting individual travel data according to the extracted travel data features, building a group interaction model, and expanding the individual travel data to group travel data through the group interaction model to obtain hub operation data;

[0009] Construct emergency event fuzzy assessment model, transport capacity operation status assessment model and hub transportation operation status assessment model;

[0010] The extracted hub operation data is input into each evaluation model for evaluation, and it is judged whether the corresponding warning level is reached according to each evaluation result. An alarm is issued based on the judgment result and the warning method is matched, and an auxiliary traffic dispatch plan is output.

[0011] Furthermore, the obtaining of comprehensive traffic information within the scope of the traffic hub and preprocessing of the comprehensive traffic information includes: the comprehensive traffic information within the scope of the traffic hub includes real-time traffic data and passenger flow data, historical traffic data and passenger flow data; the preprocessing includes cleaning and storing the data, and removing abnormal values ​​and erroneous data;

[0012] Furthermore, the extraction of corresponding travel data features includes: mining frequent item sets and performing hub correlation analysis using the Apriori algorithm and graph clustering algorithm based on the acquired comprehensive traffic information, summarizing the association rules between various types of data in the comprehensive traffic information, and fusing the data sets of various types of data in the comprehensive traffic information according to the association rules, and extracting travel data features from the fused data sets, wherein the travel data features include characteristics of various travel modes, travel time periods, travel groups, transfer behaviors, and activity types.

[0013] Further, extracting individual travel data according to the extracted travel data features includes:

[0014] Establish a spatiotemporal threshold sliding window mechanism to slice travel data features and cluster them according to activity combination patterns;

[0015] The key elements for identifying the transfer passenger flow are extracted from the travel data feature slices of different clustering categories, and the key elements corresponding to each clustering category are used to enhance the travel data feature slices; the key elements for identifying the transfer passenger flow include the travel start and end points, transfer nodes, transfer time periods and time consumption, and transfer paths of individuals or groups;

[0016] The enhanced travel data feature slices are used as input, and the multivariate framed and nonlinear time series of pedestrian and vehicle trajectories are identified through graph convolutional neural networks, and the node feature matrix of the pedestrian and vehicle trajectories under multimodal travel conditions is obtained. The node feature matrix of the pedestrian and vehicle trajectories is composed of the node feature representations of each node on the trajectory, and the feature representation of each node in the node feature matrix is ​​updated by aggregating the information of neighboring nodes to obtain the monitoring results of the pedestrian and vehicle trajectories within the hub range;

[0017] Based on the monitoring results of the trajectories of people and vehicles within the hub, combined with the historical traffic data in the comprehensive traffic information and using the extended Markov chain model, the future trajectories of people and vehicles within the hub are predicted; the predicted results of the trajectories of people and vehicles within the hub are combined with the monitoring results to obtain individual travel data.

[0018] Further, the expanding the individual travel data to group travel data through the group interaction model includes:

[0019] The individual travel data is input into the group interaction model. The group interaction model abstracts each travel individual represented by the individual travel data into an entity that makes decisions based on its own status and the status of neighboring individuals, analyzes the interaction between each travel individual, and then expands the individual travel data to group travel data, that is, obtaining the hub operation data.

[0020] Furthermore, the emergency event fuzzy assessment model performs assessment including:

[0021] Construct the first-level monitoring indicators and second-level monitoring indicators for emergency events. The first-level monitoring dimension indicators include emergency events and equipment alarms. The second-level monitoring dimension indicators under emergency events include extreme weather, fire accidents, and large passenger flow events. The second-level monitoring dimension indicators under equipment alarms include temperature, humidity, smoke, and video equipment alarms. The weights of various indicators are calculated using the hierarchical entropy analysis method.

[0022] Based on the hub operation data, calculate the secondary monitoring indicators and their weights for emergency events, and sum them up to obtain the comprehensive score of emergency situation to evaluate the probability and severity of emergency events;

[0023] Determine the risk status level to which the comprehensive score of the emergency situation belongs, wherein the risk status level includes four levels: very high, high, medium, and low; and output the risk status level judgment result as the evaluation result of the comprehensive score model of the emergency situation.

[0024] Furthermore, the capacity operation status assessment model performs assessment including: constructing capacity evaluation indicators including passenger flow diffusion response time, capacity loading rate, capacity filling coefficient and waiting time, and analyzing the capacity evaluation indicators in real time according to hub operation data; if one of the capacity evaluation indicators exceeds a preset threshold, it is judged that the current capacity operation status is abnormal; if it does not exceed the threshold, the current capacity operation status is normal.

[0025] Furthermore, the hub traffic operation status assessment model performs an assessment including: constructing two evaluation indicators, namely, average passenger flow density and passenger flow information, wherein the average passenger flow density and passenger flow information are respectively divided into five levels, and judging the current level of the average passenger flow density and passenger flow information according to the hub operation data, and judging the regional congestion degree by comprehensively considering the current level of the average passenger flow density and passenger flow information, and the regional congestion degree is the hub traffic operation status assessment result.

[0026] Furthermore, judging whether the corresponding warning level is reached according to each evaluation result, giving an alarm based on the judgment result matching the warning mode, and outputting the auxiliary traffic dispatching plan include:

[0027] Based on the traffic operation situation, hub capacity situation and traffic incident emergency situation obtained from each evaluation result, it is judged whether the corresponding warning level is reached. The maximum warning level is level 3, and a corresponding alarm signal is generated. According to the generated alarm signal, a preset alarm mode is matched to issue a situation warning. If the minimum warning level is not reached, no alarm is issued.

[0028] According to the original travel data characteristics, hub operation data and current warning level, the corresponding historical dispatch response plan with the highest similarity is output according to similarity matching.

[0029] It can be seen from the above technical solution that the advantages of the present invention are:

[0030] 1. The present invention constructs a group interaction model to analyze group travel information by fusing and analyzing the collected data, and then judges the road traffic status. In view of the complex passenger and vehicle flow conditions in the comprehensive transportation hub, the corresponding indicator parameters are used to evaluate the operation status, capacity status and emergency event status of the transportation hub in real time, and to carry out early warning and output scheduling plans, so as to facilitate urban transportation management departments to better respond to traffic incidents and emergencies.

[0031] 2. The present invention can timely analyze and effectively evaluate the traffic status within the road network and hub, and carry out early warning processing of traffic dynamics, so as to facilitate the evacuation of passengers and the dispatch of traffic flow, improve the comfort and safety of passengers, and enhance the robustness and resilience of the entire transportation system. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0033] Figure 1 It is a schematic diagram of the steps of the traffic status monitoring and dynamic early warning method based on the group interaction model of the present invention.

[0034] Figure 2It is a structural schematic diagram of the present invention. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments and the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0036] By collecting statistics and analyzing the information of vehicles or passengers passing through the transportation hub, the changes and trends of traffic flow can be obtained. Based on the changes and trends, the capacity and load conditions of the transportation hub and the matching degree of the transport capacity of rail transit systems of different types can be evaluated, and the operation of vehicles inside or near the transportation hub can be evaluated to reflect the traffic diversion capacity of the transportation hub. It can be determined whether there are obstacles to the transportation hub, and early warning reminders and further scheduling improvements can be made accordingly.

[0037] like Figure 1 and Figure 2 As shown, this embodiment provides a traffic status monitoring and dynamic early warning method based on a group interaction model. Aiming at the complex passenger and vehicle flow conditions in a comprehensive transportation hub, by real-time monitoring and analyzing the traffic status of the transportation hub, it can help the urban transportation management department to better respond to traffic incidents and emergencies, and improve the robustness and resilience of the entire transportation system. Specifically, it includes:

[0038] Step S1: Obtain comprehensive traffic information within the scope of the traffic hub, pre-process the comprehensive traffic information, and extract corresponding travel data features and hub operation data features.

[0039] Step S1 is specifically as follows: the step of obtaining comprehensive traffic information within the scope of the traffic hub and preprocessing the comprehensive traffic information includes:

[0040] Step S10: Use the deployment method based on auction bidding criteria to deploy traffic node sensing equipment, and use the dynamic task acceptance algorithm to schedule tasks at hub traffic nodes, collect real-time traffic data, passenger flow data, and environmental data, and obtain historical traffic data and passenger flow data, and clean and store the data to remove abnormal values ​​and erroneous data; the traffic data includes the speed and traffic volume of each means of transportation, as well as traffic accident information on the route (time, location, type and number of vehicles involved, etc.) and road condition information (road construction, road closure, traffic control, etc.). The passenger flow data includes the number of passengers arriving / leaving the station, passenger flow density, and passenger waiting time, etc. The environmental data includes the weather conditions of the day, air quality, noise level, and other factors that have an important impact on traffic operation.

[0041] Step S11: Based on the acquired comprehensive traffic data, the Apriori algorithm and graph clustering algorithm are used to mine frequent item sets and perform hub correlation analysis, generate interest indexes, summarize the association rules between multi-source data sets, and fuse the data sets according to the association rules. The travel data features are extracted and the correlation indicators are sorted for each type of travel mode, travel time, travel group, transfer behavior, and activity type, so as to realize the parameterization and structuring of the travel process within the transportation hub; the travel refers to the traffic behavior of pedestrians or vehicles moving in the transportation hub through any transportation mode in the hub.

[0042] Step S12: Screen and extract corresponding model evaluation data features according to the evaluation indicators of each evaluation model to obtain hub operation characterization data, and input the characterization data into each evaluation model.

[0043] During data preprocessing, the present invention first analyzes association rules based on multi-source transportation big data (such as multi-period hub entry and exit data, public transportation card swiping data, residents' travel attributes, etc.), uses Apriori, graph clustering and other algorithms to mine frequent item sets and perform hub correlation analysis, generate interest indicators, summarize the association rules between multi-source data sets, and then perform data set fusion and travel data feature extraction.

[0044] Step S2: Dynamically identify and predict individual travel based on the extracted travel data features, build a group interaction model, and expand individual travel to group travel through the group interaction model to obtain hub operation data. This step can analyze the path selection of individual travelers and the traffic status of groups through the hierarchical travel structured modeling of "individual-group".

[0045] The dynamic identification and prediction of individual travel is specifically to dynamically identify and predict the travel and transfer behavior, travel mode and trajectory of hub passenger flow based on the extracted travel data characteristics.

[0046] Specifically, the dynamic identification and prediction of the travel and transfer behavior, travel mode and trajectory of hub passenger flow based on the extracted travel data features includes:

[0047] Step S20: Establishing a temporal and spatial threshold sliding window mechanism, extracting feature slices in the travel time series through a dynamic window range, and clustering according to activity combination patterns to panoramically visualize hub passenger flow and travel chain forms; the activity combination patterns include daily commuting patterns, holiday leisure patterns, passenger patterns, etc., and each activity combination pattern defines the travel time, travel start and end points, and traffic patterns within a day;

[0048] Step S21: Based on the travel chain slice, the trip breakpoint information, travel attributes, and time domain characteristics can be integrated to analyze the transfer behavior, and the key elements of transfer passenger flow identification can be derived, including the travel start and end point, transfer node, transfer period and time, and transfer path.

[0049] Step S22: Identify the pedestrian and vehicle trajectories of multivariate frames and nonlinear time series through a graph convolutional neural network, obtain a node feature matrix of pedestrian and vehicle trajectories under multimodal travel conditions, and update the feature representation of each node by aggregating neighboring node information.

[0050] As a deep learning model in the field of graph data processing, Graph Convolutional Neural Network (GCN) has shown good processing capabilities for point, line, and surface features, as well as irregular time series data. It can accurately identify key information in travel trajectories, including multi-dimensional factors such as activity patterns and travel methods, and can extract nonlinear combination influence relationships between factors.

[0051] Step S23: Based on the human and vehicle trajectory recognition results, combined with historical traffic data and using the extended Markov chain model, predict future path slices.

[0052] Based on the perspective of individual travel status, travel trajectories can be reconstructed according to individual attributes, travel preferences, mode choices, and activity patterns. It can also predict information such as the intersection status of travel trajectories within the hub, the density of pedestrian and vehicle traffic, and the load status of road facilities, forming an important basis for judging traffic status.

[0053] After obtaining the results of human and vehicle trajectory recognition and prediction, a group interaction model is constructed. The group interaction model abstracts the traveling individuals into entities with certain states and behavioral strategies. It makes decisions based on its own and neighboring individual states and experience factors, and transforms the interactions between individuals into local neighboring relationships, network connections, and information transmission methods, thereby expanding individual travel to group travel states.

[0054] The group interaction model can be constructed based on classic simulation models such as agent-based models (ABM), multi-agent systems (MAS) or social force models. In the group interaction model, the transformation of interaction between individuals and the expansion of travel status are realized as follows:

[0055] 1) Define neighbor relationships based on the spatial location and interaction radius of individuals;

[0056] 2) Establishing network connections based on interactions or functional connections, through which individuals can share or receive information and influence the decisions and behaviors of distant individuals;

[0057] 3) define the mechanisms by which information is transmitted between individuals, such as visual, auditory, or digital communication;

[0058] 4) Through comprehensive analysis of individual decisions and behavioral feedback, the model can predict and simulate the travel status of the entire group, including flow distribution, traffic congestion, and evacuation efficiency.

[0059] At present, the online evaluation and analysis of the operation status of comprehensive transportation hubs mainly focuses on the real-time monitoring of hub passenger flow. It lacks the evaluation and monitoring of the overall transportation hub status from the three dimensions of emergency, capacity and surrounding traffic operation, and it is difficult to provide hub management departments with fast, accurate and comprehensive hub operation status information. Therefore, step S3 can achieve a more comprehensive monitoring and analysis of the hub operation status.

[0060] Step S3: extract the main factors that reflect the hub emergency event situation, capacity operation situation and traffic operation situation respectively, and construct emergency event fuzzy assessment model, capacity operation situation assessment model and hub traffic operation situation assessment model.

[0061] Different evaluation index systems for hub traffic objects are set up for the fuzzy assessment model of emergency events, the capacity operation status assessment model and the hub traffic operation status assessment model.

[0062] For the emergency event fuzzy evaluation model, a first-level monitoring indicator is set, and a second-level monitoring indicator is set under the first-level monitoring indicator. The first-level monitoring dimension indicators include emergency events and equipment alarms. The second-level monitoring dimension indicators under emergency events include extreme weather, fire accidents, large passenger flow events, etc. The second-level monitoring dimension indicators under equipment alarms include temperature, humidity, smoke sensing equipment alarms, video equipment alarms, etc. The hierarchical entropy analysis method is used to calculate the weights of various indicators, as follows:

[0063]

[0064] Among them, w j is the weight of the j-th secondary indicator, E j is the information entropy of the indicator, n is the number of indicators, p ij is the proportion of the jth indicator in the i-th emergency event;

[0065] The emergency event fuzzy evaluation model uses a hierarchical entropy analysis method to comprehensively score the emergency situation based on an index system constructed according to the primary index and secondary index corresponding to the emergency event.

[0066] In this embodiment, a fuzzy evaluation model for emergency events is constructed based on the MECE (Mutually Exclusive Collectively Exhaustive, MECE) principle. According to the hub operation data combined with real-time perception data such as weather, fire sensors, cameras, videos, and checkpoints, a secondary monitoring indicator system is constructed in the dimensions of extreme weather, fire accidents, large passenger flow events, temperature, humidity, smoke sensing equipment alarms, and video equipment alarms. On top of the secondary monitoring dimension indicators, a primary monitoring indicator system in the dimensions of emergency events and equipment alarms is established. The weights of various indicators are calculated using the hierarchical entropy analysis method, and a comprehensive scoring model for emergency situations is constructed with the probability and severity of risk events as the purpose. Specifically, Among them, s i is the comprehensive score of the ith solution, p ij is the proportion of the jth indicator in the ith solution, x ij is the ith value of the jth indicator. Finally, based on the emergency pressure state, the comprehensive score is judged to belong to the risk state level, which includes four levels: very high, high, medium, and low, to achieve real-time monitoring and early warning of emergency events in comprehensive transportation hubs.

[0067] The capacity operation status assessment model analyzes capacity evaluation indicators such as passenger flow diffusion response time, capacity loading rate, capacity filling coefficient and waiting time of each transportation mode (such as railways, buses, subways, online car-hailing, and private cars) based on hub operation data. Specifically, through historical statistics and real-time monitoring of passenger flow data of railways, buses, subways, online ride-hailing cars, and private cars, an algorithm for extracting information about the spatiotemporal evolution of passenger flow based on spatial domain graph convolution is constructed to monitor the pre-diffusion, synchronous diffusion, and post-diffusion states of passenger flow at the station, and analyze and evaluate the passenger flow diffusion response time of major transportation nodes; based on historical statistics and real-time monitoring of the capacity carrying data of railways, buses, subways, online ride-hailing cars, and private cars, the adjacent cycle changes of the capacity of each mode of transportation based on the diffusion index model are established, and the capacity carrying rate (the ratio of actual passenger capacity to maximum passenger capacity) of each mode of transportation is analyzed; and based on the passenger flow diffusion trend and capacity filling coefficient of the station, which reflects the filling degree of various transportation tools during peak and non-peak periods, a passenger waiting time model for transportation hubs based on the probability of time window state is constructed to analyze the waiting time. Data from various sensors and monitoring equipment are integrated, and hub operation data is updated in real time. The data is analyzed using the above method to generate a real-time operation status evaluation. When abnormal fluctuations in passenger flow or capacity are monitored (a certain capacity evaluation indicator exceeds the preset threshold), the current capacity operation situation is judged to be abnormal and the system will automatically trigger the early warning mechanism.

[0068] In this embodiment, in order to solve the problem of passenger detention caused by the imbalance between passenger flow and capacity supply and demand inside and outside the comprehensive transportation hub, a transportation hub capacity operation status assessment model based on the passenger flow and capacity diffusion trend is used to conduct capacity assessment, so as to realize real-time monitoring and early warning of subsequent capacity load.

[0069] In this embodiment, the hub traffic operation situation assessment model uses a fuzzy comprehensive evaluation model to score the surrounding traffic operation situation based on the indicator system of the traffic hub operation situation.

[0070] Specifically, a traffic data fusion algorithm based on wavelet neural network is constructed to collect and extract macro traffic data around the comprehensive hub under different weather conditions and different types of traffic events. Traffic state discrimination indicators are selected, and traffic state discrimination parameters such as speed, flow, occupancy rate, headway, etc. are combined. The Greenshields model is used as the modeling basis to construct the traffic operation state evaluation index around the first-level comprehensive hub. The specific form is as follows: Where v is the average speed of vehicles in the traffic network, v f is the free flow speed, k is the traffic density of the traffic network, k j is the blocking density.

[0071] Based on the road network health index data such as the comprehensive regional congestion level, average road network speed, number of vehicles in transit, and the current number of traffic congestion incidents; the station security health index data such as the congestion level of the east and west drop-off platforms, the number of vehicles entering and leaving the platform on the day and the current average speed of vehicles, the total number of parking lots and the number of currently parked vehicles; and the station security health index data such as the number of passengers sent by the railway today, the current capacity (including subway, bus, taxi, online car-hailing and others) fullness index, the passenger flow to be dispatched and the current capacity situation, the hierarchical analysis method is used to calculate the evaluation weights of various second- and third-level indicators, and a multi-dimensional feature and fuzzy comprehensive model is established to evaluate the traffic operation situation around the hub.

[0072] Specifically, the short-term passenger flow of the hub is predicted using a time series model based on historical passenger flow data and real-time passenger flow data. The fuzzy evaluation method is used to form a fuzzy rule matrix based on the relationship between the average passenger flow density and passenger flow and the regional congestion level. Then, the characteristic data of the average passenger flow density and passenger flow are used as input, and the regional congestion status is inferred through fuzzy operations. The regional congestion level at the corresponding time, the traffic information of subways, buses and taxis, and the vehicle operation information are extracted to obtain unified characterization data, which are input into the online evaluation model of the operation status of urban comprehensive transportation hubs.

[0073] Among them, the prediction of short-term passenger flow of the hub using a time series model based on historical passenger flow data and real-time passenger flow data includes: using a clustering algorithm to classify passenger flow, and using a time series model to model different categories of passenger flow time series; training the SVM model, determining the kernel function and parameters, using the firefly algorithm to optimize the parameters of the RBF kernel function in the SVM model, and using the trained SVM model to perform nonlinear regression prediction on passenger flow to obtain passenger flow prediction data. The integrated data of average passenger flow density and passenger flow is used as input, and the degree of regional congestion is inferred through fuzzy operation, including:

[0074] Determine the evaluation index, which includes the average passenger flow density and passenger flow information, and construct the factor set U corresponding to the average passenger flow density P = {K1, K2} and the factor set U corresponding to the passenger flow information q ={S1,S2}; where the average passenger flow density K1 is the ratio of the number of passengers in a specific area to the area of ​​the area based on a fixed time period. For example, the ratio of the number of passengers in a subway station to the platform area during the morning rush hour. The average passenger flow density K2 is the passenger flow density at different entrances or exits in the same time period, reflecting the utilization rate of a specific channel. The passenger flow information S1 is the number of people entering and exiting each transportation station in a specific time period, and the passenger flow information S2 is the passenger flow at different time points in a specific transportation station.

[0075] The fuzzy set domains of the two evaluation indicators, average passenger flow density and passenger flow information, are defined as K1=K2={very small, small, medium, large, very large}, S1=S2={very small, small, medium, large, very large} respectively;

[0076] Based on the evaluation of the average passenger flow density and passenger flow information, thresholds are set for the average passenger flow density and passenger flow information respectively, and an evaluation set V = {v1, v2, v3, v4} is established. The regional congestion degree is used as the evaluation result of the passenger flow operation state of the comprehensive transportation hub, and is divided into four levels, V = {v1, v2, v3, v4} = {A, B, C, D}. In this embodiment, if K1, K2, S1 and S2 exceed the corresponding preset thresholds at the same time, it is determined to be level A, indicating that the passenger flow operation is severely congested; if only the average passenger flow density K1 or K2 exceeds the preset threshold, it is determined to be level B, indicating that the passenger flow operation is moderately congested; if only the passenger flow information S1 or S2 exceeds the corresponding preset threshold, it is determined to be level C, indicating that the passenger flow operation is locally congested; if K1, K2, S1 and S2 do not exceed the corresponding preset thresholds, it is determined to be level D, indicating that the passenger flow operation is unblocked.

[0077] The fuzzy relationship and membership function are determined according to the fuzzy control rules, and the regional congestion degree is calculated through the fuzzy relationship and membership function. The maximum membership method is used to perform anti-fuzzy operation to convert the fuzzy quantity corresponding to the regional congestion degree into an accurate quantity.

[0078] In this embodiment, the process of using the hub traffic operation status assessment model for evaluation is as follows: determine the evaluation target, the geographical scope and the time scope of the evaluation, screen the main characteristic indicators that reflect the operation status of the hub comprehensive transportation according to the evaluation target to obtain key indicators, and form an evaluation index system. The key indicators include three layers. The first layer of indicators are determined by using the hierarchical analysis method to calculate the evaluation weights of various second and third layer indicators, and establish a multi-dimensional feature and fuzzy comprehensive model to evaluate the traffic operation status around the hub.

[0079] Step S4: Input the extracted hub operation data features into each evaluation model to evaluate the traffic situation of the hub, and judge whether the corresponding warning level is reached according to each evaluation result, match the warning method based on the judgment result, and output the auxiliary traffic scheduling plan.

[0080] Specifically, the corresponding warning level is determined based on each evaluation result, and an alarm is issued based on the judgment result matching the warning method. The auxiliary traffic dispatch plan is output including:

[0081] Step S40: Generate corresponding alarm signals based on the traffic operation situation, hub capacity situation and traffic incident emergency situation obtained from each evaluation result, and carry out situation warning according to the generated alarm signal matching the preset alarm mode; the warning level is set to level 1 to level 3, and the higher the warning level, the more serious the current hub traffic situation;

[0082] For the first-level warning, it represents extreme passenger density or passenger volume, which may immediately affect public safety or cause serious congestion. The corresponding alarm mode is the highest level of alarm, prompting staff to conduct emergency evacuation, temporarily close entrances or some areas, increase security personnel investment, and urgently notify relevant departments; for the second-level warning, it represents passenger density or passenger volume exceeds the normal operating capacity but does not reach an extreme level. The corresponding alarm mode is a medium-level alarm, prompting staff to pay attention to possible escalating risks and take appropriate measures such as strengthening on-site management, optimizing passenger flow lines and broadcasting to remind passengers to pay attention to safety to prevent the situation from deteriorating; for the third-level warning, it represents a more common increase in passenger flow, such as an increase in passenger flow during holidays or events. The corresponding alarm mode is a low-level alarm, prompting staff to monitor and take appropriate preventive measures, such as notifying the public in advance to avoid peak hours, so as to optimize passenger flow and enhance passenger experience.

[0083] Step S41: According to the original travel data characteristics, hub operation characteristics and warning levels, a scheme recommendation algorithm based on cosine similarity is adopted to output the corresponding historical dispatch response scheme with the highest similarity in the traffic dispatch scheme database according to similarity matching.

[0084] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the embodiments of the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A traffic status monitoring and dynamic early warning method based on a group interaction model, characterized in that: The steps include: Obtaining comprehensive traffic information within the scope of the traffic hub, preprocessing the comprehensive traffic information, and extracting corresponding travel data features; Extracting individual travel data according to the extracted travel data features, building a group interaction model, and expanding the individual travel data to group travel data through the group interaction model to obtain hub operation data; Construct emergency event fuzzy assessment model, transport capacity operation status assessment model and hub transportation operation status assessment model; The extracted hub operation data is input into each evaluation model for evaluation, and it is judged whether the corresponding warning level is reached according to each evaluation result. An alarm is issued based on the judgment result and the warning method is matched, and an auxiliary traffic dispatch plan is output.

2. The traffic status monitoring and dynamic early warning method based on the group interaction model according to claim 1 is characterized in that: The step of acquiring comprehensive traffic information within the scope of the traffic hub and preprocessing the comprehensive traffic information includes: The comprehensive traffic information within the scope of the traffic hub includes real-time traffic data and passenger flow data, historical traffic data and passenger flow data; the preprocessing includes cleaning and storing the data, removing abnormal values ​​and erroneous data; The extracting of corresponding travel data features includes: Based on the acquired comprehensive traffic information, the Apriori algorithm and graph clustering algorithm are used to mine frequent item sets and perform hub correlation analysis, summarize the association rules between various types of data in the comprehensive traffic information, and fuse the data sets of various types of data in the comprehensive traffic information according to the association rules. The travel data features are extracted from the fused data sets, and the travel data features include the features of various travel modes, travel time periods, travel groups, transfer behaviors, and activity types.

3. The traffic status monitoring and dynamic early warning method based on group interaction model according to claim 1 is characterized in that: The extracting of individual travel data according to the extracted travel data features comprises: Establish a spatiotemporal threshold sliding window mechanism to slice travel data features and cluster them according to activity combination patterns; The key elements for identifying the transfer passenger flow are extracted from the travel data feature slices of different clustering categories, and the key elements corresponding to each clustering category are used to enhance the travel data feature slices; the key elements for identifying the transfer passenger flow include the travel start and end points, transfer nodes, transfer time periods and time consumption, and transfer paths of individuals or groups; The enhanced travel data feature slices are used as input, and the multivariate framed and nonlinear time series of pedestrian and vehicle trajectories are identified through graph convolutional neural networks, and the node feature matrix of the pedestrian and vehicle trajectories under multimodal travel conditions is obtained. The node feature matrix of the pedestrian and vehicle trajectories is composed of the node feature representations of each node on the trajectory, and the feature representation of each node in the node feature matrix is ​​updated by aggregating the information of neighboring nodes to obtain the monitoring results of the pedestrian and vehicle trajectories within the hub range; Based on the monitoring results of the trajectories of people and vehicles within the hub, combined with the historical traffic data in the comprehensive traffic information and using the extended Markov chain model, the future trajectories of people and vehicles within the hub are predicted; the predicted results of the trajectories of people and vehicles within the hub are combined with the monitoring results to obtain individual travel data.

4. The traffic status monitoring and dynamic early warning method based on group interaction model according to claim 1 is characterized in that: The expanding individual travel data to group travel data through the group interaction model includes: The individual travel data is input into the group interaction model. The group interaction model abstracts each travel individual represented by the individual travel data into an entity that makes decisions based on its own status and the status of neighboring individuals, analyzes the interaction between each travel individual, and then expands the individual travel data to group travel data, that is, obtaining the hub operation data.

5. The traffic status monitoring and dynamic early warning method based on group interaction model according to claim 1 is characterized in that: The emergency event fuzzy assessment model includes: Construct the first-level monitoring indicators and second-level monitoring indicators for emergency events. The first-level monitoring dimension indicators include emergency events and equipment alarms. The second-level monitoring dimension indicators under emergency events include extreme weather, fire accidents, and large passenger flow events. The second-level monitoring dimension indicators under equipment alarms include temperature, humidity, smoke, and video equipment alarms. The weights of various indicators are calculated using the hierarchical entropy analysis method, as follows: Among them, w j is the weight of the j-th secondary monitoring dimension indicator, E j is the information entropy of the indicator, n is the number of indicators in the secondary monitoring dimension, p ij is the proportion of the jth secondary monitoring dimension indicator in the i-th emergency event; Based on the hub operation data, calculate the secondary monitoring indicators and their weights for emergency events, and sum them up to get the comprehensive score of emergency situation to evaluate the probability and severity of emergency events. Specifically: Among them, s i is the comprehensive score of the probability and severity of the ith emergency event, n is the number of secondary monitoring dimension indicators for emergency events, x ij is the i-th value of the j-th secondary monitoring dimension indicator; Determine the risk status level to which the comprehensive score of the emergency situation belongs, wherein the risk status level includes four levels: very high, high, medium, and low; and output the risk status level judgment result as the evaluation result of the comprehensive score model of the emergency situation.

6. The traffic status monitoring and dynamic early warning method based on group interaction model according to claim 1 is characterized in that: The capacity operation status assessment model performs assessment including: constructing capacity evaluation indicators including passenger flow diffusion response time, capacity loading rate, capacity filling coefficient and waiting time, and analyzing the capacity evaluation indicators in real time according to hub operation data; if one of the capacity evaluation indicators exceeds a preset threshold, it is judged that the current capacity operation status is abnormal; if not, the current capacity operation status is normal.

7. The traffic status monitoring and dynamic early warning method based on group interaction model according to claim 1 is characterized in that: The evaluation of the hub traffic operation status evaluation model includes: constructing two evaluation indicators, average passenger flow density and passenger flow information, the average passenger flow density and passenger flow information are respectively divided into five levels, judging the current level of the average passenger flow density and passenger flow information according to the hub operation data, and judging the regional congestion level by comprehensively considering the current level of the average passenger flow density and passenger flow information, and the regional congestion level is the hub traffic operation status evaluation result.

8. The traffic status monitoring and dynamic early warning method based on group interaction model according to claim 1 is characterized in that: The method of judging whether the corresponding warning level is reached according to each evaluation result, giving an alarm based on the judgment result matching the warning mode, and outputting the auxiliary traffic dispatch plan includes: Based on the traffic operation situation, hub capacity situation and traffic incident emergency situation obtained from each evaluation result, it is judged whether the corresponding warning level is reached. The maximum warning level is level 3, and a corresponding alarm signal is generated. According to the generated alarm signal, a preset alarm mode is matched to issue a situation warning. If the minimum warning level is not reached, no alarm is issued. According to the original travel data characteristics, hub operation data and current warning level, the corresponding historical dispatch response plan with the highest similarity is output according to similarity matching.

Images