CN111640304B - Automatic quantitative extraction method for traffic jam propagation characteristics of continuous flow traffic facility - Google Patents

Automatic quantitative extraction method for traffic jam propagation characteristics of continuous flow traffic facility Download PDF

Info

Publication number
CN111640304B
CN111640304B CN202010498564.8A CN202010498564A CN111640304B CN 111640304 B CN111640304 B CN 111640304B CN 202010498564 A CN202010498564 A CN 202010498564A CN 111640304 B CN111640304 B CN 111640304B
Authority
CN
China
Prior art keywords
space
time
traffic
congestion
flow
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202010498564.8A
Other languages
Chinese (zh)
Other versions
CN111640304A (en
Inventor
欧吉顺
聂庆慧
张韦华
安成川
梁程
邓社军
于世军
刘路
刘欣
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Yangzhou University
Original Assignee
Yangzhou University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Yangzhou University filed Critical Yangzhou University
Priority to CN202010498564.8A priority Critical patent/CN111640304B/en
Publication of CN111640304A publication Critical patent/CN111640304A/en
Application granted granted Critical
Publication of CN111640304B publication Critical patent/CN111640304B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G08G1/01—Detecting movement of traffic to be counted or controlled
    • G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
    • G08G1/0125—Traffic data processing
    • G08G1/0129—Traffic data processing for creating historical data or processing based on historical data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • G06V20/54—Surveillance or monitoring of activities, e.g. for recognising suspicious objects of traffic, e.g. cars on the road, trains or boats
    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G08G1/01—Detecting movement of traffic to be counted or controlled
    • G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
    • G08G1/0125—Traffic data processing
    • G08G1/0133—Traffic data processing for classifying traffic situation

Abstract

The invention relates to an automatic and quantitative extraction method for traffic jam propagation characteristics of continuous flow traffic facilities. The method uses the route as an analysis object, firstly constructs a traffic space-time contour map according to the collected traffic flow state parameters, and utilizes the collected traffic flow state parametersk-clustering historical traffic flow average speed data by means of a means algorithm and automatically determining thresholds for dividing congested and non-congested traffic states on the basis of the clustered boundary speeds; then, identifying independent (space-time discontinuous) space-time congestion areas in the traffic space-time contour map by using a graph connectivity clustering algorithm; and then, according to the space-time range corresponding to each independent space-time jam area, by combining with the traffic flow state parameters in the space-time range, various characteristic indexes related to traffic jam propagation are analyzed and extracted from the traffic space-time contour map, and the fast automatic identification and quantitative extraction of the traffic jam bottleneck and the traffic jam propagation characteristics of the continuous flow traffic facility are realized.

Description

Automatic quantitative extraction method for traffic jam propagation characteristics of continuous flow traffic facility
Technical Field
The invention relates to a method for analyzing and quantitatively extracting space-time propagation characteristics of traffic jam generated by a continuous flow traffic facility by using a computer automatic analysis technical means, in particular to an automatic quantitative extraction method of the traffic jam propagation characteristics for the continuous flow traffic facility, belonging to the field of intelligent monitoring and information management of road traffic.
Background
Traffic congestion has a series of negative effects on countries in the world, such as significant increase in economic loss and travel costs, resulting in severe exhaust and noise pollution. An important premise for relieving and treating traffic jam is that the traffic jam occurring on a road needs to be intelligently monitored, the bottleneck of the road traffic jam is quickly identified, and the traffic jam propagation characteristics are scientifically quantized, so that effective decision support is provided for reasonable traffic jam management.
In order to achieve the above objective, researchers have proposed a series of analysis and extraction methods for traffic congestion propagation characteristics (such as congestion bottleneck position, congestion bottleneck activation time, congestion propagation maximum time range, congestion propagation maximum space range, and vehicle delay caused by congestion). For continuous flow traffic facilities (highways, restricted access roads or urban expressways), the above methods can be divided into two main categories, namely methods based on local analysis and methods based on global analysis.
The method based on local analysis uses a road section as an analysis object (one road section is usually defined by using an upstream and downstream fixed detector as a boundary, and can also be defined in a manual mode, for example, a traffic route is divided into a plurality of road sections with the same length), judges whether the road section has traffic jam or not by extracting and analyzing the change curve of traffic flow state parameters collected by the upstream and downstream detectors, and analyzes and extracts various traffic jam characteristics. And then, analyzing and processing all road sections on the whole traffic facility in the same process, thereby extracting the traffic jam characteristics of all road sections. And finally, identifying the traffic jam bottleneck by integrating the congestion characteristics of all road sections, the congestion occurrence time sequence and the adjacency relation of the congested road sections, and extracting the traffic jam propagation characteristics. There are two main disadvantages to the method based on local analysis: one is that a traffic flow state parameter threshold method (such as a speed difference threshold method and an occupancy difference threshold method) is usually adopted when judging whether a certain road section has traffic jam. Since the determination of the optimal parameter threshold value depends on manual experience, rapid automatic processing cannot be realized for a large number of different continuous flow traffic facilities. In addition, the determination of the parameter threshold has subjectivity, and the reliability of the constructed model cannot be fully guaranteed; secondly, analysis and comparison of traffic states of every two road sections are involved in the modeling process, and when the number of the road sections is large, the model calculation complexity is easily overhigh. When the traffic facilities are large in scale or the traffic facilities need to be monitored and analyzed for multiple days, the working efficiency of the model is low, and the online automatic monitoring and management of road traffic are not facilitated.
The method based on the global analysis takes a route as an analysis object, a traffic space-time contour map facing to the whole route is constructed according to traffic flow state parameters (such as average speed, travel time, delay and the like) detected by each road section on the route, and traffic jam propagation characteristics are analyzed and extracted according to the traffic space-time contour map. Compared with a method based on local analysis, the method based on global analysis analyzes the spatial and temporal evolution condition of traffic jam on the route from a global angle by constructing a traffic space-time contour map, avoids comparison and analysis between every two road sections, can identify all traffic jam bottlenecks on the route through one-time analysis, and extracts traffic jam propagation characteristic indexes. Meanwhile, the method visually and visually presents the whole traffic jam condition of the route through the traffic space-time profile map, and is beneficial to visually and interactively monitoring and managing traffic by traffic researchers and engineering managers. However, the main disadvantages of the methods are that the manual work is required to identify the traffic jam bottleneck from the traffic space-time profile and extract the traffic jam propagation characteristics, the accuracy of identifying the traffic jam bottleneck and extracting the congestion propagation characteristic indexes is not high, the analysis efficiency is low, and the labor cost is increased. In addition, some traffic jam characteristic indexes such as vehicle delay cannot be directly extracted through manual visual observation, which limits the universality and the practicability of the method to a great extent.
Disclosure of Invention
Aiming at the problems, the invention provides an automatic quantitative extraction method for traffic jam propagation characteristics of continuous flow traffic facilities in order to overcome the main defects of the two methods. The method takes a route as an analysis object, firstly, a traffic space-time contour map is constructed according to collected traffic flow state parameters, a k-means algorithm is utilized to cluster historical traffic flow average speed data, and threshold values for dividing congestion and non-congestion traffic states are automatically determined according to the clustered boundary speed; then, identifying independent (space-time discontinuous) space-time congestion areas in the traffic space-time contour map by using a graph connectivity clustering algorithm; and then, according to the space-time range corresponding to each independent space-time jam area, by combining with the traffic flow state parameters in the space-time range, various characteristic indexes related to traffic jam propagation are analyzed and extracted from the traffic space-time contour map, and the fast automatic identification and quantitative extraction of the traffic jam bottleneck and the traffic jam propagation characteristics of the continuous flow traffic facility are realized.
The object of the invention is achieved in that,
a traffic jam propagation characteristic automatic quantitative extraction method for continuous flow traffic facilities is characterized by comprising the following steps:
step 1, constructing a traffic space-time contour map based on traffic flow state parameters;
(1-1) setting the maximum time range to be analyzed as M (the maximum value of a time axis of a traffic space-time contour map), the maximum space range as N (the maximum value of a space axis of the traffic space-time contour map) and the resolution of the traffic space-time contour map as N multiplied by M, wherein the direction of traffic flow in the traffic space-time contour map is defined as being from near to far in space;
(1-2) collecting time-series data of traffic flow average speed and traffic flow parameter on the continuous flow traffic facility according to M and N, respectively;
(1-3) constructing the average traffic speed under a given resolution n x m by using a space-time interpolation method by taking time as a horizontal axis and taking space as a vertical axis and combining the average traffic speed and a flow time sequence collected on the time and space coordinate positionsA space-time contour map of degree and a space-time contour map of traffic flow; for the traffic average speed space-time profile map, N × M data samples (data points) are corresponded in total, each sample corresponds to a space-time block, and the information which represents the average speed (assuming that the average speed is constant in the space-time range) of the traffic flow passing through the road space range with the length of M/M in the time range with the length of N/N is recorded as the information
Figure BDA0002523878780000031
Figure BDA0002523878780000032
Wherein, TiAnd SiRespectively, the temporal and spatial eigenvalues, V, of the ith sampleiIs the average velocity feature value of the ith sample; similarly, for the traffic flow space-time contour map, a total of N × M data samples (data points) are corresponded, each sample corresponds to a space-time block, and the traffic flow of the traffic flow in the road space range with the length of M/M in the time range with the length of N/N is represented and recorded as<Ti,Si,Qi>(ii) a Wherein, TiAnd SiTaking the same value as before, QiIs the flow characteristic value of the ith sample;
step 2, segmenting congested areas and non-congested areas in the traffic space-time contour map based on a k-means clustering algorithm;
(2-1) acquiring historical time series data of a traffic flow average speed parameter within a road space range defined by a traffic space-time profile diagram; here, in order to ensure the clustering quality, the length of the time range corresponding to the historical time-series data is set to at least one week;
(2-2) clustering the acquired historical time series data of the average speed parameter by using a k-means algorithm in an unsupervised clustering technology; the clustering number is set to be 2, and the corresponding traffic flow running state is the congestion state and the non-congestion state; after clustering is finished, 2 data clusters (clusters) are output, and the minimum value and the maximum value of the average speed in the 1 st data cluster are respectively set as V1_minAnd V1_maxLet the average velocity minimum and maximum values in the 2 nd data cluster be V respectively2_minAnd V2_maxThen, the average speed threshold for judging the traffic jam state is calculated as:
Figure BDA0002523878780000041
(2-3) according to the average speed threshold value judged by the traffic jam state, dividing the traffic (average speed and flow) space-time profile map into a space-time jam area and a space-time non-jam area, wherein the division rule is as follows:
Figure BDA0002523878780000042
in the above formula, CiRepresenting the traffic flow running state of the ith sample corresponding to the traffic space-time profile chart, C i1 represents that the traffic flow corresponding to the sample is in a congestion state in the corresponding space-time range, and otherwise, the traffic flow is in a non-congestion state; space-time congestion zone consisting ofiThe space-time block corresponding to the sample of 1 is formed, and the space-time non-congestion area is formed by CiThe spatio-temporal block corresponding to the sample of 0;
step 3, dividing the space-time congestion area into one or more independent (space-time discontinuous) space-time congestion areas based on a graph connectivity clustering algorithm;
(3-1) screening C from samples corresponding to the traffic average speed space-time profile mapiA sample of 1; that is, this step analyzes only the spatiotemporal congestion regions in the spatiotemporal profile; respectively extracting a T characteristic column and an S characteristic column from the screened sample;
(3-2) respectively carrying out normalization processing on the characteristic values in the T characteristic column and the S characteristic column, and calculating by the following formula:
Figure BDA0002523878780000043
Figure BDA0002523878780000044
Figure BDA0002523878780000051
Figure BDA0002523878780000052
Figure BDA0002523878780000053
Figure BDA0002523878780000054
in the above formula, T'jIs the normalized time characteristic value of the jth sample, wherein, j is more than or equal to 1 and less than or equal to alpha and less than or equal to nxm, and
Figure BDA0002523878780000055
alpha is CiNumber of samples,. mu.1TIs the mean value, σ, of the eigenvalues of the T-eigenvaluesSIs the standard deviation, S 'of the eigenvalue of the T characteristic line'jIs the normalized spatial feature value, μ, of the j-th sampleSIs the mean value, σ, of the eigenvalues of the S-signature columnSIs the standard deviation of the eigenvalues of the S-signature column;
(3-3) calculating the Euclidean distance between each sample corresponding to the space-time congestion area and the rest samples in the area in the time and space dimensions, thereby obtaining a distance matrix D; the distance between two samples is calculated by the following formula:
Figure BDA0002523878780000056
in the above formula, Da,bIs the value of the element, T ', corresponding to the a-th row and b-th column in the distance matrix D'aAnd T'bRespectively, a-th sum in the corresponding sample set of spatio-temporal congestion areasNormalized temporal feature value, S 'of the b-th sample'aAnd S'bNormalized space characteristic values of the a-th sample and the b-th sample in the sample set corresponding to the space-time congestion area are obtained; wherein a is more than or equal to 1 and less than or equal to n x m, b is more than or equal to 1 and less than or equal to alpha and less than or equal to n x m, and
Figure BDA0002523878780000057
(3-4) setting each space-time block in the space-time congestion area as a vertex, and setting edges according to the adjacency relation of the space-time blocks on the space-time dimension, so as to construct an undirected acyclic topological graph; the adjacency matrix a of the topological graph is calculated by the following formula:
Figure BDA0002523878780000058
Figure BDA0002523878780000059
dconn＝max(dT,dS)(12)
Figure BDA0002523878780000061
in the above formula, dTIs the shortest distance between every two space-time blocks in the time axis direction of the traffic space-time contour map, dSIs the shortest distance between every two space-time blocks in the space axis direction of the traffic space-time contour map, dconnWhether a connecting edge exists between every two space-time blocks is determined, Aa，bThe element value corresponding to the a-th row and the b-th column in the adjacent matrix A;
(3-5) searching all connected branches from the constructed topological graph by using a depth-first search algorithm in a graph theory, and classifying all vertexes in the same connected branch into a cluster; the number of the connected branches in the topological graph corresponds to the number of the categories in the clustering result; finally, the space-time congestion area can be further divided into one or more independent space-time congestion areas which are discontinuous in space-time dimension through connected branch search; each independent space-time congestion area is caused by different traffic congestion bottlenecks; by analyzing and quantifying each space-time congestion area, a plurality of independent traffic congestion bottlenecks existing in one route can be identified, and corresponding traffic congestion propagation characteristic indexes are quantitatively extracted;
step 4, fine adjustment is carried out on the independent space-time congestion area;
setting a to-be-fine-tuned space-time congestion area
Figure BDA0002523878780000062
BθRepresenting the theta-th space-time block in the space-time congestion area B to be finely adjusted, wherein theta represents the number of the space-time blocks in B, and the specific fine adjustment process comprises 2 links;
the first step is to finely adjust the active jam bottleneck position of the B:
(4-1-1) finding the spatial position S of the most downstream detector corresponding to Bdown：
Sdown＝min1≤δ≤ξ{Sδ≥Smax_B}(14)
In the above formula, Smax_BIs the position of the spatially farthest spatio-temporal block in B, SδIs spatially located at or farther than Smax_BIs a spatial position equal to or farther than Smax_BThe number of detectors of (a);
(4-1-2) search and SdownSpatial position S of nearest neighbor upstream detectorupThen B's active congestion bottleneck is defined as a binary group<Sup,Sdown>；
(4-1-3) search for the spatial position in B to be SupAnd SdownSet of space-time blocks in between, denoted as Bbott；
(4-1-4) from BbottFinding the earliest time space-time block in the space-time block
Figure BDA0002523878780000071
Time-latest spatio-temporal block
Figure BDA0002523878780000072
Calculating the duration of the active congestion bottleneck of B
Figure BDA0002523878780000073
Wherein
Figure BDA0002523878780000074
Is composed of
Figure BDA0002523878780000075
The time characteristic value of (a) is,
Figure BDA0002523878780000076
is composed of
Figure BDA0002523878780000077
A time characteristic value of (a);
(4-1-5) performing fine adjustment judgment on the active congestion bottleneck position of the B, wherein the rule is as follows:
Figure BDA0002523878780000078
in the above formula, IbottIs an indicator function, Ibott1 denotes the need for fine tuning of the active congestion bottleneck position of B, IbottWhen the value is 0, the active congestion bottleneck position of the B does not need to be finely adjusted, and epsilon is a duration threshold value;
(4-1-6) if the active congestion bottleneck position of B needs to be finely adjusted, B is adjustedbottThe space-time block in B is removed from B, and then the adjusted B is processed by the processes (1) to (6) until the space-time block meeting I is foundbottFinishing the fine adjustment process of the link as the rule B of 0; if the active congestion bottleneck position of the link B does not need to be finely adjusted, finishing the fine adjustment process of the link;
the second step is to finely adjust the space-time discontinuous region in B:
traversing the space-time blocks contained in the B from near to far (or from far to near) according to the space positions of the space-time blocks, namely scanning the space-time blocks in the B in sequence in a traffic space-time contour diagram; the space-time block set of each spatial position describes the traffic flow operation state of the traffic flow changing with time at the spatial position. For a set of spatio-temporal blocks at each spatial position:
(4-2-1) finding the spatio-temporal block in which the time is the earliest
Figure BDA0002523878780000079
Time-latest spatio-temporal block
Figure BDA00025238787800000710
For time eigenvalue between
Figure BDA00025238787800000711
And
Figure BDA00025238787800000712
checking all the space-time blocks in the traffic space-time contour map between the time characteristic values (the space-time blocks are not necessarily the space-time blocks contained in B), if the value of the space-time blocks is 1, continuing the processing, and if the value of the space-time blocks is 0, changing the traffic flow running state value of the space-time contour map to 1; the purpose of implementing the above operation is to ensure that the traffic flow at the same spatial position maintains time continuity during congestion;
(4-2-2) repeating the processing in (4-2-1) until the spatial positions of the spatio-temporal blocks in B are traversed; after processing, each independent space-time congestion area keeps continuous in space-time;
step 5, quantitatively extracting traffic jam propagation characteristic indexes;
setting a spatio-temporal congestion zone to be analyzed
Figure BDA0002523878780000081
Is a finely adjusted independent space-time congestion area, wherein B'μRepresents the mu-th space-time block in B ', psi represents the number of space-time blocks in B';
the steps are to quantitatively extract the traffic jam propagation characteristic index from the B'; the calculation process of each index is as follows:
Figure BDA0002523878780000082
Figure BDA0002523878780000083
Tspan(B′)＝Tclear(B′)-Tonset(B′)(18)
Figure BDA0002523878780000084
Figure BDA0002523878780000085
Sspan(B′)＝Send(B′)-Sstart(B′)(21)
Figure BDA0002523878780000086
Ω(B′)＝<Sup(B′),Sdown(B′)>(23)
in the above formula, Tonset(B ') is the start time of congestion propagation for the spatiotemporal congestion region B',
Figure BDA0002523878780000087
is the time characteristic value T of the mu-th space-time block in the space-time congestion area Bclear(B ') is the end time of the congestion propagation of the spatio-temporal congestion region B', Tspan(B ') is the duration of the congestion propagation of the spatio-temporal congestion region B', Sstart(B ') is a spatial start position of congestion propagation of the spatiotemporal congestion region B',
Figure BDA0002523878780000088
is the first in a space-time congestion area BSpatial eigenvalues of mu spatio-temporal blocks, Send(B ') is a spatial end position of the congestion propagation of the spatio-temporal congestion region B', Sspan(B ') is the maximum queue length of the congestion propagation of the spatio-temporal congestion region B', R (B ') is the total vehicle delay (in units of) caused by the congestion propagation of the spatio-temporal congestion region B',
Figure BDA0002523878780000091
is the flow characteristic value (unit is vehicle/5 min) of the mu-th space-time block in the space-time congestion area BintervIs the time interval for traffic flow data acquisition (typically taking 5 minutes),
Figure BDA0002523878780000092
is the average velocity eigenvalue (in kilometers per hour or miles per hour), V, of the μ -th spatio-temporal block in the spatio-temporal congestion area BfreeThe free flow speed of the traffic flow on the continuous flow traffic facility can be obtained by sorting the average speed data of the historical traffic flow from small to large and then calculating the 95 percent quantile of the data; omega (B ') is the active congestion bottleneck of the spatio-temporal congestion zone B', Sdown(B ') is the spatial position of the most downstream detector corresponding to the spatio-temporal congestion zone B', Sup(B') is a group with Sdown(B') the spatial position of the nearest upstream detector.
In the step (1-2), the average speed and flow parameter data of the traffic flow are acquired directly by a fixed traffic detection mode or acquired by indirect estimation through traffic state parameter data collected by a mobile traffic detection mode.
The fixed traffic detection device used in the fixed traffic detection mode is an induction coil or a microwave detector, and the fixed traffic detection device can directly acquire traffic flow and average speed of the traffic flow;
the mobile traffic detection mode is that the instantaneous speed and the positioning information of each vehicle are obtained through floating vehicles provided with GPS equipment or RFID chips, on the basis, the traffic flow rate is obtained through converging the number of vehicles of the traffic flow in a given space-time range, and the average speed of the traffic flow is obtained through averaging the instantaneous speed of the vehicles of the traffic flow in the given space-time range.
In the step (1-3), the space-time interpolation method is two-dimensional linear interpolation or two-dimensional nonlinear interpolation.
Compared with the prior art, the method has the advantages that the automatic analysis and quantitative extraction of the transmission characteristics of the traffic jam occurring on the continuous flow traffic facility in the space-time dimension are realized, the manual intervention process is omitted, the risk of unreliable analysis caused by manual subjective experience is reduced, and the automation level and the working efficiency of the traffic jam monitoring and management for the continuous flow traffic facility are obviously improved.
Drawings
FIG. 1 is a workflow of a method for automated analysis and extraction of traffic congestion propagation characteristics for continuous flow traffic facilities;
FIG. 2 is a workflow of a spatiotemporal congestion region segmentation method based on graph connectivity search clustering;
FIG. 3 is a continuous flow transportation facility analyzed in an embodiment of the present invention;
FIG. 4 is a traffic average speed space-time profile at 3/2015 and 12/12 days in an embodiment of the invention;
fig. 5 is a graph of the segmentation result of the empty congestion area at 3/2015 and 12/yen in the embodiment of the present invention;
FIG. 6 is a graph illustrating the fine adjustment result of the independent spatio-temporal congestion area at 3, 12 and 3 months in 2015 according to the embodiment of the invention;
fig. 7 is a diagram illustrating the fine adjustment result of 22 working day independent spatio-temporal congestion areas between 3/month and 2/year 2015 and 3/month and 27/year 2015 in the embodiment of the present invention;
fig. 8 is a visualization result diagram of total vehicle delays in 22 working day independent space-time congestion areas between 3/month and 2/year 2015 and 3/month and 27/year 2015 in the embodiment of the invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings.
The continuous flow transportation facility analyzed by the present example is a highway corridor in a city, as shown in fig. 3. The highway corridor is approximately 26 miles in length starting at point a and ending at point B. A total of 29 traffic detectors are arranged on the corridors, and traffic flow data passing through the detectors are collected at intervals of 5 minutes. The corridor contains a total of 30 road segments.
An analytical data set was created by aggregating traffic flow average speeds and traffic flow rates for 5 minute intervals of 22 weekdays (weekend and weekday disregarded) between 3 months 2 days 2015 and 3 months 27 days 2015 acquired by 29 detectors (circled 1-29 in fig. 3). And (2) constructing a traffic average speed space-time contour map and a traffic flow space-time contour map of each day by the construction method of the traffic space-time contour map in the step 1. The resolution of the constructed traffic space-time profile is set to 200 x 200. Fig. 4 shows a traffic average speed space-time profile at 3, 12 and 2015. As can be seen from the figure, there are 4 particularly significant independent spatio-temporal congestion areas (spatio-temporal congestion area 1, spatio-temporal congestion area 2, spatio-temporal congestion area 3, spatio-temporal congestion area 4) on the day. For each independent space-time congestion area, the congestion propagation characteristic indexes which can be extracted comprise an active congestion bottleneck (identified by a binary group of an upstream detector position and a downstream detector position at the bottleneck space position), congestion propagation starting time, congestion propagation ending time, congestion propagation duration, congestion propagation starting position, congestion propagation ending position, maximum congestion propagation queuing length and total vehicle delay (calculated by means of a traffic average speed space-time profile map and a traffic flow space-time profile map).
After the traffic space-time contour map is constructed, the congested areas and the non-congested areas in the traffic space-time contour map are segmented according to the step 2. Firstly, clustering is carried out on traffic flow average speed data collected by 29 detectors on a road section defined by a traffic space-time contour map by utilizing a k-means clustering algorithm. In the present example, the time range for clustering data is selected from 3/2/2015 to 3/2015/8/2015. And clustering and threshold calculation are carried out, so that the traffic jam state discrimination average speed threshold is 45.5 miles per hour. In other words, in the present example, the proposed method identifies areas in the traffic space-time profile having a traffic flow average speed value less than or equal to 45.5 mph as congested areas, and areas in the traffic space-time profile having a traffic flow average speed value greater than 45.5 mph as non-congested areas. Next, the congestion area is further divided into one or more independent spatio-temporal congestion areas by using the spatio-temporal congestion area division method based on graph connectivity clustering proposed in step 3. Fig. 5 shows 9 independent spatiotemporal congestion regions processed by the spatiotemporal congestion region segmentation step.
After the independent spatio-temporal congestion areas are obtained, each independent spatio-temporal congestion area is finely adjusted according to step 4, so that 4 independent spatio-temporal congestion areas as shown in fig. 6 are obtained. It can be seen that 5 of the 9 independent spatio-temporal congestion areas processed in step 3 are filtered out by the proposed method because their active congestion bottleneck duration is less than a given threshold (20 minutes in the present example). Meanwhile, the proposed method repairs the discontinuous areas of the 4 independent space-time congestion areas. Fig. 7 shows the independent spatiotemporal congestion region refinement results for 22 working days from 3/2/2015 to 27/3/2015.
After the fine-tuned independent space-time congestion areas are obtained, the traffic congestion propagation characteristics of each independent space-time congestion area are quantitatively extracted according to the step 5. Table 1 shows traffic congestion propagation characteristic indexes of independent spatio-temporal congestion areas quantitatively extracted from 3/2/2015 to 3/2015 and 6/2015. Fig. 8 gives a visualization of the total vehicle delays for 22 working days of independent spatio-temporal congestion areas quantitatively extracted from 2015 year 3, month 2 to 2015 year 3, month 27 (darker colors indicate greater total vehicle delays).
TABLE 1 traffic jam propagation characteristic index of independent space-time jam area quantitatively extracted
Figure BDA0002523878780000121

Claims (4)

1. A traffic jam propagation characteristic automatic quantitative extraction method for continuous flow traffic facilities is characterized by comprising the following steps:
step 1, constructing a traffic space-time contour map based on traffic flow state parameters;
(1-1) setting the maximum time range to be analyzed as M, the maximum space range as N and the resolution of a traffic space-time profile map as N multiplied by M, wherein the direction of traffic flow in the traffic space-time profile map is defined as from near to far in space; m corresponds to the maximum value of the time axis of the traffic space-time profile graph; n corresponds to the maximum value of the space axis of the traffic space-time contour map;
(1-2) collecting time-series data of traffic flow average speed and traffic flow parameter on the continuous flow traffic facility according to M and N, respectively;
(1-3) constructing a traffic average speed space-time contour map and a traffic flow space-time contour map under a given resolution n x m by using a space-time interpolation method by taking time as a horizontal axis and taking space as a vertical axis and combining a traffic flow average speed and flow time sequence collected on a time and space coordinate position; for the traffic average speed space-time profile map, the data samples are corresponding to N multiplied by M in total, each sample is corresponding to a space-time block, and the average speed information of the traffic flow passing through the road space range with the length of M/M in the time range with the length of N/N is expressed and recorded as<Ti,Si,Vi>(1≤i≤n×m,i∈Z+) (ii) a Wherein, TiAnd SiRespectively, the temporal and spatial eigenvalues, V, of the ith sampleiIs the average velocity feature value of the ith sample; similarly, for the traffic flow space-time contour map, a total of N × M data samples are corresponded, each sample corresponds to a space-time block, and the traffic flow of the traffic flow in the road space range with the length of M/M in the time range with the length of N/N is represented and recorded as<Ti,Si,Qi>(ii) a Wherein, TiAnd SiTaking the same value as before, QiIs the flow characteristic value of the ith sample;
step 2, segmenting congested areas and non-congested areas in the traffic space-time contour map based on a k-means clustering algorithm;
(2-1) acquiring historical time series data of a traffic flow average speed parameter within a road space range defined by a traffic space-time profile diagram; here, in order to ensure the clustering quality, the length of the time range corresponding to the historical time-series data is set to at least one week;
(2-2) clustering the acquired historical time series data of the average speed parameter by using a k-means algorithm in an unsupervised clustering technology; the clustering number is set to be 2, and the corresponding traffic flow running state is the congestion state and the non-congestion state; after clustering is finished, outputting 2 data clusters, and respectively setting the minimum value and the maximum value of the average speed in the 1 st data cluster as V1_minAnd V1_maxLet the average velocity minimum and maximum values in the 2 nd data cluster be V respectively2_minAnd V2_maxThen, the average speed threshold for judging the traffic jam state is calculated as:
Figure FDA0002774763020000021
(2-3) according to the average speed threshold value judged by the traffic jam state, dividing the traffic average speed space-time profile map and the traffic flow space-time profile map into a space-time jam area and a space-time non-jam area respectively, wherein the division rule is as follows:
Figure FDA0002774763020000022
in the above formula, CiRepresenting the traffic flow running state of the ith sample corresponding to the traffic space-time profile chart, Ci1 represents that the traffic flow corresponding to the sample is in a congestion state in the corresponding space-time range, and otherwise, the traffic flow is in a non-congestion state; space-time congestion zone consisting ofiThe space-time block corresponding to the sample of 1 is formed, and the space-time non-congestion area is formed by CiThe spatio-temporal block corresponding to the sample of 0;
step 3, dividing the space-time congestion area into one or more independent space-time congestion areas based on a graph connectivity clustering algorithm;
(3-1) screening C from samples corresponding to the traffic average speed space-time profile mapiA sample of 1; that is, this step analyzes only the spatiotemporal congestion regions in the spatiotemporal profile; respectively extracting from the screened samplesTaking a T characteristic column and an S characteristic column;
(3-2) respectively carrying out normalization processing on the characteristic values in the T characteristic column and the S characteristic column, and calculating by the following formula:
Figure FDA0002774763020000031
Figure FDA0002774763020000032
Figure FDA0002774763020000033
Figure FDA0002774763020000034
Figure FDA0002774763020000035
Figure FDA0002774763020000036
in the above formula, T'jIs the normalized time characteristic value of the jth sample, wherein j is more than or equal to 1 and less than or equal to alpha and less than or equal to nxm, and j, alpha belongs to Z+Alpha is CiNumber of samples,. mu.1TIs the mean value, σ, of the eigenvalues of the T-eigenvaluesSIs the standard deviation, S 'of the eigenvalue of the T characteristic line'jIs the normalized spatial feature value, μ, of the j-th sampleSIs the mean value, σ, of the eigenvalues of the S-signature columnSIs the standard deviation of the eigenvalues of the S-signature column;
(3-3) calculating the Euclidean distance between each sample corresponding to the space-time congestion area and the rest samples in the area in the time and space dimensions, thereby obtaining a distance matrix D; the distance between two samples is calculated by the following formula:
Figure FDA0002774763020000037
in the above formula, Da,bIs the value of the element, T ', corresponding to the a-th row and b-th column in the distance matrix D'aAnd T'bNormalized time characteristic values, S ', of the a-th and b-th samples in the sample set corresponding to the spatio-temporal congestion area respectively'aAnd S'bNormalized space characteristic values of the a-th sample and the b-th sample in the sample set corresponding to the space-time congestion area are obtained; wherein a is more than or equal to 1 and less than or equal to n multiplied by m, b is more than or equal to 1 and less than or equal to alpha and less than or equal to n multiplied by m, and a, b belongs to Z+；
(3-4) setting each space-time block in the space-time congestion area as a vertex, and setting edges according to the adjacency relation of the space-time blocks on the space-time dimension, so as to construct an undirected acyclic topological graph; the adjacency matrix a of the topological graph is calculated by the following formula:
Figure FDA0002774763020000041
Figure FDA0002774763020000042
dconn＝max(dT,dS) (12)
Figure FDA0002774763020000043
in the above formula, dTIs the shortest distance between every two space-time blocks in the time axis direction of the traffic space-time contour map, dSIs the shortest distance between every two space-time blocks in the space axis direction of the traffic space-time contour map, dconnIs to make a pairWhether the space-time block has a connected distance of connected edges, Aa,bThe element value corresponding to the a-th row and the b-th column in the adjacent matrix A;
(3-5) searching all connected branches from the constructed topological graph by using a depth-first search algorithm in a graph theory, and classifying all vertexes in the same connected branch into a cluster; the number of the connected branches in the topological graph corresponds to the number of the categories in the clustering result; finally, the space-time congestion area is further divided into one or more independent space-time congestion areas which are discontinuous in space-time dimension through connected branch search; each independent space-time congestion area is caused by different traffic congestion bottlenecks; by analyzing and quantifying each space-time congestion area, a plurality of independent traffic congestion bottlenecks existing in one route can be identified, and corresponding traffic congestion propagation characteristic indexes are quantitatively extracted;
step 4, fine adjustment is carried out on the independent space-time congestion area;
setting a to-be-fine-tuned space-time congestion area B as { B ═ Bθ/1≤θ≤Θ,θ∈Z+,Θ∈Z+}，BθRepresenting the theta-th space-time block in the space-time congestion area B to be finely adjusted, wherein theta represents the number of the space-time blocks in B, and the specific fine adjustment process comprises 2 links;
the first step is to finely adjust the active jam bottleneck position of the B:
(4-1-1) finding the spatial position S of the most downstream detector corresponding to Bdown：
Figure FDA0002774763020000044
In the above formula, Smax_BIs the position of the spatially farthest spatio-temporal block in B, SδIs spatially located at or farther than Smax_BIs a spatial position equal to or farther than Smax_BThe number of detectors of (a);
(4-1-2) search and SdownSpatial position S of nearest neighbor upstream detectorupThen B's active congestion bottleneck is defined as a binary group<Sup,Sdown>；
(4-1-3) search for the spatial position in B to be SupAnd SdownSet of space-time blocks in between, denoted as Bbott；
(4-1-4) from BbottFinding the earliest time space-time block in the space-time block
Figure FDA0002774763020000051
Time-latest spatio-temporal block
Figure FDA0002774763020000052
Calculating the duration of the active congestion bottleneck of B
Figure FDA0002774763020000053
Wherein
Figure FDA0002774763020000054
Is composed of
Figure FDA0002774763020000055
The time characteristic value of (a) is,
Figure FDA0002774763020000056
is composed of
Figure FDA0002774763020000057
A time characteristic value of (a);
(4-1-5) performing fine adjustment judgment on the active congestion bottleneck position of the B, wherein the rule is as follows:
Figure FDA0002774763020000058
in the above formula, IbottIs an indicator function, Ibott1 denotes the need for fine tuning of the active congestion bottleneck position of B, IbottWhen the value is 0, the active congestion bottleneck position of the B does not need to be finely adjusted, and epsilon is a duration threshold value;
(4-1-6) if the active jam bottleneck position of B needs to be finely adjusted, B is adjustedbottThe space-time blocks in B are removed from B, and then the adjusted B is processed by the processes from (4-1-1) to (4-1-6) until the space-time blocks meeting I are foundbottFinishing the fine adjustment process of the link as the rule B of 0; if the active congestion bottleneck position of the link B does not need to be finely adjusted, finishing the fine adjustment process of the link;
the second step is to finely adjust the space-time discontinuous region in B:
traversing the space-time blocks contained in the B from near to far or from far to near according to the space positions of the space-time blocks, namely scanning the space-time blocks in the B in sequence in a traffic space-time contour diagram; the space-time block set of each space position describes the traffic flow running state of the traffic flow changing with time at the space position; for a set of spatio-temporal blocks at each spatial position:
(4-2-1) finding the spatio-temporal block in which the time is the earliest
Figure FDA0002774763020000061
Time-latest spatio-temporal block
Figure FDA0002774763020000062
For time eigenvalue between
Figure FDA0002774763020000063
And
Figure FDA0002774763020000064
checking all space-time blocks in the traffic space-time contour map among the time characteristic values, if the value of the space-time blocks is 1, continuing processing, and if the value of the space-time blocks is 0, changing the traffic flow running state value of the space-time blocks to 1; the purpose of implementing fine adjustment operation on the space-time discontinuous region in the B is to ensure that the traffic flow at the same space position keeps time continuity in the congestion process;
(4-2-2) repeating the processing in (4-2-1) until the spatial positions of the spatio-temporal blocks in B are traversed; after processing, each independent space-time congestion area keeps continuous in space-time;
step 5, quantitatively extracting traffic jam propagation characteristic indexes;
setting a space-time congestion area B ' ═ B ' to be analyzed 'μ/1≤μ≤Ψ,μ∈Z+,Ψ∈Z+Is an independent space-time congestion area after fine adjustment, wherein B'μRepresents the mu-th space-time block in B ', psi represents the number of space-time blocks in B';
the steps are to quantitatively extract the traffic jam propagation characteristic index from the B'; the calculation process of each index is as follows:
Figure FDA0002774763020000065
Figure FDA0002774763020000066
Tspan(B′)＝Tclear(B′)-Tonset(B′) (18)
Figure FDA0002774763020000067
Figure FDA0002774763020000068
Sspan(B′)＝Send(B′)-Sstart(B′) (21)
Figure FDA0002774763020000069
Ω(B′)＝<Sup(B′),Sdown(B′)> (23)
in the above formula, Tonset(B') is the start of the congestion propagation of the spatio-temporal congestion region BIn the middle of the furnace, the gas-liquid separation chamber,
Figure FDA00027747630200000610
is the time characteristic value T of the mu-th space-time block in the space-time congestion area Bclear(B ') is the end time of the congestion propagation of the spatio-temporal congestion region B', Tspan(B ') is the duration of the congestion propagation of the spatio-temporal congestion region B', Sstart(B ') is a spatial start position of congestion propagation of the spatiotemporal congestion region B',
Figure FDA0002774763020000071
is the spatial characteristic value S of the mu-th space-time block in the space-time congestion area Bend(B ') is a spatial end position of the congestion propagation of the spatio-temporal congestion region B', Sspan(B ') is the maximum queue length of the congestion propagation of the spatio-temporal congestion region B', R (B ') is the total vehicle delay caused by the congestion propagation of the spatio-temporal congestion region B', and the unit is vehicle hour,
Figure FDA0002774763020000072
is the flow characteristic value of the mu time space block in the time-space congestion area B', the unit is vehicle/5 min, TintervIs the time interval for traffic flow data acquisition, typically taking 5 minutes,
Figure FDA0002774763020000073
is the average speed characteristic value of the mu-th space-time block in the space-time congestion area B', the unit is kilometer/hour or mile/hour, VfreeIs the free flow velocity of the traffic flow on the continuous flow traffic facility; omega (B ') is the active congestion bottleneck of the spatio-temporal congestion zone B', Sdown(B ') is the spatial position of the most downstream detector corresponding to the spatio-temporal congestion zone B', Sup(B') is a group with Sdown(B') the spatial position of the nearest upstream detector.
2. The method for automatically and quantitatively extracting the traffic jam propagation characteristics facing the continuous flow traffic facility as claimed in claim 1, wherein in the step (1-2), the average speed and the parameter data of the traffic flow are directly acquired by a fixed traffic detection mode or indirectly estimated by the traffic state parameter data collected by a mobile traffic detection mode.
3. The method for automatically and quantitatively extracting the traffic jam propagation characteristics facing the continuous flow traffic facility according to claim 2, wherein when the traffic flow average speed and the flow parameter data are directly acquired by a fixed traffic detection mode, the fixed traffic detection device used in the fixed traffic detection mode is an induction coil or a microwave detector, and the fixed traffic detection device can directly acquire the traffic flow and the traffic flow average speed; when the average speed and flow parameter data of the traffic flow are indirectly estimated and acquired through the traffic state parameter data collected by a mobile traffic detection mode, the mobile traffic detection mode acquires the instantaneous speed and positioning information of each vehicle through a floating vehicle provided with GPS equipment or an RFID chip, on the basis, the traffic flow is acquired by converging the number of vehicles of the traffic flow in a given space-time range, and the average speed of the traffic flow is acquired by averaging the instantaneous speeds of the vehicles of the traffic flow in the given space-time range.
4. The method for automatically and quantitatively extracting the traffic congestion propagation characteristic for the continuous flow traffic facility according to claim 1, wherein in the step (1-3), the space-time interpolation method is two-dimensional linear interpolation or two-dimensional nonlinear interpolation.
CN202010498564.8A 2020-06-04 2020-06-04 Automatic quantitative extraction method for traffic jam propagation characteristics of continuous flow traffic facility Active CN111640304B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010498564.8A CN111640304B (en) 2020-06-04 2020-06-04 Automatic quantitative extraction method for traffic jam propagation characteristics of continuous flow traffic facility

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010498564.8A CN111640304B (en) 2020-06-04 2020-06-04 Automatic quantitative extraction method for traffic jam propagation characteristics of continuous flow traffic facility

Publications (2)

Publication Number Publication Date
CN111640304A CN111640304A (en) 2020-09-08
CN111640304B true CN111640304B (en) 2021-01-08

Family

ID=72332506

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010498564.8A Active CN111640304B (en) 2020-06-04 2020-06-04 Automatic quantitative extraction method for traffic jam propagation characteristics of continuous flow traffic facility

Country Status (1)

Country Link
CN (1) CN111640304B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112233418B (en) * 2020-09-27 2021-09-03 东南大学 Secondary traffic accident prevention control method under intelligent network-connected mixed traffic flow environment
CN112233458B (en) * 2020-09-27 2021-09-07 东南大学 Secondary accident prevention oriented dynamic lane and variable speed limit cooperative control method
CN113362605B (en) * 2021-07-23 2022-06-21 上海交通大学 Distributed highway optimization system and method based on potential homogeneous area analysis
CN115527366B (en) * 2022-09-09 2023-07-25 扬州大学 Large-scale urban road network traffic jam bottleneck identification method

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106960571A (en) * 2017-03-30 2017-07-18 百度在线网络技术（北京）有限公司 Congestion in road bottleneck point determines method, device, server and storage medium
US10332391B2 (en) * 2016-12-06 2019-06-25 Here Global B.V. Split lane traffic jam detection and remediation
CN110176139A (en) * 2019-02-21 2019-08-27 淮阴工学院 A kind of congestion in road identification method for visualizing based on DBSCAN+
CN111008223A (en) * 2019-10-21 2020-04-14 北京交通大学 Regional traffic jam correlation calculation method based on space-time association rule
CN111159247A (en) * 2019-11-20 2020-05-15 北京交通大学 Regional traffic jam propagation track mining method

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8543320B2 (en) * 2011-05-19 2013-09-24 Microsoft Corporation Inferring a behavioral state of a vehicle

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10332391B2 (en) * 2016-12-06 2019-06-25 Here Global B.V. Split lane traffic jam detection and remediation
CN106960571A (en) * 2017-03-30 2017-07-18 百度在线网络技术（北京）有限公司 Congestion in road bottleneck point determines method, device, server and storage medium
CN110176139A (en) * 2019-02-21 2019-08-27 淮阴工学院 A kind of congestion in road identification method for visualizing based on DBSCAN+
CN111008223A (en) * 2019-10-21 2020-04-14 北京交通大学 Regional traffic jam correlation calculation method based on space-time association rule
CN111159247A (en) * 2019-11-20 2020-05-15 北京交通大学 Regional traffic jam propagation track mining method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
Systematic clustering method to identify and characterise spatiotemporal congestion on freeway corridors;Jishun Ou,etc.;《IET Intelligent Transport Systems》;20181031;第12卷(第8期);第826-837页 *
信息化条件下交通拥堵瓶颈识别与扩散判别;叶晓飞等;《武汉理工大学学报（交通科学与工程版）》;20141031;第38卷(第5期);第1060-1064页 *

Also Published As

Publication number Publication date
CN111640304A (en) 2020-09-08

Similar Documents

Publication Publication Date Title
CN111640304B (en) Automatic quantitative extraction method for traffic jam propagation characteristics of continuous flow traffic facility
CN110176139B (en) DBSCAN + -based road congestion identification visualization method
CN111539454B (en) Vehicle track clustering method and system based on meta-learning
CN111210612B (en) Method for extracting bus route track based on bus GPS data and station information
CN108665703B (en) Road network state transition point discrimination method based on macroscopic basic graph
CN116644373B (en) Automobile flow data analysis management system based on artificial intelligence
CN105374209B (en) A kind of urban area road network running status characteristics information extraction method
CN101807345A (en) Traffic jam judging method based on video detection technology
CN108922174B (en) Dynamic classification method for paths in group of intersections around expressway entrance ramp
CN112512020B (en) Traffic state weak signal perception studying and judging method based on multi-source data fusion
CN114049765B (en) Urban road network traffic flow OD estimation method based on automatic vehicle number plate identification data
CN114783183A (en) Monitoring method and system based on traffic situation algorithm
CN109164450B (en) Downburst prediction method based on Doppler radar data
CN115691120A (en) Congestion identification method and system based on highway running water data
CN111768619A (en) Express way vehicle OD point determining method based on checkpoint data
CN111062648A (en) Method for evaluating comprehensive performance of asphalt pavement
Luo et al. Grouped travel time estimation in signalized arterials using point-to-point detectors
Biswas et al. Assessment of level-of-service on urban arterials: a case study in Kolkata metropolis
CN112767684A (en) Highway traffic jam detection method based on charging data
WO2017107790A1 (en) Method and apparatus for predicting road conditions using big data
CN107066501B (en) motor vehicle exhaust remote measuring equipment distribution method based on road similarity
CN109147320B (en) Road section traffic state discrimination method based on checkpoint data
CN116168356B (en) Vehicle damage judging method based on computer vision
CN106446194B (en) Method and device for determining acquisition parameters of geographic area
CN113516850B (en) Pipeline traffic flow data acquisition method based on space syntactic analysis

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant