Traffic jam control method based on oversampling GPS data

Abstract

The invention discloses a traffic jam control method based on oversampling GPS data, and relates to the technical field of traffic jam control, and the method comprises the following steps: collecting urban traffic historical data during an activity; the method comprises the following steps of: preprocessing collected urban traffic historical data, simulating a vehicle flow during an activity based on oversampling, and performing data processing by using a Grehilz speed-density linear model to obtain vehicle GPS track data; inputting the GPS track data of the vehicle into a grid traffic jam identification model, and carrying out the identification of grid traffic jam; and based on an identification result, performing index evaluation on different original traffic jam control methods by using a VISSIM simulation model, and selecting a plan of an optimal traffic jam control method to realize traffic jam control based on oversampling GPS data. The problems that an existing method has limitation, data collection cost is high, calculation difficulty is large, and dense traffic flow acquisition or real-time traffic jam area analysis is difficult to achieve are solved.

Description

Technical Field

[0001] The invention relates to the technical field of traffic congestion control, and in particular to a traffic congestion control method based on over-sampling GPS data. Background Art

[0002] With the comprehensive improvement of national strength, social development has also made significant progress. Major cities in my country have begun to host large-scale international and national events. During large-scale events, population and vehicles gather, causing a sharp increase in traffic demand. The congestion of large-scale events is different from daily congestion. It is related to the location and scale. If key sections are not controlled in time, the scope of congestion will continue to expand, thus affecting the progress of the event. To this end, event organizers should predict congested sections in advance and formulate corresponding traffic control methods.

[0003] At present, the research on traffic congestion control methods for large-scale events mainly proposes relevant traffic congestion control methods based on the characteristics of different types of events and the actual situation of the road network around the event venues; or predicts the flow of people and traffic at large-scale events through statistical methods such as Bayesian, and uses the prediction results as the premise of traffic congestion control methods to formulate specific traffic congestion control measures for the event.

[0004] Most of the methods related to traffic flow identification of urban traffic congestion use traditional traffic congestion detection data such as loop detector data, video data, and floating vehicle data. This type of data requires fixed equipment for collection, has high collection costs, and has certain limitations on the scope of data collection.

[0005] At the same time, most algorithms for mining traffic congestion using GPS data of operating vehicles need to be combined with map matching algorithms. The process of matching a large amount of GPS data with maps is computationally difficult and time-consuming, and it is difficult to achieve real-time analysis of traffic congestion areas. Summary of the invention

[0006] In view of the above-mentioned deficiencies in the prior art, the present invention provides a traffic congestion control method based on oversampled GPS data to solve the limitations of the existing methods, including high data collection costs, great computational difficulty, and difficulty in performing real-time traffic congestion area analysis.

[0007] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: a traffic congestion control method based on oversampled GPS data, comprising the following steps: S1: Collect historical urban traffic data during the activity; S2: Preprocess the collected urban traffic historical data, simulate the vehicle flow during the activity based on oversampling, and use the Greenshields speed-density linear model for data processing to obtain vehicle GPS trajectory data; S3: inputting vehicle GPS trajectory data into a grid traffic congestion identification model to identify grid traffic congestion; S4: Based on the identification results, the VISSIM simulation model is used to evaluate the indicators of the original different traffic congestion control methods, and the plan of the optimal traffic congestion control method is selected to realize traffic congestion control based on oversampled GPS data.

[0008] The beneficial effect of the above scheme is that the present invention simulates the trajectory travel data during large-scale events through ticket registration information and historical traffic data of large-scale events. The generated data is put into the model to identify key congestion areas, including congested sections, congestion ranges, congestion directions and congestion times, and the traffic flow characteristics of the congestion range are extracted and put into the simulation model to generate indicators to evaluate the feasibility of different traffic congestion control methods, so as to pre-formulate a better traffic congestion control method, which solves the problems of high data collection cost, high calculation difficulty and difficulty in real-time traffic congestion area analysis in existing methods.

[0009] Furthermore, the urban traffic history data in S1 includes GPS data of operating vehicles, loop detector data and video data; The fields of the operating vehicle GPS data are: VehicleId, Time, Lon, Lat, Speed, Direction, Mileage and State; The VehicleId is the license plate number, which is used to uniquely identify the operating vehicle; The Time is the collection time of the GPS data of the operating vehicle; The Lon and Lat are the latitude and longitude coordinates of the operating vehicle's GPS data, which are used to determine the location of the operating vehicle; The Speed ​​is the speed of the operating vehicle, which is used to record the instantaneous speed of the operating vehicle at the time of uploading the GPS track; The Direction is the direction angle of the operating vehicle, which is used to record the direction of the vehicle head when the GPS track is uploaded; Mileage is the mileage of the operating vehicle, which is used to record the total mileage of the operating vehicle at the time of uploading the GPS track; The State is the operating state of the vehicle, including two values ​​0 and 1. The value 0 indicates an empty vehicle, and the value 1 indicates a passenger-carrying state.

[0010] The beneficial effect of the above further scheme is that the acquired GPS data of the operating vehicle contains the above 8 fields. Through the above technical scheme, the GPS data of the vehicle can be acquired for subsequent model identification.

[0011] Furthermore, S2 includes the following sub-steps: S2-1: After cleaning and deduplication of the operating vehicle GPS data, construct an operating vehicle GPS trajectory data set T, and create a new set D for storing the newly added trajectory data; S2-2: Divide the road network in the study area into grids and determine whether the operating vehicle GPS trajectory dataset T is empty. If so, proceed to step S2-5; otherwise, proceed to step S2-3: S2-3: Select a track Traj from the operating vehicle GPS track data set T, delete the track Traj from the data set T, and perform oversampling and densification on the track Traj; S2-4: storing the oversampled and densified trajectory data and the original trajectory data in a set D, and returning to step S2-2; S2-5: Create a new set G, and divide the data in set D according to the grid division range, and store them in set G in units of grids; S2-6: Determine whether the set G is empty, if so, proceed to step S2-9, otherwise proceed to step S2-7; S2-7: Select a grid data set Grid_ij from the set G, and delete the data set Grid_ij from the set G; S2-8: Aggregate the trajectories in the dataset Grid_ij within a specific time and calculate the traffic density of the grid , and calculate the average vehicle speed of the grid in combination with the Greenshields speed-density linear model, and return to step S2-6; The traffic density of the grid The calculation formula is:

[0012] in, is the number of vehicles on the road at a certain moment, is the length of the road; S2-9: The speed aggregated to a specific time is matched with the grid number to the newly added trajectory data, and the vehicle mileage is calculated according to the latitude and longitude coordinates of each trajectory segment. The time of the trajectory data is adjusted in combination with the vehicle's driving speed to obtain the vehicle GPS trajectory data.

[0013] The beneficial effect of the above further scheme is: based on the vehicle trajectory data of the urban road network under normal conditions, combined with the Greenshields speed-density linear model, the vehicle trajectory data during large-scale events is simulated, and the GPS trajectory is generated based on the oversampling method.

[0014] Furthermore, constructing the operating vehicle GPS trajectory dataset T in S2-1 includes the following sub-steps: S2-1-1: Filter all GPS point data of the same VehicleId, recorded as set P; S2-1-2: Create new trajectory data and ; S2-1-3: Sort all the trajectory data in set P by Time, and determine whether set P is an empty set. If so, output the trajectory data as the operating vehicle GPS trajectory data set T, otherwise proceed to step S2-1-4; S2-1-4: Select the earliest trajectory point in set P , and the trajectory points Delete from the set P and determine the trajectory point Is the operating status equal to the trajectory point , if yes, go to step S2-1-5, otherwise go to step S2-1-6; S2-1-5: Determine the trajectory point Is the operating status of 1? If so, the trajectory point Adding track data , otherwise the trajectory point Add track data , and return to step S2-1-3; S2-1-6: Create a new track number in the track data Increase by 1 and return to step S2-1-5.

[0015] The beneficial effect of the above further scheme is that the present invention needs to model the GPS trajectory data of the operating vehicle. The original trajectory data of the operating vehicle is based on GPS points, and each data record is the time and space information of a point. However, the identification of traffic congestion requires the continuous time and speed changes of a trip, so the point data needs to be tracked.

[0016] Furthermore, the oversampling and densification of the trajectory Traj in S2-3 includes the following sub-steps: S2-3-1: Divide the trajectory Traj into GPS points, and insert a set number of GPS points equidistantly between two adjacent GPS points; S2-3-2: Calculate the longitude and latitude coordinates of the inserted GPS points, and generate a new track for each GPS point with the same serial number between two adjacent points; S2-3-3: Simulate the newly added trajectory as the trajectory of the newly added vehicle to achieve oversampling and densification of the trajectory Traj.

[0017] The beneficial effect of the above further scheme is: by predicting the traffic demand of the road network during large-scale events, the number of trajectories that need to be increased is determined in combination with the number of GPS data trajectories of existing operating vehicles, thereby achieving the purpose of increasing the number of trajectories.

[0018] Furthermore, the grid traffic congestion identification model in S3 combines the number of trajectories in the grid area with the average vehicle speed inside the grid to form a two-dimensional vector Indicates the congestion pattern within the grid and the congestion pattern at different time periods and The difference is measured by the Mahalanobis distance, the formula is:

[0019] in, is the Mahalanobis distance, is the covariance matrix of the whole sample, is the transpose of the matrix; The difference between the congestion pattern of a specific grid and the congestion pattern of the whole sample for the whole day is calculated as:

[0020] in, For the A congestion mode, For the whole day The mean of the congestion pattern in the time period, As the basis for judging the degree of congestion, The bigger it is, the more severe the congestion. This is the situation under free flow state;

[0021] in, is the mean value under free flow state; Constraints are imposed on the average speed and number of trajectories of the vehicles. When the constraints are met, the grid is Internal congestion, the constraint formula is:

[0022]

[0023] in, For the time period The number of inner tracks, For the time period The average speed inside is the average number of trajectories of the grid over the entire day, is the average speed of the grid throughout the day.

[0024] The beneficial effect of the above further solution is: through the above technical solution, the average speed and number of trajectories of each grid within each time step range are calculated, and the difference between the average speed and number of trajectories of the grid within the range of the whole day (0:00-24:00) is used to determine whether the grid is congested in the time period. The difference between the grid congestion pattern of each time step and the grid congestion pattern of the whole day is measured by the Mahalanobis distance, and the larger the Mahalanobis distance, the greater the difference.

[0025] Furthermore, in S3, grid traffic congestion is identified, and the degree of congestion is determined specifically as follows:

[0026] in, To index the congestion level; when When , the grid traffic congestion is determined to be unblocked; when , and is not satisfied and When , the grid traffic congestion is determined to be unblocked; when , and satisfies , not satisfied When , the grid traffic congestion is determined to be unblocked; when , and is not satisfied ,satisfy When , the grid traffic congestion is determined to be in a slow-moving state; when , and satisfies and When , the grid traffic congestion is determined to be in a general congestion state; when , and is not satisfied and When , the grid traffic congestion is determined to be unblocked; when , and satisfies , not satisfied When , the grid traffic congestion is determined to be unblocked; when , and is not satisfied ,satisfy When , the grid traffic congestion is determined to be in a general congestion state; when , and satisfies and When , the grid traffic congestion is judged to be in a serious state; when , and is not satisfied and When , the grid traffic congestion is determined to be unblocked; when , and satisfies , not satisfied When , the grid traffic congestion is determined to be unblocked; when , and is not satisfied ,satisfy When , the grid traffic congestion is determined to be in a severe congestion state; when , and satisfies and When , the grid traffic congestion is determined to be in a severe congestion state.

[0027] The beneficial effect of the above further scheme is: through the above technical scheme, the congestion degree of the grid is determined, the average speed and number of trajectories of the grid in the current time period are used to construct the congestion pattern, and the difference between the congestion pattern of the all-day data sample is used to determine whether the grid is congested.

[0028] Furthermore, the S4 comprises the following sub-steps: S4-1: Collect relevant road data in the study area and obtain road network files; S4-2: Based on the road data and road network files, use VISSIM to simulate the road network in the study area and establish a VISSIM simulation model; S4-3: Based on the VISSIM simulation model, the indicators of the original different traffic congestion control methods are analyzed by adjusting the simulation parameters, and the optimal traffic congestion control method is selected to realize traffic congestion control based on oversampled GPS data.

[0029] The beneficial effect of the above further scheme is that in order to construct a simulation model of road network sections with high incidence of road congestion during large-scale events, it is necessary to investigate the actual situation of the road network to ensure the accuracy of the road network model.

[0030] Furthermore, the indicator includes the average delay , average speed , average queue length and traffic density .

[0031] Furthermore, the average delay , average speed , average queue length and traffic density The calculation formula is:

[0032] in, For road section Traffic delays, For road section Traffic volume;

[0033] in, is the distance traveled by the vehicle on the road section, For the The travel time of a vehicle on a road segment, For traffic volume;

[0034] in, For road section The length of the queue of moving vehicles, is the number of vehicle queues on all road segments;

[0035] in, is the number of vehicles on the road at a certain moment, For road section The traffic density, For road section Length, is the total length of the road network.

[0036] The beneficial effect of the above further scheme is: in order to optimize the original traffic congestion control method, the traffic flow characteristic indicators of the road are analyzed through simulation parameter adjustment, and the main evaluation indicators include average delay, average speed, queue length and traffic density. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 The figure is a flow chart of a traffic congestion control method based on oversampled GPS data.

[0038] Figure 2 Schematic diagram of the key links of a traffic congestion control method based on oversampled GPS data.

[0039] Figure 3 Simulate a flow chart for large-scale event GPS trajectory data.

[0040] Figure 4 Extract flow chart for raw trajectory.

[0041] Figure 5 Schematic diagram of trajectory oversampling and densification.

[0042] Figure 6 Modeling flow chart for traffic simulation.

[0043] Figure 7 Figure 2 is a graph showing the relationship between the number of grid trajectories and the spatial step size.

[0044] Figure 8 This is a graph showing the relationship between the failure grid ratio and the spatial step size.

[0045] Fig. 9 Figure 2 is a graph showing the relationship between the failure mesh ratio and the time step. DETAILED DESCRIPTION

[0046] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0047] like Figure 1 and Figure 2 As shown, a traffic congestion control method based on over-sampled GPS data comprises the following steps: S1: Collect historical urban traffic data during the activity; S2: Preprocess the collected urban traffic historical data, simulate the vehicle flow during the activity based on oversampling, and use the Greenshields speed-density linear model for data processing to obtain vehicle GPS trajectory data; S3: inputting vehicle GPS trajectory data into a grid traffic congestion identification model to identify grid traffic congestion; S4: Based on the identification results, the VISSIM simulation model is used to evaluate the indicators of the original different traffic congestion control methods, and the plan of the optimal traffic congestion control method is selected to realize traffic congestion control based on oversampled GPS data.

[0048] The urban traffic history data in S1 includes GPS data of operating vehicles, loop detector data and video data; Get loop detector data. Extract the traffic volume, vehicle density and average speed of the road section from the loop detector and video data. Combined with the three-parameter model of traffic flow, the relationship between vehicle density and vehicle speed between each road section can be calculated. The three parameters of traffic flow refer to vehicle density, driving speed and traffic volume, which are the three basic parameters of traffic flow characteristics. The relationship between the three parameters is:

[0049] Where Q represents the traffic volume (vehicles / h), K represents the average traffic density; The fields of the operating vehicle GPS data are: VehicleId, Time, Lon, Lat, Speed, Direction, Mileage and State; The VehicleId is the license plate number, which is used to uniquely identify the operating vehicle; The Time is the collection time of the GPS data of the operating vehicle; The Lon and Lat are the latitude and longitude coordinates of the operating vehicle's GPS data, which are used to determine the location of the operating vehicle; The Speed ​​is the speed of the operating vehicle, which is used to record the instantaneous speed of the operating vehicle at the time of uploading the GPS track; The Direction is the direction angle of the operating vehicle, which is used to record the direction of the vehicle head when the GPS track is uploaded; Mileage is the mileage of the operating vehicle, which is used to record the total mileage of the operating vehicle at the time of uploading the GPS track; The State is the operating state of the vehicle, including two values ​​0 and 1. The value 0 indicates an empty vehicle, and the value 1 indicates a passenger-carrying state.

[0050] Cleaning of GPS data of operating vehicles. This step mainly includes: identification and processing of duplicate data, identification and processing of invalid data, identification and processing of parking data, and screening of data with operating status of passenger carrying as research data.

[0051] Based on the vehicle trajectory data of the urban road network under normal conditions, the three-parameter model is combined to simulate the vehicle trajectory data during large-scale events. The process is as follows: Figure 3 As shown in the figure, the original operating vehicle GPS data is grouped according to different vehicle numbers VehicleId and track numbers Traj. The steps for simulating track data during large-scale activities are as follows: The S2 includes the following sub-steps: S2-1: After cleaning and deduplication of the operating vehicle GPS data, construct an operating vehicle GPS trajectory data set T, and create a new set D for storing the newly added trajectory data; S2-2: Divide the road network in the study area into grids and determine whether the operating vehicle GPS trajectory dataset T is empty. If so, proceed to step S2-5; otherwise, proceed to step S2-3: S2-3: Select a track Traj from the operating vehicle GPS track data set T, delete the track Traj from the data set T, and perform oversampling and densification on the track Traj; S2-4: storing the oversampled and densified trajectory data and the original trajectory data in a set D, and returning to step S2-2; S2-5: Create a new set G, and divide the data in set D according to the grid division range, and store them in set G in units of grids; S2-6: Determine whether the set G is empty, if so, proceed to step S2-9, otherwise proceed to step S2-7; S2-7: Select a grid data set Grid_ij from the set G, and delete the data set Grid_ij from the set G; S2-8: Aggregate the trajectories in the dataset Grid_ij within a specific time and calculate the traffic density of the grid , and calculate the average vehicle speed of the grid in combination with the Greenshields speed-density linear model, and return to step S2-6; The traffic density of the grid The calculation formula is:

[0052] in, is the number of vehicles on the road at a certain moment, is the length of the road; S2-9: The speed aggregated to a specific time is matched with the grid number to the newly added trajectory data, and the vehicle mileage is calculated according to the latitude and longitude coordinates of each trajectory segment. The time of the trajectory data is adjusted in combination with the vehicle's driving speed to obtain the vehicle GPS trajectory data.

[0053] The present invention needs to model the GPS trajectory data of operating vehicles. The original trajectory data of operating vehicles is based on GPS points, and each piece of data records the time and space information of a point. However, the identification of traffic congestion requires the continuous time and speed changes of a trip, so it is necessary to track the point data. A point in the GPS trajectory data is defined as:

[0054] Among them, VehicleId represents the vehicle number, Lat and Lon represent the latitude and longitude of GPS positioning, Time represents the time when the GPS point is positioned, State represents the operating status of the operating vehicle when the GPS point is positioned, and Speed ​​represents the speed of the operating vehicle at the GPS point.

[0055] Trajectory data: A segment of the operating vehicle’s journey (i.e., a segment in which the operating state of the operating vehicle is continuously 0 or continuously 1) is recorded as a trajectory. A trajectory consists of n GPS points, so a segment of trajectory data is defined as:

[0056] Where TrajID represents the number of the trajectory, which is the unique identification mark of the trajectory; It indicates the set of points in this track sorted in time order; State indicates whether the operating vehicle is unloaded (State=0) or in operation (State=1) in this track. VehicleId indicates which operating vehicle this track belongs to.

[0057] According to the above definition of GPS trajectory data, it is necessary to classify the operating vehicle GPS data according to the vehicle operating state and vehicle number VehicleId, identify and extract the corresponding trajectory data. The extraction trajectory flow chart is as follows Figure 4 As shown in the figure, the original operating vehicle GPS data is grouped according to different VehicleIds, and the steps for extracting tracks according to different vehicle numbers are as follows: The construction of the operating vehicle GPS trajectory dataset T in S2-1 includes the following steps: S2-1-1: Filter all GPS point data of the same VehicleId, recorded as set P; S2-1-2: Create new trajectory data and ; S2-1-3: Sort all the trajectory data in set P by Time, and determine whether set P is an empty set. If so, output the trajectory data as the operating vehicle GPS trajectory data set T, otherwise proceed to step S2-1-4; S2-1-4: Select the earliest trajectory point in set P , and the trajectory points Delete from the set P and determine the trajectory point Is the operating status equal to the trajectory point , if yes, go to step S2-1-5, otherwise go to step S2-1-6; S2-1-5: Determine the trajectory point Is the operating status of 1? If so, the trajectory point Add track data , otherwise the trajectory point Add track data , and return to step S2-1-3; S2-1-6: Create a new track number in the track data Increase by 1 and return to step S2-1-5.

[0058] Gridding the road network in the study area: Since urban road network data has the characteristics of complex structure, low accuracy, and high time-variability, it is very costly to obtain accurate road network data. And directly analyzing and processing the GPS data of operating vehicles in conjunction with the road network will require extremely high computer configuration and a long time to calculate and process the data. Therefore, the road network data is gridded.

[0059] The urban road network is divided into square grids, the grid numbers can be defined as:

[0060] Among them, a represents the horizontal number of the grid, and b represents the vertical number of the grid.

[0061] The steps for grid drawing are as follows: ① Select the longitude and latitude of the grid drawing starting point , start drawing the grid from the upper left; ② Determine the grid step length L (unit: meter), and determine the number of grids based on the actual grid step length ( ), thereby determining the longitude and latitude step of each grid. The longitude and latitude step formula is as follows:

[0062]

[0063] in, , Indicates the latitude and longitude step of the grid, is the latitude range of the study area.

[0064] ③ Calculate the starting longitude and latitude of all grids, the formula is as follows, and draw the grids.

[0065]

[0066]

[0067] in,( ) is the grid The starting longitude and latitude of .

[0068] The urban area grid can reflect the traffic flow information of the urban traffic road network within a period of time and is a special traffic status detector. The step size of the grid and the division of the time period are called the spatial step size and the time step size. Different step sizes show different traffic flow information. The value of the urban area spatial step size is negatively correlated with the number of grids in the urban area. The number of grids increases as the spatial step size decreases, that is, the grid granularity also increases as the spatial step size decreases. Similar to the model spatial step size, the smaller the value of the time step size, the more subtle the traffic conditions of the city can be shown in the time dimension.

[0069] In S2-3, the trajectory Traj is subjected to oversampling densification, including the following sub-steps: S2-3-1: Divide the trajectory Traj into GPS points, and insert a set number of GPS points equidistantly between two adjacent GPS points; S2-3-2: Calculate the longitude and latitude coordinates of the inserted GPS points, and generate a new track for each GPS point with the same serial number between two adjacent points; S2-3-3: Simulate the newly added trajectory as the trajectory of the newly added vehicle to achieve oversampling and densification of the trajectory Traj.

[0070] In this embodiment, the traffic demand of the road network during the large-scale event is predicted, and the number of tracks that need to be increased is determined in combination with the number of GPS data tracks of existing operating vehicles. Assuming that there is one vehicle track on the road section under normal traffic network conditions, and it is estimated that there will be 10 tracks during the large-scale event, the number of tracks needs to be increased by 9 times. The specific operation of increasing these 9 times vehicle tracks is as follows: Figure 5 As shown in the figure, this track segment consists of 5 GPS points. 9 GPS points are inserted equidistantly between two adjacent points. The longitude and latitude coordinates of the 9 GPS points are calculated. A new track is generated for each GPS point with the same serial number between two adjacent points. For example, the first newly added track is recorded as , the newly added 9 trajectories are simulated as 9 trajectories of 9 newly added vehicles, thereby achieving the purpose of adding the number of trajectories.

[0071] Construct a grid traffic congestion identification model: Calculate the average speed V and the number of trajectories N of each grid within each time step, and judge whether the grid is congested in this time period by the difference between the average speed and the number of trajectories of the grid within the range of the whole day (0:00-24:00). The difference between the grid congestion pattern of each time step and the grid congestion pattern of the whole day is measured by the Mahalanobis distance. The larger the Mahalanobis distance, the greater the difference.

[0072] The grid traffic congestion identification model in S3 combines the number of trajectories in the grid area with the average vehicle speed inside the grid to form a two-dimensional vector Indicates the congestion pattern within the grid and the congestion pattern at different time periods and The difference is measured by the Mahalanobis distance, the formula is:

[0073] in, is the Mahalanobis distance, is the covariance matrix of the whole sample, is the transpose of the matrix; The difference between the congestion pattern of a specific grid and the congestion pattern of the whole sample for the whole day is calculated as:

[0074] in, For the A congestion mode, For the whole day The average of the congestion patterns in the time period, As a basis for judging the degree of congestion, The bigger it is, the more severe the congestion. This is the situation under free flow state;

[0075] in, is the mean value under free flow state; In order to accurately extract the aggregation phenomenon of low-speed vehicles from the GPS trajectory data of operating vehicles, it is necessary to constrain the average speed and number of trajectories of the vehicles: Constraints are imposed on the average speed and number of trajectories of the vehicles. When the constraints are met, the grid is Internal congestion, the constraint formula is:

[0076]

[0077] in, For the time period The number of inner tracks, For the time period The average speed inside is the average number of trajectories of the grid over the entire day, is the average speed of the grid throughout the day.

[0078] In S3, grid traffic congestion is identified, and the degree of congestion is determined as follows:

[0079] in, Index for congestion level classification, index It is calculated by dividing the difference between the congestion pattern at a certain time step and the congestion pattern of the entire day by the mean of the total difference for the entire day. The larger the index is, the more serious the congestion within the grid is.

[0080] when When , the grid traffic congestion is determined to be unblocked; when , and is not satisfied and When , the grid traffic congestion is determined to be unblocked; when , and satisfies , not satisfied When , the grid traffic congestion is determined to be unblocked; when , and is not satisfied ,satisfy When , the grid traffic congestion is determined to be in a slow-moving state; when , and satisfies and When , the grid traffic congestion is determined to be in a general congestion state; when , and is not satisfied and When , the grid traffic congestion is determined to be unblocked; when , and satisfies , not satisfied When , the grid traffic congestion is determined to be unblocked; when , and is not satisfied ,satisfy When , the grid traffic congestion is determined to be in a general congestion state; when , and satisfies and When , the grid traffic congestion is judged to be in a serious state; when , and is not satisfied and When , the grid traffic congestion is determined to be unblocked; when , and satisfies , not satisfied When , the grid traffic congestion is determined to be unblocked; when , and is not satisfied ,satisfy When , the grid traffic congestion is determined to be in a severe congestion state; when , and satisfies and When , the grid traffic congestion is determined to be in a severe congestion state.

[0081] The present invention controls traffic congestion from five aspects: traffic facility planning, road traffic management, parking traffic control, public transportation control and pedestrian traffic control.

[0082] To build a simulation model of high-incidence road congestion sections during large-scale events, it is necessary to investigate the actual situation of the road network to ensure the accuracy of the road network model. Figure 6 As shown: The S4 comprises the following sub-steps: S4-1: Collect relevant road data in the study area and obtain road network files; By mining the GPS trajectory data of operating vehicles simulated during large-scale events, we determined the sections of the road network with high congestion incidence as the research areas for simulation modeling, and collected relevant road data in the research areas, including road types, road capacity, intersections, traffic signals, and bus station routes; S4-2: Based on the road data and road network files, use VISSIM to simulate the road network in the study area and establish a VISSIM simulation model; Combined with the relevant road traffic data collected in the study area, the VISSIM software was used to reproduce the road network in the study area, the simulation parameters were adjusted to establish a traffic simulation model, and the traffic signal and traffic demand data were set according to the actual situation; S4-3: Based on the VISSIM simulation model, the indicators of the original different traffic congestion control methods are analyzed by adjusting the simulation parameters, and the optimal traffic congestion control method is selected to realize traffic congestion control based on oversampled GPS data.

[0083] In order to optimize the original traffic congestion control method, the traffic flow characteristic indicators of the road are analyzed by adjusting the simulation parameters. The main evaluation indicators include average delay, average speed, queue length and traffic density. Traffic delay is calculated by the difference between the actual driving time of the vehicle on the driving section and the time the vehicle drives in the free flow state of the section. The average delay is the average of the traffic delays of all vehicles on the road; the average speed refers to the average of the speeds of all vehicles on the section, calculated by the ratio of the length of the vehicle on the section to the time the vehicle travels on the section; the average queue length refers to the average of the queue lengths of vehicles on all sections in the study area within the divided time interval; the traffic density refers to the density of vehicles on the section, that is, the number of vehicles on a unit length section at a certain moment.

[0084] The indicators include average delay , average speed , average queue length and traffic density .

[0085] The average delay , average speed , average queue length and traffic density The calculation formula is:

[0086] in, For road section Traffic delays, For road section Traffic volume;

[0087] in, is the distance traveled by the vehicle on the road section, For the The travel time of a vehicle on a road segment, For traffic volume;

[0088] in, For road section The length of the queue of moving vehicles, is the number of vehicle queues on all road segments;

[0089] in, is the number of vehicles on the road at a certain moment, For road section The traffic density, For road section Length, is the total length of the road network.

[0090] In one embodiment of the present invention, since the value of the spatial step length of the urban area is negatively correlated with the number of grids in the urban area, the number of grids increases as the value of the spatial step length decreases, that is, the grid granularity also increases as the spatial step length decreases. Finer grid granularity can show the subtle traffic conditions of the city in the spatial dimension, but it makes more grids invalid. Even in the grids divided by the spatial step length, a large number of grids do not have the GPS point information of operating vehicles, which increases the noise of the final data and directly affects the accuracy of traffic congestion judgment. Therefore, the setting of the spatial step length needs to consider the effective information of the grid, that is, there must be a sufficient number of effective trajectories inside the grid; at the same time, it is also necessary to consider the proportion of invalid grids, and a smaller proportion has a faster computing speed. Similar to the model spatial step length, the smaller the value of the time step length, the more subtle traffic conditions of the city can be shown in the time dimension. However, it will also lead to an increase in the proportion of invalid grids, increase the noise of the final data, and directly affect the accuracy of traffic congestion judgment.

[0091] This embodiment selects the area within the Chengdu ring road as the research object for analysis, with a longitude range of [103.92918, 104.21106] and a latitude range of [30.526748, 30.796598]. The time range of the selected data is from 06:00 in the morning to 24:00 in the evening. There are fewer vehicles in the urban area road network in the early morning, and it is difficult to form congestion. Statistics are made on the average number of GPS tracks per hour in the grid and the number and ratio of failed grids under different spatial step settings.

[0092] Determine the spatial step length. Draw a graph of the relationship between the number of grid tracks and the spatial step length based on the statistical results. Figure 7 The relationship between the failure mesh ratio and the spatial step size Figure 8 .Depend on Figure 7It can be seen that the number of trajectories in a grid is exponentially related to the spatial step length. When the spatial step length is 100 to 250m, the number of trajectories in the grid increases slowly. When the spatial step length is greater than 250m, the number of trajectories inside the grid begins to rise sharply. Figure 8 It can be seen that the spatial step length is exponentially related to the proportion of failed grids. When the spatial step length of the grid division is less than 200m, the proportion of failed grids decreases rapidly. Failed grids refer to grids without GPS track information. When the spatial step length of the division is smaller, there will be areas such as communities, schools and shopping malls without GPS information, resulting in more grid failures. The increase in the spatial step length will lead to an increase in the area of ​​the grid, and the possibility of GPS information falling into the grid will be greater, so the number of failed grids will decrease. It is meaningless to put failed grids into the model, and the increase in the proportion of failed grids will only increase the calculation time of the model. In summary, in this case, the spatial step length of the urban area grid division is selected as 200m.

[0093] Determine the time step. The area within the Chengdu ring road was selected as the research object for analysis. The time range of the selected data was 06:00 in the morning to 24:00 in the evening, and the grid step was 200m. The number of trajectories inside the grid in different time steps was counted, and a line graph of the failure grid and the time step was drawn. Fig. 9 As shown in the figure, the proportion of failed grids is exponentially related to the time step. When the time step is set to be greater than 10 minutes, the proportion of failed grids in the urban area slowly decreases, so the time step is selected as 10 minutes.

[0094] Large-scale events will lead to more pedestrians and vehicles gathering, thus inducing traffic congestion in urban road networks. Different from the regular congestion during the morning and evening rush hours on weekdays, the traffic congestion induced during large-scale events is random and regional, and there is almost no regularity in the occurrence of congestion. If the traffic congestion is not discovered and alleviated in time, it will cause a large number of vehicle delays. The present invention takes the Chengdu Sports Center as an example to verify the rationality and practicality of the framework process designed by the present invention for the traffic congestion control method based on micro-traffic data decision-making during large-scale events. The operation process is organized as follows: Step 1) Chengdu City operating vehicle data set collection. The data collection time range is from April 24, 2022 to April 30, 2022 and from July 1, 2022 to July 7, 2022, and the data collection vehicle type is online car-hailing. Data collection is uploaded by the external positioning system uniformly installed by the company on the operating vehicles, and the data collection frequency is 30s; there are 49 collection fields in total. The present invention selects 6 GPS data field information including vehicle number, GPS collection time, latitude and longitude, speed, direction angle, and operating status for analysis.

[0095] Step 2): Preprocessing of GPS data of operating vehicles in Chengdu. The GPS data of operating vehicles in Chengdu was cleaned and deduplicated to obtain the research data, and descriptive statistical analysis was performed on the data collection time interval, spatial distribution and speed.

[0096] Step 3): GPS data trajectory division of Chengdu's operating vehicles. A segment of the vehicle trajectory data represents a segment of the vehicle's journey, so the trajectory division first needs to determine the basis for the journey division. If the GPS data of a vehicle has no position change or direction change for more than 2 minutes, it is considered that the vehicle is parked on the roadside or picking up or dropping off passengers.

[0097] Step 4): Grid division of Chengdu area. The grid division area in this example is selected within the Chengdu Ring Expressway, with a longitude range of [103.92918, 104.21106] and a latitude range of [30.526748, 30.796598]. The area within the Chengdu Ring Expressway is gridded, and the spatial step size L=200m is selected. The study area is divided into 150*135 grids. In this example, a 20*20 grid radiating outward from the Chengdu Sports Center is selected as an example to identify the traffic congestion inside the grid. The grid number Among the 400 grids in the study area, there are 6 invalid grids, accounting for 1.5%. Selecting the time step TS = 10 minutes, the time segment of a single day is A time slice.

[0098] Step 5): Generate GPS trajectory data during large-scale events at Chengdu Sports Center. The GPS data of operating vehicles in Chengdu are tracked. The selected study area in this example contains 3,978 independent vehicles and 14,057 vehicle trajectories. The existing historical data is oversampled and densified according to the ratio of the traffic volume of large-scale events to the existing historical data. The oversampled and densified data is mapped to the grid of the urban road network, and the traffic volume and traffic density of each grid within 30 seconds are calculated. The speed of the vehicle trajectory in the grid is adjusted through the speed-density linear relationship model proposed by Greenshields. Finally, the adjusted speed is matched to the oversampled and densified GPS data, and the travel time between two adjacent GPS points in a trajectory is calculated by the distance between the two adjacent GPS points and the vehicle speed. Finally, the GPS time of the vehicle is adjusted to obtain the trajectory data of the vehicle during the large-scale event.

[0099] Step 6): Identification of key congested sections of Chengdu Sports Center. Select a set of sample data to conduct case analysis on traffic congestion identification, divide the time slices and regional grids, build a traffic congestion identification model based on the average speed and number of trajectories within the grid, select a grid for calculation and analysis, and expand the cycle to the entire study area. Calculate the number of trajectories and average speed within the grid throughout the day, and calculate the congestion pattern of a single grid in 144 time slices in a day by time period to obtain the average grid congestion pattern for the whole day. , and determine whether the grid is congested in each time period.

[0100] Step 7): Evaluation of traffic congestion control methods based on traffic simulation. Investigate and collect traffic data of the urban road network in the study area, including the number of lanes, traffic lights, road types and other data, and count the road traffic volume. Simulate modeling of the study area to judge the feasibility of traffic congestion control methods. Shudu Avenue and Shuncheng Street are the main roads in Chengdu's road network. There are dense vehicles passing through, and the road capacity requirements are high. The intersection of Shudu Avenue Renmin East Road, Shudu Avenue Zongfu Road and Shuncheng Street in Chengdu was selected as the traffic simulation area. The evaluation steps of traffic congestion control methods are as follows: (1) The traffic volume at the intersection between 17:00 and 23:00 was calculated by using the GPS oversampling and densification trajectory data of existing operating vehicles. The direction of the traffic flow was determined by the vehicle head direction from the GPS data.

[0101] (2) Construct a VISSIM simulation model, build a road network at the intersection of Shudu Avenue and Shuncheng Street, set the intersection merging volume in units of 600 seconds, perform path planning for the traffic volume merging into the intersection, and set the ratio of vehicles going straight, turning left, and turning right at the intersection.

[0102] (3) Traffic signal timing method optimization. Step 6 identifies the traffic congestion of a single grid, analyzes the areas with severe congestion and the adjacent road networks, and finds that the cause of the congestion may be insufficient vehicle traffic capacity at the intersection. Therefore, this example analyzes the traffic flow at the intersection of Shudu Avenue and Shuncheng Street, and optimizes the existing traffic timing method to alleviate traffic congestion at the intersection. Traffic cameras and traffic control systems have been installed in this study area, so this example optimizes the original signal timing of the intersection to better control traffic flow. The traffic signal light timing at the Shudu-Shuncheng intersection is three-phase, which is the east-west straight-through release of Shudu Avenue, the opposite left-turn release, and the north-south straight-through and left-turn release of Shuncheng Street. The previous traffic signal release time for left turns in the east-west direction at the Shudu-Shuncheng intersection was short, but in the actual road network data, a large number of vehicles have left-turn operations at this intersection, so the left-turn road in the east-west direction will form congestion, and the vehicle queue phenomenon is more obvious. Left turns in the north-south direction are released at the same time as straight-through vehicles. If there is a queue of left-turners waiting at the intersection, when traffic is released, the oncoming straight-through vehicles need to pass first, which will increase the queue of left-turning vehicles and cause congestion. Therefore, the traffic signal timing of the intersection is changed to four-phase, and the time for releasing left turns in the east-west direction is increased.

[0103] (4) Evaluation of traffic signal timing method. The traffic timing method proposed in (3) was implemented in the VISSIM traffic simulation software. The average delay time, average queue length, average speed and traffic volume of vehicles on the road section before and after the timing improvement were compared to judge whether the traffic signal timing method was feasible.

[0104] Average delay: The average delay at the Shudu-Shuncheng intersection during large-scale events increased by 9.4% compared to the average delay under normal road network conditions. After optimizing the traffic signal timing at the intersection, the average delay at the intersection during large-scale events improved to a certain extent, with the average delay time decreasing by 6.3%.

[0105] Average traffic density: From the results of traffic simulation, the traffic density of the road section during large-scale events increased sharply compared with the traffic density of the road section under daily traffic conditions. By optimizing the traffic signal timing at the intersection through simulation, the traffic density of the road section decreased by 16.7%. Therefore, the traffic signal timing has a certain effect on alleviating traffic congestion.

[0106] Average speed: The simulation model of traffic volume during large-scale events has a low speed before and after the event. During the event, the traffic state rebounded. At the end of the event, the participants basically left at the same time, resulting in more vehicles gathering and congestion in the road network. From the traffic simulation results after the signal timing optimization of the intersection, it can be seen that the scheme has improved the average speed of vehicles in the east-west direction of Renmin North Road, Shudu Avenue, the west-east direction of Zongfu Road, Shudu Avenue, the north-south direction of Shuncheng Avenue, and the south-north direction of Shuncheng Avenue. It has little effect on the average speed of other sections. Therefore, from the average speed, this method can alleviate traffic congestion.

[0107] Average queue length: Simulation models were constructed for daily traffic volume, traffic volume during large-scale events, and traffic volume after optimizing traffic signal timing during large-scale events. The maximum vehicle queue lengths at the four entrances of the intersection were recorded at 10-minute intervals, and the maximum queue length was finally calculated. The queue length on Shudu Avenue is generally longer than that on Shuncheng Street. The queue length during large-scale events is much longer than the daily traffic flow state. After the traffic congestion control method was optimized, the queue length was improved to a certain extent, but the traffic flow could not reach the daily traffic flow state. This shows that the traffic signal timing method has a certain effect on alleviating traffic congestion.

[0108] The present invention provides a traffic congestion control method based on oversampled GPS data. First, the collected data is processed by data preprocessing, oversampling densification, speed adjustment and travel time scaling, etc., to generate vehicle flow trajectory data during large-scale events, which is used as the input of the grid internal traffic congestion identification model to identify the key sections of traffic congestion during large-scale events, and the traffic congestion control method of urban sections is optimized and designed by VISSIM micro-simulation modeling. The practical significance of the present invention lies in that the framework of the traffic congestion control method design constructed can be used for the optimization process of large-scale event traffic congestion control methods of different event scales and different event locations. It can prevent traffic congestion during large-scale events in advance, direct and guide congestion during events, save travel time for event participants, and alleviate the pressure on urban traffic road networks during large-scale events. In addition, relieving traffic congestion during large-scale events can also improve the happiness and comfort of event participants and reduce environmental pollution caused by exhaust emissions caused by long-term congestion. The present invention has certain guiding significance for the optimization of traffic congestion control methods during large-scale events.

[0109] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the invention.

Claims

1. A traffic congestion control method based on oversampled GPS data, characterized in that: The following steps are involved: S1: Collect historical urban traffic data during the activity; S2: Preprocess the collected urban traffic historical data, simulate the vehicle flow during the activity based on oversampling, and use the Greenshields speed-density linear model for data processing to obtain vehicle GPS trajectory data; S3: inputting vehicle GPS trajectory data into a grid traffic congestion identification model to identify grid traffic congestion; S4: Based on the identification results, the VISSIM simulation model is used to evaluate the indicators of the original different traffic congestion control methods, and the plan of the optimal traffic congestion control method is selected to realize traffic congestion control based on oversampled GPS data.

2. The traffic congestion control method based on oversampled GPS data according to claim 1, characterized in that: The urban traffic history data in S1 includes GPS data of operating vehicles, loop detector data and video data; The fields of the operating vehicle GPS data are: VehicleId, Time, Lon, Lat, Speed, Direction, Mileage and State; The VehicleId is the license plate number, which is used to uniquely identify the operating vehicle; The Time is the collection time of the GPS data of the operating vehicle; The Lon and Lat are the latitude and longitude coordinates of the operating vehicle's GPS data, which are used to determine the location of the operating vehicle; The Speed ​​is the speed of the operating vehicle, which is used to record the instantaneous speed of the operating vehicle at the time of uploading the GPS track; The Direction is the direction angle of the operating vehicle, which is used to record the direction of the vehicle head when the GPS track is uploaded; Mileage is the mileage of the operating vehicle, which is used to record the total mileage of the operating vehicle at the time of uploading the GPS track; The State is the operating state of the vehicle, including two values ​​0 and 1. The value 0 indicates an empty vehicle, and the value 1 indicates a passenger-carrying state.

3. The traffic congestion control method based on oversampled GPS data according to claim 2, characterized in that: The S2 includes the following sub-steps: S2-1: After cleaning and deduplication of the operating vehicle GPS data, construct an operating vehicle GPS trajectory data set T, and create a new set D for storing the newly added trajectory data; S2-2: Divide the road network in the study area into grids and determine whether the operating vehicle GPS trajectory dataset T is empty. If so, proceed to step S2-5; otherwise, proceed to step S2-3: S2-3: Select a track Traj from the operating vehicle GPS track data set T, delete the track Traj from the data set T, and perform oversampling and densification on the track Traj; S2-4: storing the oversampled and densified trajectory data and the original trajectory data in a set D, and returning to step S2-2; S2-5: Create a new set G, and divide the data in set D according to the grid division range, and store them in set G in units of grids; S2-6: Determine whether the set G is empty, if so, proceed to step S2-9, otherwise proceed to step S2-7; S2-7: Select a grid data set Grid_ij from the set G, and delete the data set Grid_ij from the set G; S2-8: Aggregate the trajectories in the dataset Grid_ij within a specific time and calculate the traffic density of the grid , and calculate the average vehicle speed of the grid in combination with the Greenshields speed-density linear model, and return to step S2-6; The traffic density of the grid The calculation formula is: in, is the number of vehicles on the road at a certain moment, is the length of the road; S2-9: The speed aggregated to a specific time is matched with the grid number to the newly added trajectory data, and the vehicle mileage is calculated according to the latitude and longitude coordinates of each trajectory segment. The time of the trajectory data is adjusted in combination with the vehicle's driving speed to obtain the vehicle GPS trajectory data.

4. The traffic congestion control method based on oversampled GPS data according to claim 3, characterized in that: The construction of the operating vehicle GPS trajectory dataset T in S2-1 includes the following steps: S2-1-1: Filter all GPS point data of the same VehicleId, recorded as set P; S2-1-2: Create new trajectory data and ; S2-1-3: Sort all the trajectory data in set P by Time, and determine whether set P is an empty set. If so, output the trajectory data as the operating vehicle GPS trajectory data set T, otherwise proceed to step S2-1-4; S2-1-4: Select the earliest trajectory point in set P , and the trajectory points Delete from the set P and determine the trajectory point Is the operating status equal to the trajectory point , if yes, go to step S2-1-5, otherwise go to step S2-1-6; S2-1-5: Determine the trajectory point Is the operating status of 1? If so, the trajectory point Adding track data , otherwise the trajectory point Adding track data , and return to step S2-1-3; S2-1-6: Set the track number in the newly created track data Increase by 1 and return to step S2-1-5.

5. The traffic congestion control method based on oversampled GPS data according to claim 3, characterized in that: In S2-3, the trajectory Traj is subjected to oversampling densification, including the following sub-steps: S2-3-1: Divide the trajectory Traj into GPS points, and insert a set number of GPS points equidistantly between two adjacent GPS points; S2-3-2: Calculate the longitude and latitude coordinates of the inserted GPS points, and generate a new track for each GPS point with the same serial number between two adjacent points; S2-3-3: Simulate the newly added trajectory as the trajectory of the newly added vehicle to achieve oversampling and densification of the trajectory Traj.

6. The traffic congestion control method based on oversampled GPS data according to claim 1, characterized in that: The grid traffic congestion identification model in S3 combines the number of trajectories in the grid area with the average vehicle speed inside the grid to form a two-dimensional vector Indicates the congestion pattern within the grid and the congestion pattern at different time periods and The difference is measured by the Mahalanobis distance, the formula is: in, is the Mahalanobis distance, is the covariance matrix of the whole sample, is the transpose of the matrix; The difference between the congestion pattern of a specific grid and the congestion pattern of the whole sample for the whole day is calculated as: in, For the Congestion mode, For the whole day The average of the congestion patterns in the time period, As the basis for judging the degree of congestion, The bigger it is, the more severe the congestion. This is the situation under free flow state; in, is the mean value under free flow state; Constraints are imposed on the average speed and number of trajectories of the vehicles. When the constraints are met, the grid is Internal congestion, the constraint formula is: in, For the time period The number of inner tracks, For the time period The average speed inside is the average number of trajectories of the grid over the entire day, is the average speed of the grid throughout the day.

7. The traffic congestion control method based on oversampled GPS data according to claim 6, characterized in that: In S3, grid traffic congestion is identified, and the degree of congestion is determined as follows: in, To index the congestion level; when When , the grid traffic congestion is determined to be unblocked; when , and is not satisfied and When , the grid traffic congestion is determined to be unblocked; when , and satisfies , not satisfied When , the grid traffic congestion is determined to be unblocked; when , and is not satisfied ,satisfy When , the grid traffic congestion is determined to be in a slow-moving state; when , and satisfies and When , the grid traffic congestion is determined to be in a general congestion state; when , and is not satisfied and When , the grid traffic congestion is determined to be unblocked; when , and satisfies , not satisfied When , the grid traffic congestion is determined to be unblocked; when , and is not satisfied ,satisfy When , the grid traffic congestion is determined to be in a general congestion state; when , and satisfies and When , the grid traffic congestion is judged to be in a serious state; when , and is not satisfied and When , the grid traffic congestion is determined to be unblocked; when , and satisfies , not satisfied When , the grid traffic congestion is determined to be unblocked; when , and is not satisfied ,satisfy When , the grid traffic congestion is determined to be in a severe congestion state; when , and satisfies and When , the grid traffic congestion is determined to be in a severe congestion state.

8. The traffic congestion control method based on oversampled GPS data according to claim 1, characterized in that: The S4 comprises the following sub-steps: S4-1: Collect relevant road data in the study area and obtain road network files; S4-2: Based on the road data and road network files, use VISSIM to simulate the road network in the study area and establish a VISSIM simulation model; S4-3: Based on the VISSIM simulation model, the indicators of the original different traffic congestion control methods are analyzed by adjusting the simulation parameters, and the optimal traffic congestion control method is selected to realize traffic congestion control based on oversampled GPS data.

9. The traffic congestion control method based on oversampled GPS data according to claim 8, characterized in that: The indicators include average delay , average speed , average queue length and traffic density .

10. The traffic congestion control method based on oversampled GPS data according to claim 9, characterized in that: The average delay , average speed , average queue length and traffic density The calculation formula is: in, For road section Traffic delays, For road section Traffic volume; in, is the distance traveled by the vehicle on the road section, For the The travel time of a vehicle on a road segment, For traffic volume; in, For road section The length of the queue of moving vehicles, is the number of vehicle queues on all road segments; in, is the number of vehicles on the road at a certain moment, For road section The traffic density, For road section Length, is the total length of the road network.

Images