CN112508225A - Multi-detail traffic cell partitioning method and system based on spectral clustering algorithm - Google Patents

Multi-detail traffic cell partitioning method and system based on spectral clustering algorithm Download PDF

Info

Publication number
CN112508225A
CN112508225A CN202011161561.1A CN202011161561A CN112508225A CN 112508225 A CN112508225 A CN 112508225A CN 202011161561 A CN202011161561 A CN 202011161561A CN 112508225 A CN112508225 A CN 112508225A
Authority
CN
China
Prior art keywords
matrix
traffic
road
similarity
degree
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.)
Pending
Application number
CN202011161561.1A
Other languages
Chinese (zh)
Inventor
胡正华
高良煜
何松翰
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.)
Ningbo University of Technology
Original Assignee
Ningbo University of Technology
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 Ningbo University of Technology filed Critical Ningbo University of Technology
Priority to CN202011161561.1A priority Critical patent/CN112508225A/en
Publication of CN112508225A publication Critical patent/CN112508225A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G06Q10/043—Optimisation of two dimensional placement, e.g. cutting of clothes or wood
    • 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
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/26—Government or public services

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • Economics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Development Economics (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Probability & Statistics with Applications (AREA)
  • Primary Health Care (AREA)
  • Artificial Intelligence (AREA)
  • Educational Administration (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention discloses a multi-detail traffic zone dividing method and a system based on a spectral clustering algorithm, wherein the method comprises the following steps: s1: constructing an undirected weighted graph of a target area road network; s2: respectively calculating a road network similarity matrix contributed by each vehicle travel track; s3: summing the road network similarity matrixes contributed by all the vehicle travel tracks to obtain a similarity matrix; s4: calculating the degree of each road section in the undirected weighted graph and constructing a degree matrix; s5: constructing a Laplace matrix based on the degree matrix and the similar matrix, calculating an eigenvalue of the Laplace matrix, and taking an eigenvector corresponding to the eigenvalue to form an eigenvector matrix; s6: and (5) clustering by using the characteristic matrix as a sample by using a K-means method. The method classifies the road sections with higher relevance in daily travel of residents into one class, and the division is more scientific and reasonable; in addition, the invention fully considers the hierarchy of traffic investigation regions, can divide the influence degree of different regions on the road network, and can solve the problem of too coarse or too fine division.

Description

Multi-detail traffic cell partitioning method and system based on spectral clustering algorithm
Technical Field
The invention belongs to the technical field of intelligent traffic, and particularly relates to a multi-detail traffic zone dividing method and system based on a spectral clustering algorithm.
Background
With the rapid development of economic construction in China and the continuous improvement of the living standard of people, the number of private cars in cities is increased day by day, the existing road traffic infrastructure cannot meet the requirements of people for going out, and a series of traffic problems such as traffic jam, traffic order disorder, tail gas and noise pollution are generated. In order to alleviate the current situation of urban road congestion, a plurality of government departments carry out large-scale traffic investigation and research and planning work of urban traffic with huge resources, and the purpose is to better understand the current traffic demand condition of the city, master the occurrence rule and development trend of traffic in an urban traffic system, establish a traffic information database and provide an effective basis for implementing traffic demand prediction. Meanwhile, with the aging and perfection of the GPS positioning technology and the wireless communication technology, the possibility of acquiring the traffic state information of the urban road network in real time is realized, so that a more comprehensive and reliable technical means is provided for the survey of urban motor vehicle traveling, and meanwhile, the simulation and analysis of the running state of large-scale urban motor vehicles by researchers are facilitated.
The concept of traffic cells is derived from traffic investigations, and a traditional traffic cell refers to a set of nodes or road segments which are related in traffic characteristics and have high traffic similarity. The method is closely associated with information such as time parameters, contact degrees, similarity indexes and the like, and reflects the time-space change condition of the urban traffic road network. The traffic community is a community group which has similar structure connotations and strong relevance and keeps relative balance in a certain range, and is the reflection of factors such as population, economy and the like in urban development. In recent years, the theory and method for dividing traffic districts are gradually improved along with the development of traffic transportation, and are widely applied in the field of traffic planning, so that the theory and method become the basis for predicting traffic demand. The basic units for analyzing traffic characteristics formed by the division are not necessarily the actual boundaries, but are artificially defined spatial ranges for the convenience of studying and analyzing the laws behind the travel phenomenon. Through dividing similar regions, the regions with similar urban road environments are linked, urban traffic concentrated conditions are visually represented, unified management and control are carried out, and therefore management cost is reduced. Analysis of traffic flow between traffic cells also reflects the degree of activity in cities and the degree of economic development in particular areas. Through analyzing the traffic flow among all the cells, the population flow condition is mastered, and therefore traffic flow resources are reasonably distributed.
The size of the divided area of the traffic cell has certain influence on the practical application, and the excessively small divided area causes the workload of investigation, analysis and prediction to be increased and the fund to be wasted; if the area division is too large, the survey data can not meet the precision requirement of the traffic planning on the current situation survey, and the scientificity and the rationality of the planning are influenced. The suitable traffic zone division scheme can effectively grasp the current situation of regional traffic, reasonably predict traffic demands, scientifically compile traffic development plans, reduce the complexity of a traffic control and management system and improve the reliability of the system.
The conventional method related to traffic cell division is mainly to divide according to administrative regions, homogeneity, subarea population and other division principles, and a large number of scholars have conducted intensive research on the division of traffic cells at present, but the traditional method for dividing traffic cells restricts the number, size and functional properties of the divided traffic cells. Therefore, it is difficult to accurately grasp the potential connection among the traffic cells, and it is difficult to balance the accuracy and workload.
Disclosure of Invention
The invention aims to introduce the similarity between road sections refracted by vehicle travel tracks, and provides a multi-detail traffic cell division method and system based on a spectral clustering algorithm.
The invention provides a multi-detail traffic zone dividing method based on a spectral clustering algorithm, which comprises the following steps:
s1: each road section X in the road section data set X of the target areaiRegarding as a point in space, construct undirected weighted graph G (X, W) of road networkN×N) The similarity matrix WN×NMiddle element WijThe similarity between the road sections i and j is represented, wherein i is 1,2, and N, j is 1,2, and N is the number of the road sections in the road section data set;
s2: after the travel tracks of passing vehicles in the target area are obtained, the contribution of the travel tracks of the vehicles is calculated respectivelyRoad network similarity matrix
Figure BDA0002744403540000021
M1, 2, M representing the total number of vehicles,
Figure BDA0002744403540000022
middle element
Figure BDA0002744403540000023
Represents the travel track T of the mth vehiclemSimilarity between the contributed road segments i, j;
Figure BDA0002744403540000024
is calculated as: when the travel track TmIf the road sections i and j are included in the middle, then order
Figure BDA0002744403540000025
Otherwise
Figure BDA0002744403540000026
S3: summing the road network similarity matrixes contributed by all the vehicle travel tracks to obtain a similarity matrix;
s4: calculating each point x in undirected weighted graphiDegree of (1)
Figure BDA0002744403540000027
Will diConstructing a degree matrix as diagonal elements;
s5: constructing a Laplace matrix based on the degree matrix and the similar matrix, calculating the eigenvalue of the Laplace matrix, sequencing the eigenvalues from small to large, and taking the top k1The eigenvectors corresponding to the eigenvalues form an eigenvector matrix, k1Is a preset value;
s6: and (5) clustering by using the characteristic matrix as a sample by using a K-means method.
Further, the vehicle travel track is obtained through license plate identification data collected by a gate system arranged on a highway lane, GPS track data of the vehicle or mobile phone signaling data.
Further, the method comprisesIn step S5, k1The value is preset according to the requirement; or after sorting the eigenvalues, the difference between two adjacent eigenvalues suddenly increases, and if the two adjacent eigenvalues are respectively the pth eigenvalue and the pth +1 eigenvalue, let k be1＝p。
Further, the method also comprises a multi-level dividing step, comprising the following steps:
s701: one or more traffic areas which do not meet the precision requirement in the upper-level clustering result are taken; the initial value of the previous-level clustering result is the first-level clustering result, namely the clustering result obtained in claim 1;
s702: taking the traffic areas which do not meet the accuracy requirement as new target areas respectively, and implementing the steps S1-S6 in claim 1;
s703: when the clustering results all meet the precision requirement, ending the division; otherwise, executing step S702 on one or more traffic areas that still do not meet the accuracy requirement in the current clustering result.
The invention provides a multi-detail traffic small area division system based on a spectral clustering algorithm, which comprises:
a first module for comparing each road segment X in a road segment data set X of a target areaiRegarding as a point in space, construct undirected weighted graph G (X, W) of road networkN×N) The similarity matrix WN×NMiddle element WijThe similarity between the road sections i and j is represented, wherein i is 1,2, and N, j is 1,2, and N is the number of the road sections in the road section data set;
a second module, configured to, after obtaining travel tracks of passing vehicles in the target area, respectively calculate a road network similarity matrix contributed by the travel tracks of the passing vehicles
Figure BDA0002744403540000031
M1, 2, M representing the total number of vehicles,
Figure BDA0002744403540000032
middle element
Figure BDA0002744403540000033
To representTravel track T of mth vehiclemSimilarity between the contributed road segments i, j;
Figure BDA0002744403540000034
is calculated as: when the travel track TmIf the road sections i and j are included in the middle, then order
Figure BDA0002744403540000035
Otherwise
Figure BDA0002744403540000036
The third module is used for summing the road network similarity matrixes contributed by all the vehicle travel tracks to obtain a similarity matrix;
a fourth module for calculating each point x in the undirected weighted graphiDegree of (1)
Figure BDA0002744403540000037
Will diConstructing a degree matrix as diagonal elements;
a fifth module for constructing a Laplace matrix based on the degree matrix and the similarity matrix, calculating eigenvalues of the Laplace matrix, sorting the eigenvalues from small to large, and taking the top k1The eigenvectors corresponding to the eigenvalues form an eigenvector matrix, k1Is a preset value;
and the sixth module is used for clustering by using the characteristic matrix as a sample and using a K-means method.
Compared with the prior art, the invention has the following advantages and beneficial effects:
(1) the method is characterized in that the relevance among road sections is described based on the travel track of the vehicle, and a similarity matrix among the road sections is established, so that traffic districts are divided; the road sections with high relevance in daily travel of residents are classified into one category, and the two problems that the traditional traffic zone division needs artificial combination and the deviation degree with the actual travel behavior of the residents is high are solved, so that the division process is more scientific and reasonable.
(2) The invention also fully considers the hierarchy of traffic investigation regions, can divide traffic cells according to the influence degree of different regions on the road network, and can solve the problem that the traffic cells are too thick or too thin.
(3) Experimental results prove that the method is practical and effective, has a good practical effect in the process of dividing the urban road network traffic cells, and has a good application prospect in analyzing regional traffic flow characteristics.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 shows a first division of a traffic cell in an embodiment;
fig. 3 shows the traffic cell secondary division result in the embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The technical principle and the corresponding specific implementation mode of the multi-detail traffic cell division method based on the spectral clustering algorithm are as follows:
first, first level traffic district division
Using undirected weighted graph G (X, W)N×N) Describing road network data set, X (X)1,x2,…,xN) Representing an original road segment data set in a target area, where xi(i ═ 1, 2.., N) denotes the ith link; every road section X in XiConstructing an undirected weighted graph as a point in space; wN×NSimilarity matrix, W, representing the correlation between road sectionsN×NSize N × N, where the ith row and jth column are denoted as Wij，i＝1,2,...,N，j＝1,2,...,N，WijRepresents the similarity between the links i, j, and Wij＝WjiWhen i is j, Wij＝0。
According to the invention, the similarity between road sections is obtained according to the travel track of the vehicle, so that the similarity matrix W is determinedN×N. There are many well-established technologies available to capture the travel trajectory of a vehicle, for exampleThe vehicle license plate identification data can be acquired through license plate identification data collected by a gate system arranged on a highway lane, and GPS track data or mobile phone signaling data of a vehicle can also be utilized.
Obtaining travel tracks of all vehicles passing through a target area in a preset time period, and respectively recording the travel tracks as T1，T2，…,TM，TmThe travel track of the mth vehicle is shown, wherein M is 1, 2. T ism＝{r1,r2,…rS}，rsDenotes the S-th route segment, S1, 2.
The present invention defines the similarity between road sections based on the travel track of the vehicle, which will be described below by way of example.
Suppose a travel trajectory T of the 1 st vehicle1＝{r1,r2,r3,r4Based on travel track T1Then the road section r is considered1And road section r2、r3、r4Are all correlated, then M12＝1，M13＝1，M 141 is ═ 1; similarly, road section r2And road section r3、r4Are also relevant, then M23＝1，M 241 is ═ 1; similarly, road section r3And road section r4Is also related, then M 341. Wherein M is12、M13、M14、M23、M24、M34Respectively represent travel tracks T1The contributed road section r1And r2Similarity between, road section r1And r3Similarity between, road section r1And r4Similarity between, road section r2And r3Similarity between, road section r2And r4Similarity between, road section r3And r4The similarity between them. Then obtain the travel track T1Middle road section r1、r2、r3、r4Similarity matrix between
Figure BDA0002744403540000051
Without loss of generality, assumeFrom the track T1Contributed road network similarity matrix
Figure BDA0002744403540000052
The size is N × N. Wherein the elements of the ith row and the jth column
Figure BDA0002744403540000053
Represents travel track T1Similarity between the contributed road sections i and j, and travel track T1Including sections i and j at the same time, then the travel track T1The degree of similarity contributed
Figure BDA0002744403540000054
Is 1; otherwise it is 0. In the matrix W1In (1),
Figure BDA0002744403540000055
and when i ═ j
Figure BDA0002744403540000056
Road network similarity matrix contributed by travel tracks of M vehicles is respectively obtained based on the method
Figure BDA0002744403540000057
For all similarity matrix
Figure BDA0002744403540000058
Summing to obtain a similar matrix W corresponding to the original road section data set XN×N：
Figure BDA0002744403540000059
Obtaining a similarity matrix WN×NThen, for any point xiDefine the point xiDegree d ofiIs equal to xiSum of weights of all edges connected, i.e.
Figure BDA00027444035400000510
Will diAs a degree matrix DThe diagonal elements of (a) construct a degree matrix D:
Figure BDA00027444035400000511
defining Laplace matrix L ═ D-WN×NNormalizing the Laplace matrix to obtain a matrix L1＝D-1/2LD-1/2. Calculating the matrix L1The eigenvalues are sorted from small to large to obtain an eigenvalue set lambda, and k is taken as the top1The characteristic values form a new characteristic value set
Figure BDA00027444035400000512
Figure BDA00027444035400000513
Denotes the tth eigenvalue in the new eigenvalue set λ', t ═ 1,21，
Figure BDA00027444035400000514
Corresponding to an independent feature vector
Figure BDA00027444035400000515
k1The value is preset according to the user's requirements, generally k1The value is large, the clustering precision is high, but the calculation speed is low. In addition, when the difference between two adjacent eigenvalues suddenly increases in the eigenvalue set λ, if the two adjacent eigenvalues are respectively the pth eigenvalue and the pth +1 eigenvalue, then k can be made1＝p。
In order to make the subsequent clustering effect more excellent, each feature vector is subjected to
Figure BDA00027444035400000516
Normalization processing is carried out to obtain a feature vector
Figure BDA00027444035400000517
Then, a feature vector set can be established
Figure BDA00027444035400000518
K in feature vector set H1A column vector
Figure BDA00027444035400000519
Composition feature matrix
Figure BDA00027444035400000520
Each column in the characteristic matrix corresponds to one characteristic, and each row in the matrix corresponds to each road section in the road section data set; by feature matrices
Figure BDA00027444035400000521
And clustering the samples by using a K-means method to obtain a road network division scheme of a first level.
The first hierarchical network is divided as follows:
(1) constructing an undirected weighted graph G (X, W) of a road network data set, and determining a similarity matrix W based on the travel track of the vehicle;
(2) constructing a degree matrix D according to the similarity matrix W;
(3) calculating the Laplace matrix L ═ D-W, and normalizing the Laplace matrix to obtain the matrix L1＝D-1/2LD-1/2；
(4) Calculating the matrix L1Minimum k1Feature vectors corresponding to the respective feature values
Figure BDA0002744403540000061
And normalizing the feature vectors
Figure BDA0002744403540000062
Obtaining a feature vector
Figure BDA0002744403540000063
By feature vectors
Figure BDA0002744403540000064
Form Nxk1A feature matrix U of the dimension;
(5) taking each row in the feature matrix U as a k1Samples of vitamin D, in K-m respectivelyClustering by using an eans method, wherein the clustering dimension is k2Obtaining a partition of clusters
Figure BDA0002744403540000065
One cluster then constitutes one traffic cell, where k2I.e., the value of K in the K-means method.
Traffic district division based on multiple detail levels
The division basis and the division purpose of the traffic cells are different, and the granularity of the cell division is also different. The method is used for dividing the traffic cells of a macroscopic traffic planning model, and has larger granularity of the division of the cells and less number of the cells; the traffic cell division granularity for the microscopic traffic simulation model is small, and the number of cells is large. The increase of the number of the traffic cell divisions will increase the workload of calculation, and the decrease of the number will affect the precision.
The traditional traffic cell division method has insufficient knowledge on the regional hierarchy of traffic investigation, cannot flexibly divide cells according to the travel characteristics of different regions, and for some regions with developed economy, large traffic demand and outstanding traffic contradiction, if the division of the traffic cells is too coarse, the accuracy of the traffic investigation result is insufficient, unnecessary deviation is generated, and further the real traffic condition cannot be objectively reflected, and the accuracy of road network planning and traffic volume prediction work of the region is influenced.
In order to fully consider the hierarchy of traffic investigation regions in traffic investigation, divide traffic cells according to the influence degree of different regions on a road network and solve the problem of too coarse division of the traffic cells, a multi-stage-partition traffic cell division method is adopted to divide the road network in the investigation region into the hierarchies and divide the traffic cells from different space division granularities.
In order to solve the problems, the invention divides the traffic cells needing to be divided into multiple detail levels into a second level on the basis of obtaining the traffic cell division of the first level. The dividing thought is divided with a first level, and specifically comprises the following steps: taking a cluster (namely a traffic cell) which needs to be subjected to multi-detail hierarchical division as a new target area, determining a similar matrix based on a travel track of passing vehicles in the new target area, and then clustering by utilizing a spectral clustering algorithm; and further dividing the clusters obtained by the first division into a plurality of sub-traffic areas.
It should be noted that the multi-detail level division is not limited to two divisions, and the road network region in the cluster obtained by the previous level division can be recursively divided according to actual requirements until the required division precision is met.
Examples
The technical solution and the technical effect of the present invention will be further described with reference to the application examples.
In the embodiment, a Ningbo city is taken as an example, and license plate identification data is collected through a bayonet system, so that a travel track of a vehicle in a preset time period is extracted. And establishing the correlation among the road sections according to the travel track, and further dividing the Ningbo city road network into traffic districts. Areas with developed economy and prominent traffic conflict in Ningbo city mainly include Haeosin, Jiangbei and \37150, and state areas. For these regions, if the traffic cells are divided by using only one-time clustering, the division is obviously too coarse, so that the regions need to be further clustered to form traffic cells with smaller granularity, and the number of the small regions is increased, so that the fineness of the division is improved, and the working cost of traffic control is increased.
(1) First-level partitioning and result analysis of traffic cells
The travel track of the vehicle indirectly reflects the correlation among the road sections in the urban road network at a certain level, and the correlation among each road section contained in the same travel track is correspondingly increased. Along with the continuous travel of the vehicles, the strength of the correlation between the road sections can be gradually reflected. The process of dividing the traffic districts is a process of dividing the urban road network once according to the relevance among road sections shown by the travel track of the vehicles. By means of the division, the relevance between the links in the same small area is as high as possible, and the relevance of the links which are not in the same area is as low as possible. And after the correlation among the road sections in the road network is obtained, combining a spectral clustering algorithm to obtain a primary division scheme. The result of one division in this example is shown in fig. 2.
It can be seen from the figure that the method of the present invention divides 3 traffic cells, classifies the urban road network region where the road section with high correlation is located into one traffic control (planning) unit, and classifies the road section with relatively weak correlation into different units. The first-level division of the traffic cell is a one-time division of an urban road network on a macroscopic level. The formed traffic district is large in space range, is a macro division of a main traffic generation place and an attraction place in the same city, and is also a main road network unit for generating traffic behaviors.
(2) Multi-level partitioning and result analysis of traffic cells
For the traffic cell after the first-level division, if the spatial range of the area in which the traffic cell is located is still larger or a more obvious mixed traffic flow still exists, the traffic cell can be further divided on the basis of the previous-level division. The division of the traffic cells in multiple detail levels is a process of reestablishing the association degree between the road sections in the area according to the track of the traveling vehicles in the area on the basis of the division of the previous level, and then further dividing the traffic cells obtained in the previous level by using the method. Referring to fig. 3, the result of the secondary division of the traffic cell 3 in fig. 2 is shown.
From the small traffic cell range obtained after the secondary division, the obtained traffic cell range is much smaller than that of the traffic cell range of the previous level, because as the division granularity becomes thinner, the landform (river, mountain, etc.) becomes more and more prominent in the process of dividing the small area. Thus, the distribution of cells as a whole is relatively loose.
In the process of traffic management and planning, the division of traffic cells is always performed around the purpose of OD survey work and the traffic travel characteristics of the area. The OD survey is a key component of a traffic planning and traffic management system, and the accuracy of the prediction of the traffic occurrence and attraction amount of each traffic cell directly influences the accuracy of the later stage and even the whole traffic prediction process, so that the traffic cell division is accurately carried out, and the OD survey is very important for traffic planning work.
Those skilled in the art will appreciate that, in the embodiments of the methods of the present invention, the sequence numbers of the steps are not used to limit the sequence of the steps, and it is within the scope of the present invention for those skilled in the art to change the sequence of the steps without inventive work. The examples described herein are intended to aid the reader in understanding the practice of the invention and it is to be understood that the scope of the invention is not limited to such specific statements and examples. Those skilled in the art can make various other specific changes and combinations based on the teachings of the present invention without departing from the spirit of the invention, and these changes and combinations are within the scope of the invention.

Claims (5)

1. The multi-detail traffic zone dividing method based on the spectral clustering algorithm is characterized by comprising the following steps:
s1: each road section X in the road section data set X of the target areaiRegarding as a point in space, construct undirected weighted graph G (X, W) of road networkN×N) The similarity matrix WN×NMiddle element WijThe similarity between the road sections i and j is represented, wherein i is 1,2, and N, j is 1,2, and N is the number of the road sections in the road section data set;
s2: after the travel tracks of passing vehicles in the target area are obtained, the road network similarity matrix contributed by the travel tracks of the vehicles is respectively calculated
Figure FDA0002744403530000011
M represents the total number of vehicles,
Figure FDA0002744403530000012
middle element
Figure FDA0002744403530000013
Represents the travel track T of the mth vehiclemSimilarity between the contributed road segments i, j;
Figure FDA0002744403530000014
is calculated as: when the travel track TmIf the road sections i and j are included in the middle, then order
Figure FDA0002744403530000015
Otherwise
Figure FDA0002744403530000016
S3: summing the road network similarity matrixes contributed by all the vehicle travel tracks to obtain a similarity matrix;
s4: calculating each point x in undirected weighted graphiDegree of (1)
Figure FDA0002744403530000017
Will diConstructing a degree matrix as diagonal elements;
s5: constructing a Laplace matrix based on the degree matrix and the similar matrix, calculating the eigenvalue of the Laplace matrix, sequencing the eigenvalues from small to large, and taking the top k1The eigenvectors corresponding to the eigenvalues form an eigenvector matrix, k1Is a preset value;
s6: and (5) clustering by using the characteristic matrix as a sample by using a K-means method.
2. The method for multi-detail traffic cell segmentation based on spectral clustering algorithm as claimed in claim 1, characterized by:
the vehicle travel track is obtained through license plate identification data collected by a gate system arranged on a highway lane, GPS track data of the vehicle or mobile phone signaling data.
3. The method for multi-detail traffic cell segmentation based on spectral clustering algorithm as claimed in claim 1, characterized by:
in step S5, k1The value is preset according to the requirement; or after sorting the eigenvalues, the difference between two adjacent eigenvalues suddenly increases, and if the two adjacent eigenvalues are respectively the pth eigenvalue and the pth +1 eigenvalue, let k be1＝p。
4. The method for multi-detail traffic cell segmentation based on spectral clustering algorithm as claimed in claim 1, characterized by:
further comprising a multi-level partitioning step comprising:
s701: one or more traffic areas which do not meet the precision requirement in the upper-level clustering result are taken; the initial value of the previous-level clustering result is the first-level clustering result, namely the clustering result obtained in claim 1;
s702: taking the traffic areas which do not meet the accuracy requirement as new target areas respectively, and implementing the steps S1-S6 in claim 1;
s703: when the clustering results all meet the precision requirement, ending the division; otherwise, executing step S702 on one or more traffic areas that still do not meet the accuracy requirement in the current clustering result.
5. The multi-detail traffic small region sub-system based on the spectral clustering algorithm is characterized by comprising the following steps:
a first module for comparing each road segment X in a road segment data set X of a target areaiRegarding as a point in space, construct undirected weighted graph G (X, W) of road networkN×N) The similarity matrix WN×NMiddle element WijThe similarity between the road sections i and j is represented, wherein i is 1,2, and N, j is 1,2, and N is the number of the road sections in the road section data set;
a second module, configured to, after obtaining travel tracks of passing vehicles in the target area, respectively calculate a road network similarity matrix contributed by the travel tracks of the passing vehicles
Figure FDA0002744403530000021
M represents the total number of vehicles,
Figure FDA0002744403530000022
middle element
Figure FDA0002744403530000023
Represents the travel track T of the mth vehiclemContributed to byThe similarity between the road sections i and j;
Figure FDA0002744403530000024
is calculated as: when the travel track TmIf the road sections i and j are included in the middle, then order
Figure FDA0002744403530000025
Otherwise
Figure FDA0002744403530000026
The third module is used for summing the road network similarity matrixes contributed by all the vehicle travel tracks to obtain a similarity matrix;
a fourth module for calculating each point x in the undirected weighted graphiDegree of (1)
Figure FDA0002744403530000027
Will diConstructing a degree matrix as diagonal elements;
a fifth module for constructing a Laplace matrix based on the degree matrix and the similarity matrix, calculating eigenvalues of the Laplace matrix, sorting the eigenvalues from small to large, and taking the top k1The eigenvectors corresponding to the eigenvalues form an eigenvector matrix, k1Is a preset value;
and the sixth module is used for clustering by using the characteristic matrix as a sample and using a K-means method.
CN202011161561.1A 2020-10-27 2020-10-27 Multi-detail traffic cell partitioning method and system based on spectral clustering algorithm Pending CN112508225A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011161561.1A CN112508225A (en) 2020-10-27 2020-10-27 Multi-detail traffic cell partitioning method and system based on spectral clustering algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011161561.1A CN112508225A (en) 2020-10-27 2020-10-27 Multi-detail traffic cell partitioning method and system based on spectral clustering algorithm

Publications (1)

Publication Number Publication Date
CN112508225A true CN112508225A (en) 2021-03-16

Family

ID=74955944

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011161561.1A Pending CN112508225A (en) 2020-10-27 2020-10-27 Multi-detail traffic cell partitioning method and system based on spectral clustering algorithm

Country Status (1)

Country Link
CN (1) CN112508225A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114548884A (en) * 2022-04-27 2022-05-27 中国科学院微电子研究所 Package identification method and system based on pruning lightweight model

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102855638A (en) * 2012-08-13 2013-01-02 苏州大学 Detection method for abnormal behavior of vehicle based on spectrum clustering
CN109816984A (en) * 2019-03-19 2019-05-28 太原理工大学 A kind of traffic network region division and dynamic adjusting method
CN109887297A (en) * 2019-04-23 2019-06-14 太原理工大学 The division methods of urban traffic control sub-district based on quick global K-means spectral clustering
CN110415523A (en) * 2019-08-13 2019-11-05 东南大学 A kind of signal control work zone division methods based on vehicle driving track data
CN111598335A (en) * 2020-05-15 2020-08-28 长春理工大学 Traffic area division method based on improved spectral clustering algorithm

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102855638A (en) * 2012-08-13 2013-01-02 苏州大学 Detection method for abnormal behavior of vehicle based on spectrum clustering
CN109816984A (en) * 2019-03-19 2019-05-28 太原理工大学 A kind of traffic network region division and dynamic adjusting method
CN109887297A (en) * 2019-04-23 2019-06-14 太原理工大学 The division methods of urban traffic control sub-district based on quick global K-means spectral clustering
CN110415523A (en) * 2019-08-13 2019-11-05 东南大学 A kind of signal control work zone division methods based on vehicle driving track data
CN111598335A (en) * 2020-05-15 2020-08-28 长春理工大学 Traffic area division method based on improved spectral clustering algorithm

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
尹卓;许甜;陈阳舟;卢佳程;: "基于集成聚类的交叉口车辆行驶路径提取方法", 计算机应用与软件, no. 07, 12 July 2020 (2020-07-12), pages 186 - 193 *
尹洪英;徐丽群;曹永荣;: "基于谱聚类算法的城市路网动态分区研究", 交通信息与安全, no. 01, 15 February 2010 (2010-02-15), pages 16 - 19 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114548884A (en) * 2022-04-27 2022-05-27 中国科学院微电子研究所 Package identification method and system based on pruning lightweight model
CN114548884B (en) * 2022-04-27 2022-07-12 中国科学院微电子研究所 Package identification method and system based on pruning lightweight model

Similar Documents

Publication Publication Date Title
Zhao et al. Truck traffic speed prediction under non-recurrent congestion: Based on optimized deep learning algorithms and GPS data
Wang et al. Truck traffic flow prediction based on LSTM and GRU methods with sampled GPS data
CN110836675B (en) Decision tree-based automatic driving search decision method
Meng et al. A two-stage short-term traffic flow prediction method based on AVL and AKNN techniques
CN113159364A (en) Passenger flow prediction method and system for large-scale traffic station
CN108961758A (en) A kind of crossing broadening lane detection method promoting decision tree based on gradient
CN109345434B (en) Method for evaluating design safety of external roads in open type community
CN115409256B (en) Route recommendation method for avoiding congestion area based on travel time prediction
Wei et al. An improved road network partition algorithm for parallel microscopic traffic simulation
Ma et al. Identifying spatiotemporal traffic patterns in large-scale urban road networks using a modified nonnegative matrix factorization algorithm
Xie et al. A spatiotemporal apriori approach to capture dynamic associations of regional traffic congestion
Shen et al. An attention-based digraph convolution network enabled framework for congestion recognition in three-dimensional road networks
CN115269758A (en) Passenger-guidance-oriented road network passenger flow state deduction method and system
Zhu et al. Calibrating travel time thresholds with cluster analysis and AFC data for passenger reasonable route generation on an urban rail transit network
CN115131966A (en) Road network key node identification method considering intersection operation characteristics
CN113553350A (en) Traffic flow partition model for similar evolution mode clustering and dynamic time zone partitioning
CN112508225A (en) Multi-detail traffic cell partitioning method and system based on spectral clustering algorithm
CN111738527B (en) Urban traffic cell division method based on hot spot detection model
Jiang et al. Partitioning of urban networks with polycentric congestion pattern for traffic management policies: Identifying protected networks
CN108764518B (en) Traffic resource dynamic optimization method based on big data of Internet of things
CN114463978B (en) Data monitoring method based on track traffic information processing terminal
Yu et al. GPS data mining at signalized intersections for congestion charging
Dong et al. An identification model of urban critical links with macroscopic fundamental diagram theory
Li et al. MGC-GAN: Multi-graph convolutional generative adversarial networks for accurate citywide traffic flow prediction
Xia et al. Deeprailway: a deep learning system for forecasting railway traffic

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