CN110609824A - Hot spot area detection method based on dynamic space network model under urban road network environment - Google Patents
Hot spot area detection method based on dynamic space network model under urban road network environment Download PDFInfo
- Publication number
- CN110609824A CN110609824A CN201910857903.4A CN201910857903A CN110609824A CN 110609824 A CN110609824 A CN 110609824A CN 201910857903 A CN201910857903 A CN 201910857903A CN 110609824 A CN110609824 A CN 110609824A
- Authority
- CN
- China
- Prior art keywords
- candidate
- edge
- flow
- hot spot
- time
- 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.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 17
- 238000000034 method Methods 0.000 claims abstract description 90
- 238000001914 filtration Methods 0.000 claims description 24
- 230000006870 function Effects 0.000 claims description 19
- 230000008569 process Effects 0.000 claims description 18
- 230000008859 change Effects 0.000 claims description 14
- 230000007704 transition Effects 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000004458 analytical method Methods 0.000 claims description 4
- 238000009412 basement excavation Methods 0.000 claims description 4
- 238000013075 data extraction Methods 0.000 claims description 3
- 230000002093 peripheral effect Effects 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000013077 scoring method Methods 0.000 claims description 3
- 238000005065 mining Methods 0.000 abstract description 23
- 230000002776 aggregation Effects 0.000 description 11
- 238000004220 aggregation Methods 0.000 description 11
- 230000000694 effects Effects 0.000 description 11
- 238000007493 shaping process Methods 0.000 description 11
- 238000011160 research Methods 0.000 description 9
- 238000002474 experimental method Methods 0.000 description 8
- 238000007667 floating Methods 0.000 description 7
- 230000033001 locomotion Effects 0.000 description 7
- 238000012360 testing method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000012554 master batch record Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 238000000926 separation method Methods 0.000 description 3
- 238000012418 validation experiment Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000007418 data mining Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000002688 persistence Effects 0.000 description 2
- 238000012358 sourcing Methods 0.000 description 2
- 230000029305 taxis Effects 0.000 description 2
- 230000001186 cumulative effect Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000006806 disease prevention Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000013439 planning Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/211—Schema design and management
- G06F16/212—Schema design and management with details for data modelling support
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2474—Sequence data queries, e.g. querying versioned data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Remote Sensing (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Probability & Statistics with Applications (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a hot spot area detection method based on a dynamic space network model under an urban road network environment, which comprises the following steps: constructing a dynamic space network DSN, designing a data structure of the dynamic space network DSN, and mining a hot spot region by adopting a partitioning-filtering-detecting method. The invention adopts the track data as the source of the mobile data, and adopts the map data of the city as the main data source of the dynamic space network; aiming at the problem of mining the hot spot areas in the urban geographic space, the hot spot area mining algorithm based on the dynamic space network is provided, and the hot spot areas in the urban space in the specified time interval can be accurately and efficiently found.
Description
Technical Field
The invention relates to a time-space data mining technology, in particular to a hot spot region detection method based on a dynamic space network model under an urban road network environment.
Background
With the continuous development of global positioning technology, wireless communication technology, and the wide spread of mobile terminal devices, the acquisition of large-scale mobile data has become possible. According to different data sources, common mobile data comprise mobile phone terminal positioning and communication records, bus card swiping records, GPS track data, social network sign-in data and the like. Such data records information in terms of space, time, direction, speed, POI (Point of interest), etc. This provides unprecedented opportunities and challenges for urban population mobility research. By analyzing and mining mobile data, powerful method guidance and decision support can be provided for application neighborhoods of traffic management, disease prevention and control, city planning, travel recommendation and the like of modern cities.
Urban population mobility research is a hot problem in spatio-temporal data mining. The research task is to find the movement patterns and the activity rules of human groups, such as commuting activities, tourist meetings, traffic jams and the like of urban groups. The studies are divided into population-oriented studies and geospatially-oriented studies, depending on the type of study.
Typical population-oriented mobile modes include Swarm mode (dock), Convoy mode (Convoy), Swarm mode (Swarm), and aggregation mode (collecting). Such research analyzes and mines the regularity of group movement from the viewpoint of the motion form, and can find a moving object group with the same motion form. However, the mining method does not involve geospatial information, so that the movement rules among regions in the geospatial space cannot be found.
The second category of research, which remedies this deficiency, studies human mobility from a geospatial perspective. For example: the method comprises the steps of researching the mobility rule of human groups in a geographic space by utilizing a mobile interaction mode among subway station research areas gathered in space, researching a mobile interaction mode among functional areas in a city by adopting OD (origin) data of taxis, a black hole mode based on a space network model and the like, and defining the research work as a subgraph discovery problem in the space network. The research work of the black hole mode is a symbolic result of urban population mobility research, but has limitations in the aspect of modeling the evolution of a crowd event, namely, the distribution change of a mobile population in a geographic space along with time cannot be reflected due to the fact that no time information is contained, and limitations exist in the aspect of researching the motion evolution trend of a mobile object population in the geographic space.
Take the "Shanghai pedaling event" as an example. The crowd in the road network near the beach forms large-scale congestion in the square area through a section of gradual convergence process. The existing black hole mode can find the blocks of crowd aggregation in a road network through a sub-graph discovery method of a spatial network, but cannot model the continuous change state of a moving crowd in a geographic space, so that the aggregation trend of the crowd cannot be tracked and discovered, and even the potential tread risk possibly caused can be predicted.
Therefore, modeling and mining of the aggregation events pay more attention to the time evolution characteristics of the aggregation events while paying attention to the population scale and the spatial regionality of the aggregation events, so that the time evolution characteristics of the aggregation events can effectively reflect the distribution situation of the mobile object population in the spatial network along with the time change, and the method can help reveal the movement rule of the population in the geographic space.
Disclosure of Invention
The purpose of the invention is as follows: the invention aims to solve the defects in the prior art, and provides a hot spot region detection method based on a dynamic spatial network model under an urban road network environment, which can find a hot spot region in an urban geographic space from mobile data, wherein the hot spot region has the performances of time continuity, spatial regionality, traffic scale and the like.
The technical scheme is as follows: the invention discloses a hot spot region detection method based on a dynamic space network model under an urban road network environment, which sequentially comprises the following steps of:
(1) constructing a Dynamic Space Network (DSN);
(1.1) constructing space and topology information of a Dynamic Space Network (DSN), namely extracting corners and sections of intersections and road directions according to map data to establish space and topology information of the DSN, namely acquiring nodes and edges in the dynamic space network;
(1.2) calculating the flow attribute of the edge, namely establishing the association between the track and the road by adopting a road network matching method, and then calculating the net flow sequence { w) of each edge in the DSNi1,wi2,wi3,…};
(2) Designing a data structure of a Dynamic Space Network (DSN):
first, all edges { e ] are stored using a linear structure1,e2,e3… }, each edge eiThe elements are respectively connected with a net flow sequence { wi1,wi2,wi3… and two nodes vi,vj}; thus, the traffic attribute of an accessible edge given an edge, and associated node information; then each node viConnecting an edge set ei1,ei2,ei3…, such that given an edge, other edge information connected to it is available;
(3) adopting a partitioning-filtering-detecting method to carry out the excavation of the hot spot area, namely: firstly, dividing a dynamic space network DSN in space to obtain a candidate area set CA; then, filtering candidate areas in different time periods by adopting estimation functions F and H of an upper flow bound; and finally, detecting hot spot areas aiming at the candidate areas to obtain a hot spot area set HS.
Further, in the step (1.1), road network data is used as input data of the dynamic space network DSN, that is, intersections (intersections of roads) and road inflection points (position points where the direction of the roads changes) of the roads in the road network are mapped to nodes in the space network graph, and road sections in the road network are mapped to edges; the output data of the dynamic space network DSN, i.e. the spatial and topological information in the DSN, is obtained by the above-mentioned input data extraction.
Further, the step (1.2) adopts an HMM-based algorithm to perform road network matching, and the specific process is as follows:
combining the time sequence characteristics and the spatial attributes (the time sequence relation and the longitude and latitude attributes) of the data points in the track, simulating observation probability transition by using Gaussian distribution, wherein the state transition probability is calculated by integrating the speed information, the GPS observation points and the real points, and searching a group of candidate road sections and candidate points for each GPS sampling point according to the road section projection process; then constructing a candidate graph, wherein nodes in the candidate graph are candidate point sets observed by each GPS, and edges are shortest path sets between any two adjacent candidate points; the nodes and the edges are distributed with corresponding weight values based on the space/time analysis result; and finally, combining the observation and the transition probability to search a path with the highest score in the candidate graph. And the global matching probability is improved to the maximum extent.
Further, the detailed process of dividing the candidate region set of the meeting in step (3) is as follows:
(A) and (3) carrying out space division on the dynamic space network G: and (3) setting the spatial threshold of the hot spot region as D, and adopting a square grid with the side length of D to perform spatial division on G to obtain a D multiplied by D grid. Taking fig. 2 as an example, which shows a 4 × 4 grid of G, the squares in the grid are numbered, and the square with row number a and column number b is marked as GabA is more than or equal to 1, b is less than or equal to D, the grid gabThe set of included edges is denoted as gab.E;
(B) Generating a candidate region set according to the space division result: the candidate area CA of the dynamic spatial network G is represented by a square GabI.e. all grid neighborhoods constitute CA ═ AR (g)ab)|1≤a,b≤D};
Because the minimum circumscribed rectangle MBR of the hot spot area is smaller than or equal to a square with a side length of d in size, the hot spot area is necessarily contained by a square neighborhood, and the square neighborhood is composed of peripheral squares surrounding the square.
Further, the specific process of filtering the candidate region in the step (3) is as follows:
(A) modeling the change process of the neighborhood by adopting a time sequence form;
the time span of the neighborhood AR (g) coincides with the time sequence T of the dynamic spatial network, the time period of this neighborhood being denoted Ti,j={ti,ti+1,…,tj1 ≦ i, j ≦ k, the time period is represented in the form of a matrix as follows:
wherein, the time threshold of the hot spot region is θ, so the time period satisfying the threshold requirement should be { T }i,j|1≤i≤j≤k,j-i≥θ}。
(B) Adopting a flow estimation filtering method based on dynamic programming to filter candidate areas for the first time, namely: carrying out flow summarizing calculation on all edges with positive net flow in the candidate region in the time period by adopting an estimation function F of an upper flow bound, wherein if a hot spot region exists in the candidate region, the net flow is less than or equal to the upper flow bound formed by all edges with positive flow in the candidate region no matter whether the hot spot region contains the edges with negative flow;
candidate area AR (g) for a specified time period Ti,jThe upper bound of the inner flow is:
whereinA side indicating that the flow rate is a positive value in the time interval t; f (AR), (g), Ti,j) Simplified to F (i, j); given time period Ti,jInner, net flow f of any subgraph S edge seta(S.E,Ti,j) Equal to the sum of the net flows of all edgesMaximum net flow based on the principle of estimation of the upper bound of flow
(C) Performing secondary filtering on the candidate result after the primary filtering by adopting an estimation function H;
the upper flow bound estimate H function is:simplified to H (i, j), whereinIs shown during the time period Ti,jThe inner net flow sum value is a positive edge;
in the candidate area AR (g), each edge e is in the time period Ti,jSum of sigma with a net flow thereine∈S.Efa(e,Ti,j) And the sum of all positive values in the candidate region in the time is
Wherein the sum of all positive sum values is smaller than the calculation result of the estimation function F, i.e.: is shown at Ti,jThe total value of the internal net flow is a positive edge, and then secondary filtration is realized.
Further, in the step (3), the current time period Ti,jThe method for searching the hot spot area of the candidate area ar (g) of (1) is: first calculate the number of edges in the square g at Ti,jNet flow summary value within; then selecting the edge with the maximum net flow sum value in the g to form an initial subgraph S; then, expanding the subgraph in an iterative mode;
in each iteration, selecting adjacent edges of the current subgraph S as a candidate edge set, scoring each edge e in the candidate edge set, then adding the edge with the highest score into the subgraph S, if the edge is a-infinity edge or an edge with MBR (S + e). DL exceeding a spatial threshold value cannot be expanded, and ending the iteration until no candidate edge can be expanded;
if the sub-graph S at this time satisfies: s.fa/|Ti,jIf | ≧ δ, outputting the result, and deleting the edge related to S from the candidate data set to prevent repeated detection; finally, recalculating the upper flow bound of the remaining edges in the candidate area, if the upper flow bound meets the threshold requirement, adding the upper flow bound into the candidate area set again for re-detection, and otherwise, deleting the upper flow bound;
repeatedly searching the hot spot areas according to the method until the candidate area set is empty;
wherein the scoring method is shown in formula 1:
where Δ is a change value of a diagonal line of the subgraph S after the edge e is added, that is, Δ ═ MBR (S + e). DL-MBR (S) · DL, where "S + e" indicates a new subgraph after e is added.
Further, in the first filtering method, a dynamic programming method is adopted to directly calculate the upper bound of the flow to be estimated through the estimated upper bound of the flow, and then the time complexity is reduced to O (mk)2);
Namely: f (i, j) ═ F (i-1, j) + F (i, j-1) -F (i-1, j-1).
Has the advantages that: the method defines the hot spot region from the aspects of time persistence, regional locality and flow scale, completely describes the forming and evolution process of the hot spot region, particularly the flow cumulative change of the hot spot region, and ensures the completeness and accuracy of the final mining result. In order to efficiently mine hot spot areas from mass mobile data, the invention designs and realizes a hot spot area detection algorithm based on dynamic programming, and adopts a problem solving framework of division-filtration-detection: firstly, dividing a dynamic space network on a space to obtain a candidate region set; then, two estimation functions F and H of an upper flow bound are designed, and the advantages of the two functions are combined to efficiently filter candidate areas in different time periods; and finally, detecting hot spot regions aiming at the candidate regions. By adopting the method, the calculation efficiency of the hot spot region mining algorithm can be effectively improved.
Drawings
FIG. 1 is a data structure diagram of a dynamic space network according to the present invention;
FIG. 2 is a graph showing the inclusion relationship between hot spot areas and grid neighborhoods in the present invention;
FIG. 3 is a time series diagram of the neighborhood AR (g) in the present invention;
FIG. 4 is a diagram of a matrix representation of time periods in a time series T according to the present invention;
FIG. 5 is a schematic diagram of upper bound of calculated flow based on dynamic programming according to the present invention;
FIG. 6 is a graph of execution time as a function of the duration of a data set in an embodiment;
FIG. 7 is a graph of the effect of time thresholds on algorithm execution time in an embodiment;
FIG. 8 is a graph of the effect of spatial thresholds on algorithm execution time in an embodiment;
FIG. 9 is a graph illustrating the effect of flow threshold on algorithm execution time in an embodiment.
Detailed Description
The technical solution of the present invention is described in detail below, but the scope of the present invention is not limited to the embodiments.
For the understanding of the technical solutions of the present patent application, the corresponding technical terms are explained and defined as follows:
definition 1 (dynamic space network) a dynamic space network G ═ (V, E), and the time interval associated with G is denoted T, V ═ V1,…,vnDenotes a vertex set, E ═ E }1,…,emDenotes a set of edges, T ═ T1,…,tkDenoted as a time sequence of unit time intervals (time intervals for short), tiDenotes the ith unit time interval in T, time period Ti,jA subsequence of T, denoted as Ti,j={ti,…,tjJ is less than or equal to 1 and less than or equal to k, and | is expressed as the total number of unit time intervals in the time sequence. The flow attribute of any edge E E is in a time series form and is marked as f (E, t)1),…,f(e,tk)}。
The ingress traffic defining the 2 (ingress/egress traffic) edge refers to the number of moving objects entering the edge. The entry rate of the edge e in the unit time interval t is denoted as fin(e,t)=∑e′∈(E-e)f (e ', e, t), where f (e ', e, t) represents the number of moving objects entering e from e ' within the time interval t. Similarly, the output of edge e is denoted as fout(e,t)=∑e′∈(E-e)f (e, e', t). Let E ' be a subset of E, the incoming flow rate of E ' in a unit time interval represents the total number of moving objects entering E ' from other edges, denoted as fin(E′,t)=∑e∈E′,e′∈(E-E′)f (e', e, t). Similarly, the output of E' is denoted as fout(E′,t)=∑e∈E′,e′∈(E-E′)f(e,e′,t)。
Definition 3 (net flow) net flow represents the net value of the incoming edge or edge set flow. The net flow rate of edge e is f in unit time interval ta(e,t)=fin(e,t)-fout(e, t). Similarly, the net flow f of the edge set Ea(E,t)=fin(E,t)-fout(E, t). In a specified time period Ti,jNet flow of inner, edge eAnd net flow of edge set EIf the inlet flow is less than the outlet flow, the net flow is negative.
Definition 4 (hot spot region) is given as (V, E, L) in the form of a triplet, where s.v and s.e are subsets G.V and G.E, respectively, and form a sub-graph of G, and s.l represents the life cycle of S, and S is called a hot spot region if S satisfies all the following constraints.
1) Connectivity: s is a connected subgraph in G;
2) regionality: the diagonal length of the Minimum Bounding Rectangle (MBR) of S is less than a threshold value d, i.e. MBR (S) < DL < d;
3) persistence and scalability: there is a time subsequence Ti,jSo that fa(S.E,Ti,j)/|Ti,j| is more than or equal to delta and | Ti,j| is ≧ theta, where δ>0,θ>0,S.L=Ti,j。
Technical problem definition: given a dynamic spatial network G and a time series T, a time threshold θ, a space threshold d, and a traffic threshold δ, the objective of the hot spot region mining algorithm is to find a hot spot region set HS ═ S in G that meets the above definition1,S2,…,SnWhere HS satisfies the following condition.
1) For theS satisfies definition 4;
2) for theS does not satisfy definition 4;
3)Sa.E∩Sbe ≠ and Sa.L∩SbL ≠ cannot be satisfied at the same time.
The invention discloses a hot spot region detection method based on a dynamic space network model under an urban road network environment, which sequentially comprises the following steps of:
(1) constructing a Dynamic Space Network (DSN);
(1.1) constructing space and topology information of a Dynamic Space Network (DSN), namely extracting inflection points and road sections of intersections and road directions according to map data (including road network data, geographic information, user labeling data and the like) to establish the space and topology information of the DSN, namely acquiring nodes and edges in the dynamic space network;
(1.2) calculating the flow attribute of the edge, namely establishing the association between the track and the road by adopting a road network matching method, and then calculating the net flow sequence { w) of each edge in the DSNi1,wi2,wi3,…};
(2) And designing a data structure (including topological relation and traffic information, specifically longitude and latitude coordinates of nodes, connection relation among the nodes and traffic information of edges) of the DSN.
As shown in FIG. 1, all edges { e } are first stored using a linear structure1,e2,e3… }, each edge eiThe elements are respectively connected with a net flow sequence { wi1,wi2,wi3… and two nodes vi,vj}; thus, the traffic attribute of an accessible edge given an edge, and associated node information; then each node viConnecting an edge set ei1,ei2,ei3…, such that given an edge, other edge information connected to it is available;
(3) adopting a partitioning-filtering-detecting method to carry out the excavation of the hot spot area, namely: firstly, dividing a dynamic space network DSN in space to obtain a candidate area set CA; then, filtering candidate areas in different time periods by adopting estimation functions F and H of an upper flow bound; and finally, detecting hot spot areas aiming at the candidate areas to obtain a hot spot area set HS.
Algorithm 1 describes the execution of the hot spot region mining algorithm. First, the dynamic space network G is spatially divided according to a given threshold to obtain a divided candidate area set CA (row 2). Then, each candidate area in the CA is divided in time to obtain candidate areas in different time periods, and a flow upper bound estimation method is adopted to filter a candidate area set TS (3-6 lines). And finally, arranging time periods meeting the continuity requirement in the TS according to the descending order of the length, and sequentially detecting the hot spot regions of the candidate regions in each time period to obtain a final hot spot region set HS (7-12 rows).
1. In the step (1.1), road network data is used as input data of a dynamic space network DSN, namely intersections (intersections of roads) and road inflection points (position points of road direction change) of roads in a road network are mapped to nodes in a space network graph, and road sections in the road network are mapped to edges; the output data of the dynamic space network DSN, i.e. the spatial and topological information in the DSN, is obtained by the above-mentioned input data extraction.
The input data format is as follows:
TABLE 1 File Format for road segment data
Attribute name | Data type | Remarks for note |
edge_id | Shaping machine | ID of edge |
from | Shaping machine | ID of a node |
to | Shaping machine | ID of a node |
name | Character string | |
length | Shaping machine | |
bridge | Boolean value | Yes or No |
oneway | Boolean value | True or False |
TABLE 2 File Format of intersection data
Attribute name | Data type | Remarks for note |
vertex_id | Shaping machine | ID of a node |
geom | Floating point number | The format is as follows: POINT (latitude, longitude) |
The output data format is as follows:
table 3 file format of space and topology information in DSN
Attribute name | Data type | Remarks for note |
edge_id | Shaping machine | ID of edge |
sp_id | Shaping machine | ID of origin |
sp_lng | Floating point number | Longitude of origin |
sp_lat | Floating point number | Latitude of origin |
ep_id | Shaping machine | ID of endpoint |
ep_lng | Floating point number | Longitude of the endpoint |
ep_lat | Floating point number | Latitude of endpoint |
2. In the step (1.2), road network matching is carried out by adopting an HMM-based algorithm, and the specific process is as follows:
combining the time sequence characteristics and the spatial attributes of data points in the track, simulating observation probability transition by using Gaussian distribution, wherein the state transition probability is calculated by integrating speed information, GPS observation points and real points, and searching a group of candidate road sections and candidate points for each GPS sampling point according to a road section projection process; then constructing a candidate graph, wherein nodes in the candidate graph are candidate point sets observed by each GPS, and edges are shortest path sets between any two adjacent candidate points; the nodes and the edges are distributed with corresponding weight values based on the space/time analysis result; and finally, combining the observation and the transition probability to search a path with the highest score in the candidate graph.
The input data includes spatio-temporal trajectory data (table 4) and road network data (tables 1 and 2).
TABLE 4 File Format of spatiotemporal trajectories
Attribute name | Data type | Remarks for note |
traj_id | Shaping machine | ID of track |
lng | Floating point number | Longitude of point |
lat | Floating point number | Latitude of point |
time | Time stamp |
The output data format is:
TABLE 5 File Format of side-flow information in DSN
Attribute name | Data type | Remarks for note |
edge_id | Shaping machine | ID of edge |
weight | Shaping machine | Weight of |
time | Time stamp |
3. The detailed process of dividing the meeting candidate region set in the step (3) is as follows:
(A) and (3) carrying out space division on the dynamic space network G: and (3) setting the spatial threshold of the hot spot region as D, and adopting a square grid with the side length of D to perform spatial division on G to obtain a D multiplied by D grid. The inclusion relationship between the hot spot region and the grid neighborhood is shown in fig. 2, wherein G is a 4 × 4 grid, the grids are numbered, and the grid with row number a and column number b is marked as GabA is more than or equal to 1, b is less than or equal to D, the grid gabThe set of included edges is denoted as gab.E;
(B) Generating a candidate region set according to the space division result: the candidate area CA of the dynamic spatial network G is represented by a square GabI.e. all grid neighborhoods constitute CA ═ AR (g)ab)|1≤a,b≤D};
Because the minimum circumscribed rectangle MBR of the hot spot area is smaller than or equal to a square grid with the side length of d in size, the hot spot area is necessarily contained by a square grid neighborhood, and the square grid neighborhood consists of peripheral square grids surrounding the square grid. In short, a square g is givenabIts neighborhood AR (g)ab) Is defined as the sum of all g satisfying the requirements of | p-a ≦ 1, | q-b ≦ 1pqThe region of composition, the edge set of the neighborhood is denoted as AR (g)ab).E。
4. The specific process of filtering the candidate region in the step (3) is as follows:
(A) modeling the change process of the neighborhood by adopting a time sequence form;
as shown in FIG. 3, the time span of the neighborhood AR (g) coincides with the time series T of the dynamic spatial network, the time period of the neighborhood denoted as Ti,j={ti,ti+1,…,tj1 ≦ i, j ≦ k, the time period being represented in matrix form;
as shown in FIG. 4, the time threshold for the hot spot region is θ, so the time period for which the threshold requirement is met should be { T }i,j|1≤i≤j≤k,j-i≥θ}。
(B) Adopting a flow estimation filtering method based on dynamic programming to filter candidate areas for the first time, namely: carrying out flow summarizing calculation on all edges with positive net flow in the candidate region in the time period by adopting an estimation function F of an upper flow bound, wherein if a hot spot region exists in the candidate region, the net flow is less than or equal to the upper flow bound formed by all edges with positive flow in the candidate region no matter whether the hot spot region contains the edges with negative flow;
candidate area AR (g) for a specified time period Ti,jThe upper bound of the inner flow is:
whereinA side indicating that the flow rate is a positive value in the time interval t; f (AR), (g), Ti,j) Simplified to F (i, j); given time period Ti,jInner, net flow f of any subgraph S edge seta(S.E,Ti,j) Equal to the sum of the net flows of all edgesMaximum net flow based on the principle of estimation of the upper bound of flow
As shown in fig. 5, in the first filtering method,directly calculating the upper bound of the flow to be estimated by adopting a dynamic programming method through the estimated upper bound of the flow, and further reducing the time complexity to O (mk)2);
Namely: f (i, j) ═ F (i-1, j) + F (i, j-1) -F (i-1, j-1),
(C) performing secondary filtering on the candidate result after the primary filtering by adopting an estimation function H;
the upper flow bound estimate H function is:simplified to H (i, j), whereinIs shown during the time period Ti,jThe inner net flow sum value is a positive edge;
in the candidate area AR (g), each edge e is in the time period Ti,jSum of sigma with a net flow thereine∈s.Efa(e,Ti,j) And the sum of all positive values in the candidate region in the time is
Wherein the sum of all positive sum values is smaller than the calculation result of the estimation function F, i.e.: is shown at Ti ,jThe total value of the internal net flow is a positive edge, and then secondary filtration is realized.
5. In step (3), the current time period Ti,jThe method for searching the hot spot area of the candidate area ar (g) of (1) is: first calculate the number of edges in the square g at Ti,jNet flow summary value within; then selecting the edge with the maximum net flow sum value in the g to form an initial subgraph S; then, expanding the subgraph in an iterative mode;
in each iteration, selecting adjacent edges of the current subgraph S as a candidate edge set, scoring each edge e in the candidate edge set, then adding the edge with the highest score into the subgraph S, if the edge is a-infinity edge or an edge with MBR (S + e). DL exceeding a spatial threshold value cannot be expanded, and ending the iteration until no candidate edge can be expanded;
if the sub-graph S at this time satisfies: s.fa/|Ti,jIf | ≧ δ, outputting the result, and deleting the edge related to S from the candidate data set to prevent repeated detection; finally, recalculating the upper flow bound of the remaining edges in the candidate area, if the upper flow bound meets the threshold requirement, adding the upper flow bound into the candidate area set again for re-detection, and otherwise, deleting the upper flow bound;
repeatedly searching the hot spot areas according to the method until the candidate area set is empty;
wherein the scoring method is shown in formula 1:
where Δ is a change value of a diagonal line of the subgraph S after the edge e is added, that is, Δ ═ MBR (S + e). DL-MBR (S) · DL, where "S + e" indicates a new subgraph after e is added.
Example (b):
(1) experimental setup
To verify the effectiveness and efficiency of the present invention, the present embodiment uses real GPS trajectory data and road network data for experiments. The experimental environment is a Windows10 operating system, the processor is Intel core i5, the main frequency is 3.20GHz, the memory size is 12GB, the hard disk capacity is 500GB, and the rotating speed of the hard disk is 7200 r/s.
The experimental data are road network data of Shanghai city and taxi track data. The road network data is composed of 286591 road segments and 262764 intersections. The trajectory data is a spatiotemporal trajectory formed by 13518 taxis in shanghai city on 4/2/2015. And constructing a dynamic space network based on the two types of data, wherein the data size is 18.5 GB.
(2) Results and analysis of the experiments
In this embodiment, the effectiveness test and the algorithm efficiency test are respectively performed on the hot spot region mining algorithm HSM.
(2.1) Algorithm validation experiment
The comparison object of the experiment is the existing black hole pattern mining algorithm BHM. Because the constraint requirements of the two modes on the group scale and the space regionality are the same, the experiments are compared from the two aspects of the overlapping condition and the time length of the mining results. The term "overlap" refers to the spatial overlap of the two mining algorithm results, i.e., the MBRs of the two results are spatially crossed, indicating that the two results are adjacent to each other, and thus the two results are determined to be "overlap". If the results of the two mining algorithms have high overlap ratio, namely the number of overlapped areas is large, the HSM method and the BHM method are effective as well. Furthermore, if the HSM method gets more results than the BHM under the condition of high coincidence, the modeling method of the hot spot region is more effective than that of the black hole mode.
On the other hand, the embodiment also needs to compare the time duration in the coincidence result. Under the same parameter setting, three conditions of inclusion, separation and intersection exist in the time interval of the mining result of the two algorithms. If the time interval of the HSM includes more time intervals of the BHM in the coincidence result, the crowd gathering activity which can be more completely discovered by the HSM method is shown.
In the validation experiment, both methods (i.e., HSM and BHM) employed a common data set, which was track data for 4 months and 2 days. The parameters of the validation experiment were set as follows.
(1) Parameter setting of HSM
The time threshold θ is 3, the spatial threshold d is 150 m, the traffic threshold δ is 5, and the unit time intervals of dynamic spatial network traffic aggregation are 1 minute, 2 minutes, 3 minutes, 4 minutes, and 5 minutes, respectively.
(2) Parameter setting of BHM
To ensure the fairness of the comparison of the two algorithms, the flow threshold of the black hole mode is θ × δ 15, the spatial threshold is 150 meters, and the unit time intervals of the spatial network traffic summary are 3 minutes, 6 minutes, 9 minutes, 12 minutes and 15 minutes.
With the above time interval settings, the minimum length of crowd sourcing activity detected by both methods was 3 minutes, 6 minutes, 9 minutes, 12 minutes and 15 minutes, respectively. According to the problem definition of the two methods, the shortest time period of the result of the HSM method is the same as that of the BHM method, and the flow rate is also the same in the time period, so that the two methods find approximately the same mining result, and the comparison of the coincidence result of the two methods is facilitated.
The coincidence of the two excavation methods is shown in table 6. Under different flow summarizing time intervals, the percentage of mining results of the BHM method exceeds 90%, which indicates that the mining results of the HSM method almost completely cover the results of the BHM. Furthermore, the percentage of HSM in the table is around 50%, which indicates that the HSM method of the present invention can detect the crowd-sourcing activity that the BHM method cannot detect. The reason for this is that the BHM method is only defined for a period of time of the spatial network, and does not consider the modeling problem of the spatial network change, so that the black hole pattern in a short period of time cannot be found.
The HSM method provided by the invention models the traffic change of the space network, and can accurately discover the crowd gathering activity in any time interval. Experimental results show that the hot spot area provided by the invention can be used for more effectively detecting the aggregation events in the city.
TABLE 6 comparison of mining results for HSM and BHM methods
Besides the comparison of the coincidence conditions of the two mining methods, the embodiment further analyzes the relationship (including inclusion, intersection and separation) between the average duration and the time period in the coincidence results. From table 7, it can be found that the separation ratio of all the results is low, not more than 6%, and the ratio of the inclusion relationship is more than 60%. This, in combination with the high BHM ratios in table 6, demonstrates that the HSM method of the present invention can not only find the results found by most BHM methods, but also obtain results that are significantly longer in duration than those of BHM. The reason for this is that the HSM method models the dynamic change process of the hot spot region, and can find the complete duration of the aggregation activity, while the detection effect of the BHM method depends on the spatial network traffic aggregation condition.
TABLE 7 comparison of mining result durations for HSM and BHM methods
(2.2) method efficiency test
After the effectiveness of the HSM method is proved, this embodiment will perform an HSM efficiency experiment, where two candidate region filtering methods are compared, namely, a baseline filtering algorithm blf (baseline filter) and a filtering algorithm FUF based on the flow upper bound estimation function.
The efficiency of the operation of the two algorithms under data sets of different durations is first compared. The input data start time of each experiment is 2015, 4 months, 2 days, 6 am, the flow summarizing time interval is 1 minute, and the scale of the test data set is controlled by setting different end times. The specific parameters are as follows: the time threshold θ is 3, the space threshold d is 100 m, and the flow threshold δ is 3.
As can be seen from fig. 6, the run time of the FUF algorithm is significantly lower than the BLF algorithm for the same data set duration. FUF is not sensitive to increases in data size compared to the BLF algorithm.
The second set of experiments will compare the impact of the time threshold, space threshold and flow threshold parameters on the run time of the two algorithms. In the experiment, when one threshold parameter is tested, the other two parameters are kept at default values. The duration of the test data was 2 hours, and the flow summary time interval was 1 minute. Default values for the threshold parameter are set to: the time threshold θ is 3, the space threshold d is 100 m, and the flow threshold δ is 3.
Fig. 7, 8 and 9 compare algorithm run times at different time thresholds, spatial thresholds and flow thresholds, respectively. It can be seen that the FUF algorithm improves operating efficiency by an order of magnitude over the baseline algorithm BLF. The execution time of both algorithms decreases with increasing time threshold and flow threshold, since an increase in both thresholds represents an increased demand for mining results, resulting in fewer candidate sets and shorter algorithm run times. Similarly, as the spatial threshold increases, the candidate set of hot spot regions becomes larger, and the algorithm execution time increases accordingly.
The embodiment can show that the invention can accurately and efficiently discover the hot spot area in the urban space in the appointed time interval.
Claims (7)
1. A hot spot region detection method based on a dynamic space network model under an urban road network environment is characterized by comprising the following steps: the method sequentially comprises the following steps:
(1) constructing a Dynamic Space Network (DSN);
(1.1) constructing space and topology information of a Dynamic Space Network (DSN), namely extracting corners and sections of intersections and road directions according to map data to establish space and topology information of the DSN, namely acquiring nodes and edges in the dynamic space network;
(1.2) calculating the flow attribute of the edge, namely establishing the association between the track and the road by adopting a road network matching method, and then calculating the net flow sequence { w) of each edge in the DSNi1,wi2,wi3,…};
(2) Designing a data structure of a Dynamic Space Network (DSN):
first, all edges { e ] are stored using a linear structure1,e2,e3… }, each edge eiThe elements are respectively connected with a net flow sequence { wi1,wi2,wi3… and two nodes vi,vj}; thus, the traffic attribute of an accessible edge given an edge, and associated node information; then each node viConnecting an edge set ei1,ei2,ei3…, such that given an edge, other edge information connected to it is available;
(3) adopting a partitioning-filtering-detecting method to carry out the excavation of the hot spot area, namely: firstly, dividing a dynamic space network DSN in space to obtain a candidate area set CA; then, sequentially filtering candidate areas in different time periods by adopting estimation functions F and H of an upper flow bound; and finally, detecting hot spot areas aiming at the candidate areas to obtain a hot spot area set HS.
2. The method according to claim 1, wherein the method comprises the following steps: in the step (1.1), road network data is used as input data of a dynamic space network DSN, that is, intersections and inflection points of roads in a road network are mapped to nodes in a space network graph, and road sections in the road network are mapped to edges; the output data of the dynamic space network DSN, i.e. the spatial and topological information in the DSN, is obtained by the above-mentioned input data extraction.
3. The method according to claim 1, wherein the method comprises the following steps: the step (1.2) adopts an HMM-based algorithm to perform road network matching, and the specific process is as follows:
combining the time sequence characteristics and the spatial attributes of data points in the track, simulating observation probability transition by using Gaussian distribution, calculating the state transition probability by integrating rate information, GPS observation points and real points, and searching a group of candidate road sections and candidate points for each GPS sampling point according to a road section projection process; then constructing a candidate graph, wherein nodes in the candidate graph are candidate point sets observed by each GPS, and edges are shortest path sets between any two adjacent candidate points; the nodes and the edges are distributed with corresponding weight values based on the space/time analysis result; and finally, combining the observation and the transition probability to search a path with the highest score in the candidate graph.
4. The method according to claim 1, wherein the method comprises the following steps: the detailed process of dividing the meeting candidate region set in the step (3) is as follows:
(A) and (3) carrying out space division on the dynamic space network G: setting the space threshold of the hot spot area as D, adopting the square grids with the side length of D to perform space division on G to obtain a grid of DxD, and carrying out square grid division on the gridLine number, and a square with a line number a and a column number b is denoted as gabA is more than or equal to 1, b is less than or equal to D, the grid gabThe set of included edges is denoted as gab.E;
(B) Generating a candidate region set according to the space division result: the candidate area CA of the dynamic spatial network G is represented by a square GabI.e. all grid neighborhoods constitute CA ═ AR (g)ab)|1≤a,b≤D};
Because the minimum circumscribed rectangle MBR of the hot spot area is smaller than or equal to a square with a side length of d in size, the hot spot area is necessarily contained by a square neighborhood, and the square neighborhood is composed of peripheral squares surrounding the square.
5. The method according to claim 1, wherein the method comprises the following steps: the specific process of filtering the candidate region in the step (3) is as follows:
(A) modeling the change process of the neighborhood by adopting a time sequence form;
the time span of the neighborhood AR (g) coincides with the time sequence T of the dynamic spatial network, the time period of this neighborhood being denoted Ti,j={ti,ti+1,…,tj},1≤i,j≤k:
Wherein, the time threshold of the hot spot region is theta, so the time period meeting the threshold requirement is { T }i,j|1≤i≤j≤k,j-i≥θ。
(B) Adopting a flow estimation filtering method based on dynamic programming to filter candidate areas for the first time, namely: carrying out flow summarizing calculation on all edges with positive net flow in the candidate region in the time period by adopting an estimation function F of an upper flow bound, wherein if a hot spot region exists in the candidate region, the net flow is less than or equal to the upper flow bound formed by all edges with positive flow in the candidate region no matter whether the hot spot region contains the edges with negative flow;
candidate area AR (g) for a specified time period Ti,jThe upper bound of the inner flow is:
whereinA side indicating that the flow rate is a positive value in the time interval t; f (AR), (g), Ti,j) Simplified to F (i, j); given time period Ti,jInner, net flow f of any subgraph S edge seta(S.E,Ti,j) Equal to the sum of the net flows of all edgesMaximum net flow based on the principle of estimation of the upper bound of flow
(C) Performing secondary filtering on the candidate result after the primary filtering by adopting an estimation function H;
the upper flow bound estimate H function is:simplified to H (i, j), whereinIs shown during the time period Ti,jThe inner net flow sum value is a positive edge;
in the candidate area AR (g), each edge e is in the time period Ti,jSum of sigma with a net flow thereine∈S.Efa(e,Ti,j) And the sum of all positive values in the candidate region in the time is
Wherein the sum of all positive sum values is smaller than the calculation result of the estimation function F, i.e.: is shown at Ti,jThe total value of the internal net flow is a positive edge, and then secondary filtration is realized.
6. The method according to claim 1, wherein the method comprises the following steps: in the step (3), the current time period Ti,jThe method for searching the hot spot area of the candidate area ar (g) of (1) is: first calculate the number of edges in the square g at Ti,jNet flow summary value within; then selecting the edge with the maximum net flow sum value in the g to form an initial subgraph S; then, expanding the subgraph in an iterative mode;
in each iteration, selecting adjacent edges of the current subgraph S as a candidate edge set, scoring each edge e in the candidate edge set, then adding the edge with the highest score into the subgraph S, if the edge is a-infinity edge or an edge with MBR (S + e). DL exceeding a spatial threshold value cannot be expanded, and ending the iteration until no candidate edge can be expanded;
if the sub-graph S at this time satisfies: s.fa/|Ti,jIf | ≧ δ, outputting the result, and deleting the edge related to S from the candidate data set to prevent repeated detection; finally, recalculating the upper flow bound of the remaining edges in the candidate area, if the upper flow bound meets the threshold requirement, adding the upper flow bound into the candidate area set again for re-detection, and otherwise, deleting the upper flow bound;
repeatedly searching the hot spot areas according to the method until the candidate area set is empty;
wherein the scoring method is shown in formula 1:
where Δ is a change value of a diagonal line of the subgraph S after the edge e is added, that is, Δ ═ MBR (S + e). DL-MBR (S) · DL, where "S + e" indicates a new subgraph after e is added.
7. The method according to claim 5, wherein the method comprises the following steps: in the first filtering method, the upper bound of the flow to be estimated is directly calculated by adopting a dynamic programming method through the estimated upper bound of the flow, and the time complexity is reduced to O (mk)2);
Namely: f (i, j) ═ F (i-1, j) + F (i, j-1) -F (i-1, j-1).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910857903.4A CN110609824B (en) | 2019-09-09 | 2019-09-09 | Hot spot area detection method based on dynamic space network model under urban road network environment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910857903.4A CN110609824B (en) | 2019-09-09 | 2019-09-09 | Hot spot area detection method based on dynamic space network model under urban road network environment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110609824A true CN110609824A (en) | 2019-12-24 |
CN110609824B CN110609824B (en) | 2022-09-09 |
Family
ID=68892694
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910857903.4A Active CN110609824B (en) | 2019-09-09 | 2019-09-09 | Hot spot area detection method based on dynamic space network model under urban road network environment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110609824B (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111626582A (en) * | 2020-05-20 | 2020-09-04 | 交通运输部公路科学研究所 | Urban traffic trip problem hotspot grading method and device |
CN111680743A (en) * | 2020-06-05 | 2020-09-18 | 腾讯科技(深圳)有限公司 | Data processing method and device, computer equipment and storage medium |
CN112579920A (en) * | 2020-12-09 | 2021-03-30 | 成都中科大旗软件股份有限公司 | Cross-space-time propagation analysis method based on events |
CN113724489A (en) * | 2021-07-22 | 2021-11-30 | 东南大学 | Traffic jam tracing method based on multi-source data |
CN114090937A (en) * | 2021-11-29 | 2022-02-25 | 重庆市地理信息和遥感应用中心 | Automatic urban spatial feature area division system |
CN114495514A (en) * | 2022-02-16 | 2022-05-13 | 中南大学 | Multi-source data collaborative vehicle illegal turning hot spot area identification method |
CN116631195A (en) * | 2023-07-20 | 2023-08-22 | 江西师范大学 | Regional abnormality detection method based on urban sub-region hot spot crossing mining |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102750361A (en) * | 2012-06-13 | 2012-10-24 | 中国矿业大学 | Discovery method of track data hot spot based on local multilayer grids |
US20150199380A1 (en) * | 2014-01-16 | 2015-07-16 | Microsoft Corporation | Discovery of viewsheds and vantage points by mining geo-tagged data |
CN108108407A (en) * | 2017-12-11 | 2018-06-01 | 南京师范大学 | Group movement mobile cluster pattern sort method based on space-time track |
WO2018099578A1 (en) * | 2016-12-02 | 2018-06-07 | Fleetmatics Ireland Limited | System and method for deriving operation information of a fleet and analyzing the same |
CN108427965A (en) * | 2018-03-05 | 2018-08-21 | 重庆邮电大学 | A kind of hot spot region method for digging based on road network cluster |
CN109830102A (en) * | 2019-02-14 | 2019-05-31 | 重庆邮电大学 | A kind of short-term traffic flow forecast method towards complicated urban traffic network |
-
2019
- 2019-09-09 CN CN201910857903.4A patent/CN110609824B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102750361A (en) * | 2012-06-13 | 2012-10-24 | 中国矿业大学 | Discovery method of track data hot spot based on local multilayer grids |
US20150199380A1 (en) * | 2014-01-16 | 2015-07-16 | Microsoft Corporation | Discovery of viewsheds and vantage points by mining geo-tagged data |
WO2018099578A1 (en) * | 2016-12-02 | 2018-06-07 | Fleetmatics Ireland Limited | System and method for deriving operation information of a fleet and analyzing the same |
CN108108407A (en) * | 2017-12-11 | 2018-06-01 | 南京师范大学 | Group movement mobile cluster pattern sort method based on space-time track |
CN108427965A (en) * | 2018-03-05 | 2018-08-21 | 重庆邮电大学 | A kind of hot spot region method for digging based on road network cluster |
CN109830102A (en) * | 2019-02-14 | 2019-05-31 | 重庆邮电大学 | A kind of short-term traffic flow forecast method towards complicated urban traffic network |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111626582A (en) * | 2020-05-20 | 2020-09-04 | 交通运输部公路科学研究所 | Urban traffic trip problem hotspot grading method and device |
CN111680743A (en) * | 2020-06-05 | 2020-09-18 | 腾讯科技(深圳)有限公司 | Data processing method and device, computer equipment and storage medium |
CN111680743B (en) * | 2020-06-05 | 2021-05-11 | 腾讯科技(深圳)有限公司 | Data processing method and device, computer equipment and storage medium |
CN112579920A (en) * | 2020-12-09 | 2021-03-30 | 成都中科大旗软件股份有限公司 | Cross-space-time propagation analysis method based on events |
CN113724489A (en) * | 2021-07-22 | 2021-11-30 | 东南大学 | Traffic jam tracing method based on multi-source data |
CN113724489B (en) * | 2021-07-22 | 2022-07-01 | 东南大学 | Traffic jam tracing method based on multi-source data |
CN114090937A (en) * | 2021-11-29 | 2022-02-25 | 重庆市地理信息和遥感应用中心 | Automatic urban spatial feature area division system |
CN114495514A (en) * | 2022-02-16 | 2022-05-13 | 中南大学 | Multi-source data collaborative vehicle illegal turning hot spot area identification method |
CN116631195A (en) * | 2023-07-20 | 2023-08-22 | 江西师范大学 | Regional abnormality detection method based on urban sub-region hot spot crossing mining |
CN116631195B (en) * | 2023-07-20 | 2023-10-13 | 江西师范大学 | Regional abnormality detection method based on urban sub-region hot spot crossing mining |
Also Published As
Publication number | Publication date |
---|---|
CN110609824B (en) | 2022-09-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110609824B (en) | Hot spot area detection method based on dynamic space network model under urban road network environment | |
Zhao et al. | Urban human mobility data mining: An overview | |
CN110298500B (en) | Urban traffic track data set generation method based on taxi data and urban road network | |
Lai et al. | Urban traffic Coulomb’s law: A new approach for taxi route recommendation | |
EP2590151A1 (en) | A framework for the systematic study of vehicular mobility and the analysis of city dynamics using public web cameras | |
Li et al. | Path-finding through flexible hierarchical road networks: An experiential approach using taxi trajectory data | |
CN108734955A (en) | The method and device of predicting road conditions state | |
Lassarre et al. | A GIS-based methodology for identifying pedestrians’ crossing patterns | |
Karagiorgou et al. | A layered approach for more robust generation of road network maps from vehicle tracking data | |
Blanke et al. | Crowdsourced pedestrian map construction for short-term city-scale events | |
Thakurzx et al. | Modeling and characterization of urban vehicular mobility using web cameras | |
Wilbur et al. | A decentralized approach for real time anomaly detection in transportation networks | |
CN102376025A (en) | Method for simulating mobile phone data and evaluating urban road network traffic condition | |
CN113342873B (en) | Population analysis unit division method based on city morphology and convergence mode | |
Muñoz-Villamizar et al. | Study of urban-traffic congestion based on Google Maps API: the case of Boston | |
CN108665723A (en) | A kind of information acquisition method and device | |
JP5901392B2 (en) | Information processing system and information processing method | |
Papageorgiou et al. | Developing a smart pedestrian network big data platform for municipal organizations | |
Pierdicca et al. | Identifying the use of a park based on clusters of visitors' movements from mobile phone data | |
CN110873571B (en) | Intersection stop line data acquisition method, system and device | |
Sun et al. | Excessive-emission vehicles real-time track matching algorithm based on road network topology and weights | |
Yadlowsky et al. | Link density inference from cellular infrastructure | |
Shein et al. | Discovery of loose group companion from trajectory data streams | |
Hilario et al. | Forecasting urban population distribution of Iloilo City using GIS and spatial autocorrelation models | |
Chang et al. | Addressing COVID-induced changes in spatiotemporal travel mobility and community structure utilizing trip data: An innovative graph-based deep learning approach |
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 |