CN104952248A - Automobile convergence predicting method based on Euclidean space - Google Patents
Automobile convergence predicting method based on Euclidean space Download PDFInfo
- Publication number
- CN104952248A CN104952248A CN201510310640.7A CN201510310640A CN104952248A CN 104952248 A CN104952248 A CN 104952248A CN 201510310640 A CN201510310640 A CN 201510310640A CN 104952248 A CN104952248 A CN 104952248A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- line segment
- track
- distance
- running orbit
- 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
Landscapes
- Navigation (AREA)
- Position Fixing By Use Of Radio Waves (AREA)
Abstract
The invention discloses an automobile convergence predicting method based on Euclidean space. The automobile convergence predicting method includes steps of (1) acquiring driving tracks of all cabs in an area to be predicted and preprocessing each driving track; (2) subjecting the preprocessing results obtained in the step (1) to segment clustering to obtain cluster results, and mining motion modes according to the cluster results; (3) predicting cabs arriving preset positions within a preset time according to the positions of the cabs at this moment and the motion mode mining results. Under the condition that road conditions are unknown, future tracks of the cabs can be predicted according to historical track data of the cabs, and on this basis, the automobile convergence predicting method can be applied to convergence prediction of the cabs in given places.
Description
Technical field
The present invention relates to technical field of intelligent traffic, be specifically related to a kind of vehicle based on Euclidean space and converge Forecasting Methodology.
Background technology
Nowadays, facing to the development trend of global IT application, the demand for development of traditional traffic technique and means no longer adapt to economic development society.Intelligent transportation system is the inevitable choice of communication development.Intelligent transportation develops along with the development of sensor technology, the communication technology, GIS technology (Geographic Information System), 3S technology (remote sensing technology, Geographic Information System, GPS three kinds of technology) and computer technology.
In recent years, along with the maturation of satellite positioning tech is with general, the present vehicles, comprise vehicle, are substantially provided with mobile satellite location equipment and carry out time recording to the track of vehicle.Therefore, have a large amount of vehicle satellite positioning track data every day to produce.The study hotspot that data mining is intelligent transportation field is carried out to the satellite location data produced in car operation process.
Existing method of carrying out excavating for satellite location data has a lot.Such as Guo Q, Luo J, Deng people at document " A ■ data-driven approach for convergence prediction on road network " (Web and Wireless Geographical Information Systems.Springer Berlin Heidelberg, a kind of method of the prediction vehicle Future Trajectory based on road network is proposed 2013:41-53), do not need in the method to know any individual driving habits in advance, and propose a statistical model to predict that the real-time vehicle based on road network converges situation.
Due to the impact of some practical factors, the timing position update strategy etc. that packet loss for example in transmitting procedure, the time interval are long, the location updating of the mobile objects such as vehicle may holiday routing information, therefore causes the vehicle satellite positioning track in database to exist uncertain.
For this problem, the people such as Guo Limin are at document " the uncertain trajectory predictions based on road network " (Journal of Computer Research and Development, 2010,47 (1): 104-112.) propose a kind of method of the uncertain Track Pick-up based on road network and the algorithm of representation and uncertain track Frequent Pattern Mining thereof in, also provided is a kind of index structure of fast finding trajectory model.This method provide a kind of can reduce run in real process as the situations such as packet loss the effective ways of error that brought.But also there are the following problems: prerequisite is when must know road network information, just can carry out subsequent operation
Summary of the invention
For the deficiencies in the prior art, the present invention proposes a kind of vehicle based on Euclidean space and converge Forecasting Methodology.
Vehicle based on Euclidean space converges a Forecasting Methodology, comprises the steps:
(1) obtain the carrying running orbit of all vehicles in region to be predicted, and pre-service is carried out to each carrying running orbit;
(2) line segment cluster is carried out to the pre-processed results of described step (1) and obtain cluster result, and carry out motor pattern excavation according to described cluster result;
(3) according to position and the motor pattern Result of each vehicle of current time, the vehicle of desired location in prediction setting-up time, is arrived.
In the present invention, region to be predicted is interpreted as the area at desired location place, is generally a certain setting regions in a city or city and sets according to application demand.
In the present invention, carrying track obtains by following method:
First the historical satellite locator data (directly obtaining from vehicle company) that all vehicles in region to be predicted run in setting-up time section is obtained.The timestamp that described satellite location data comprises license plate number, each satellite location data obtains, position (position coordinates comprises longitude and latitude), empty and load state etc.Wherein, empty and load state and passenger carrying status, loaded vehicle state representation is in passenger carrying status, and light car state representation is in non-passenger carrying status.
Then for each vehicle, its historical satellite locator data is screened, filter out the satellite location data of (when being loaded vehicle state) during vehicle carrying in satellite location data, and by the satellite location data that filters out according to license plate number and timestamp ordering, then draw the carrying running orbit obtaining this vehicle using each satellite location data as tracing point.
As follows pre-service is carried out to each carrying running orbit in described step (1):
(1-1) determine the time step point of this carrying running orbit and the position of space leaping point, and with the position of space leaping point, track is carried out to this carrying running orbit according to time step point and be separated;
(1-2) adopt Douglas-Peucker algorithm to carry out track simplification to the result after track separation and obtain pretreated carrying running orbit.
As preferably, determine the position of time step point by the following method:
If the interval of the timestamp that any two adjacent satellite location datas (tracing point) are corresponding is greater than the time threshold of setting in carrying running orbit, then think life period jump between these two adjacent satellite location datas.
Further preferably, the position of space leaping point is determined by the following method:
If the capacity-threshold being greater than setting of space Euclidean distance (i.e. Euclidean distance) between the position that in carrying running orbit, any two adjacent satellite location datas (tracing point) are corresponding, then think Existential Space jump between these two adjacent satellite location datas.
Wherein, time threshold and capacity-threshold set according to the moving law of vehicle, and time threshold is generally 10 ~ 20 minutes, are preferably 15 minutes.Capacity-threshold is 2 ~ 3km, is preferably 2.5km.
The object that track is separated running orbit is divided into multiple single running orbit.Single running orbit refers to that vehicle on purpose moves to the movement locus the process in another place from the three unities.Single running orbit excavates the basis of motor pattern after being, because motor pattern reflection is the exercise habit of vehicle when completing a single running orbit and routing preference.Track is separated by identifying that the mode of burble point realizes, and burble point comprises time step point, space leaping point, velocity sag point, long-time dwell point (handling well when pre-service) etc.
As preferably, in described step (1-1), track is carried out to carrying running orbit and is separated as follows:
(1-11) corresponding single running orbit is obtained as track separating resulting according to the position of time step point using carrying out track classification to the carrying track of each vehicle;
(1-12) for any one single running orbit, according to the position of space leaping point, division is carried out to present single running orbit and obtain some sub-running orbits as track separating resulting.
For improving precision of prediction further, further preferably, carry out also carrying out speed judgement to described sub-running orbit after track separation obtains sub-running orbit to carrying running orbit in described step (1-1):
If the speed in described sub-running orbit between any two continuity points exceedes default threshold speed, then think that point is abnormity point below, and give up this abnormity point, obtain revising sub-running orbit as track separating resulting.Negotiation speed judgement effectively can remove the satellite location data of mistake, effectively can improve precision of prediction.
Described step (1-2) adopts the Douglas-Peucker algorithm based on high line to carry out track simplification to each track separating resulting.The track separating resulting obtained in the present invention remains track, carries out track simplification specifically comprise the steps: it
A two of track separating resulting end points are connected into line segment by ();
B () to be determined on track separating resulting, from this line segment distance point farthest, to calculate the high linear distance of this distance point farthest to line segment, and proceed as follows:
If this high linear distance is less than default high linear distance threshold value, then give up this point, and using the track separating resulting after giving up as predigested running track (i.e. pretreated carrying running orbit);
Otherwise, retain distance point farthest, and it be connected with two end points, obtain two strip line segments;
C () performs step (b) respectively for the two strip line segments obtained.
The scope of described high linear distance threshold value is 50 ~ 150m, is preferably 80m.
Due to track be separated after usually can contain the record of a lot of redundancy in the result that obtains.Such as will report when certain object (vehicle) stops in certain position for a long time the positional information that a series of longitude and latitude is identical continuously, by the multiple point on same straight line of report during certain object straight-line travelling.The record of these redundancies may be nonsensical or can be inferred by the mode of linear fit by other record.The object that track simplifies removes these redundant recordings in single running orbit exactly, only retains the key point in track.
Described step carries out line segment cluster based on Euclidean space distance to all pre-processed results in (2), comprises the steps:
(2-1) each pre-processed results is divided into some line segments;
Pre-processed results in the present invention, through pretreated carrying running orbit, is in fact also running orbit, when line segment is divided to it, directly using between adjacent two points as a line segment.For each pre-processed results, the line segment quantity finally obtained depends on actual conditions, but owing to carrying out pre-service to data, actual every bar carrying running orbit is separated the line segment quantity obtained and can not differs too large.
(2-2) for any line segment that current pre-processed results is corresponding, in the line segment that other pre-processed results are corresponding, the line segment nearest with it is determined:
If the distance between these two line segments is less than default line segment distance threshold, then these two line segments are gathered for same class;
Otherwise, using this line segment separately as a class;
(2-3) for any two classes, calculate distance therebetween, if described distance is less than default class distance threshold, then the two is merged into a class;
Otherwise, do not process;
(2-4) execution step (2-3) is returned until stop when total class number is constant;
(2-5) add up the line segment quantity in each class, give up the class that line segment quantity is less than default line segment amount threshold.
Described line segment distance threshold is 150m, and described class distance threshold is 150m, described line segment amount threshold 15.During practical application, line segment distance threshold, class distance threshold and line segment amount threshold all can set according to practical application request.
Motor pattern mining process is carried out as follows in described step (2):
(S1) for any two classes, the distance between any two classes is calculated, and according to distance and when the annexation of the first two class middle conductor judges the proximity relations between the first two class:
If the distance between the first two class is less than default splicing distance threshold, the logarithm of the line segment that position judgment is wherein sliceable in each pre-processed results of then (namely originating) corresponding to it according to each line segment, if sliceable logarithm is greater than default logarithmic threshold, then think that these two classes are adjacent, otherwise, non-conterminous.
Described splicing distance threshold is generally 500 ~ 1500, and the present invention preferably 1000.
Described logarithmic threshold is 3 ~ 7, is preferably 3 in the present invention.
For two line segments belonging to different cluster in the present invention:
If the pre-processed results in these two line segment correspondence sources, and (namely in two line segments, the terminal of the preceding line segment of sequential is adjacent with the starting point of the posterior line segment of sequential continuously in the two position in the pre-processed results of correspondence, namely there is not other points in the middle of), then think that these two line segments are sliceable.
The sequential of middle conductor of the present invention is determined by the acquisition time of satellite location data corresponding to the beginning or end of line segment.Such as, when being determined by starting point acquisition time, the acquisition time of the satellite location data that starting point is corresponding front, then think this line segment be sequential front, on the contrary, if the acquisition time of satellite location data corresponding to starting point is rear, then think that this line segment is that sequential is rear.
Determine the source of each line segment for guarantee in the present invention and determine sliceable line segment, when each pre-processed results being divided into some line segments in cluster process, can add mark to each line segment, this mark is for illustration of the source (pre-processed results namely corresponding to it) of this line segment and the position of this line segment in the pre-processed results of correspondence.
Because line segment derives from continuous print track (pre-processed results), the source of line segment can first be numbered pre-processed results, represents corresponding relation with numbering.
The sequencing numbers of starting point in pre-processed results at every turn dividing the line segment obtained when dividing line segment, can be assigned to this line segment as line segment numbering to represent the position of line segment in the pre-processed results of correspondence by the position of this line segment in the pre-processed results of correspondence.
Accordingly, for belonging to inhomogeneous two line segments:
If the pre-processed results that these two line segments are corresponding identical, and the two position in the pre-processed results of correspondence continuously (namely continuous for representing the numbering of the position of line segment in the pre-processed results of correspondence), then think that these two line segments are sliceable.
(S2) build frequent pattern tree (fp tree) according to the proximity relations of all classes, in described frequent pattern tree (fp tree), root node is for preserving the list of the child node be connected with this root node, its child node and line segment cluster one_to_one corresponding; In described frequent pattern tree (fp tree), each non-root node comprises cluster and support two attributes, the class that this node of cluster attribute representation is corresponding, and Support value represents from this node to the degree of depth quantity of the vehicle operating track of the node being 1.
First according to the neighbouring relations between each cluster when building frequent pattern tree (fp tree) according to the proximity relations of all classes, build the oriented connected relation for representing neighbouring relations, node representation class in connected graph, while represent the annexation between adjacent two classes, and the arrow on limit represents the position relationship between two classes.
Then according to this directed connected graph, generate frequent pattern tree (fp tree), represent the frequent movement locus that we obtain.Wherein, the representative line segment of cluster corresponding for this motor pattern, according to the proximity relations of each cluster, by adjacent cluster splicing generation motor pattern, connects according to neighbouring relations and namely obtains frequent movement locus by frequent movement locus.
The node that each degree of depth that it should be noted that in scheme-tree is greater than 1 is the path of the father node of 1 to its degree of depth is a motor pattern, and all motor patterns exist all in this form in this tree.
The root node of frequent pattern tree (fp tree) only for preserving the list of the child node be connected with this root node, all child nodes and line segment cluster one_to_one corresponding.With the degree of depth be 1 node be root subtree saves with all frequent movement locus of line segment cluster ending corresponding to this node.
Frequent movement locus refers to that vehicle is by the higher section track of frequency, its reflection be the characteristics of motion and the exercise habit of vehicle.It also can regard the common sub-trajectory of one group of historical track as, but directly comparing historical track, to go to find common sub-trajectory complexity too high, and therefore we adopt the method extending to long common sub-trajectory from short common sub-trajectory.The shortest common sub-trajectory is the cluster of line segment, and its length is 1.Length be 2 common sub-trajectory can obtain by connecting two line segment clusters.Two line segment clusters connect into length, and to be the condition of the common sub-trajectory of 2 be: 1) geographic position of two classes is connected; 2) there is the historical track of some successively by the region at these two class places.The common sub-trajectory that can be N by length by similar process extends into the common sub-trajectory that length is N+1, thus excavates long common sub-trajectory, and these sub-trajectories just can as the basis of next step prediction.
Distance in the present invention between any two classes calculates by the following method:
Determine the representative line segment of each class, the distance between using the distance between the representative line segment of each class as two classes.
Described representative line segment is the starting point of all line segments in corresponding class and the position mean of terminal, be averaging the starting point coordinate of representatively line segment after being added by the coordinate of all starting points, will the coordinate of all terminals be added after be averaging the terminal point coordinate of representatively line segment.
In the present invention, starting point and terminal are determined according to the acquisition time (i.e. timestamp) of satellite location data corresponding to line segment two end points, and sequential is preceding is starting point, and sequential is posterior is terminal.
Proceed as follows for any vehicle in described step (3):
(3-1) position of desired location (i.e. position to be predicted) and this vehicle of current time is integrated on described frequent pattern tree (fp tree), and calculates the timestamp of desired location and the projected position of each vehicle of current time in frequent movement locus and correspondence according to merger result; Described timestamp represents the acquisition time of the satellite location data that this projected position is corresponding in this frequent movement locus.Concrete grammar is as follows:
Determine the class nearest with predeterminated position, calculate predeterminated position point represents line segment distance to such:
If a () this distance is greater than estimation range threshold value (usually get 30 ~ 100, be preferably 50 in the present invention), then prediction of failure (namely not measuring this vehicle running orbit in the future in advance by existing frequent movement locus);
(b) otherwise, carry out continuing to determine the class minimum with the positional distance of current time vehicle:
(b1) if this minor increment is greater than default estimation range threshold value, then think that this vehicle can not arrive desired location through setting-up time, and give up this vehicle;
(b2) otherwise, make vertical line from the position of this vehicle to the representative line segment apart from minimum class, and using intersection point position as the projected position of this vehicle in frequent movement locus, and calculate timestamp corresponding to this projected position:
C () position relationship of representative line segment two end points of class belonging to projected position and projected position, proportionally calculates the timestamp that this projected position is corresponding.
Calculate the timestamp that this projected position is corresponding as follows:
Belonging to supposing, two end points of the representative line segment of class are respectively A, corresponding timestamp is the time stamp T 2 that T1, B are corresponding, projected position is N, and corresponding timestamp is T3, calculates time stamp T corresponding to projected position 3 by following formulae discovery according to length ratio relation:
T3＝T1+(AN/AB)*(T2-T1))，
Wherein, AN is the distance between A to N, and AB is the distance between A to B.
In the present invention first by determine predeterminated position and all kinds of between distance compare with the threshold value preset, thus determine whether this position can predict, by setting predicted condition, substantially increase practicality and forecasting efficiency that vehicle converges Forecasting Methodology.
(3-2) this vehicle position residing after Preset Time is calculated as predicted position according to projected position in frequent movement locus of the position of this vehicle of current time and timestamp;
(3-3) distance between computational prediction position and desired location, if be less than threshold value (usually to get 30 ~ 80,50 are preferably in the present invention) and the Support value of this projected position node of correspondence in frequent pattern tree (fp tree) is greater than 3, then think that this vehicle can arrive desired location through setting-up time, otherwise, think that this vehicle can not arrive desired location through setting-up time.
Forecasting Methodology of the present invention, can also directly predict by adding up the quantity that can arrive the vehicle of setting position the quantity obtaining arriving the vehicle of setting position at setting-up time.
Do not make specified otherwise, the distance in the present invention all refers to Euclidean space distance, i.e. Euclidean distance.
Compared with prior art, tool of the present invention has the following advantages:
Do not needing under the condition knowing road network situation, can according to the historical trajectory data of vehicle, to its in the future track predict, and based on this, be applied to predict the convergence of given place vehicle.By this application, some helpful information can be obtained to a certain extent: as predicted the vehicle flow situation of a certain traffic intersection in following a period of time; Can by the convergence situation of prediction vehicle, perceive somewhere and some important events are occurring, such as concert is held in this ground, football match etc.; Oneself can wait for how long vehicle needs for needing people by bus to carry out predicting; Or the safety of some important departments as government organs and other places is monitored, if there is more vehicle can converge in these places within ensuing a period of time, then should cause certain attention.
Accompanying drawing explanation
Fig. 1 is the process flow diagram of the convergence of the vehicle based on the Euclidean space Forecasting Methodology of the present embodiment;
Fig. 2 is linear L
iwith line segment L
jbetween distance schematic diagram;
Fig. 3 is the structural representation of directed connected graph;
Fig. 4 is the structural representation of frequent pattern tree (fp tree).
Embodiment
Describe the present invention below in conjunction with the drawings and specific embodiments.
Vehicle based on Euclidean space converges a Forecasting Methodology, as shown in Figure 1, comprises the steps:
(1) satellite location data that in region to be predicted, all vehicles run in setting-up time section (directly obtaining from company belonging to vehicle) is obtained.
The satellite location data of the present embodiment comprises license plate number, timestamp, position (longitude+latitude), empty and load state etc.Wherein, empty and load state and passenger carrying status, loaded vehicle state representation is in passenger carrying status, and light car state representation is in non-passenger carrying status.
(2) pre-service is carried out to the satellite location data obtained.
First, filter out the satellite location data of (when being loaded vehicle state) during vehicle carrying in satellite location data, and the satellite location data filtered out is sorted respectively according to license plate number and timestamp obtain the carrying running orbit of vehicle.
(3) for each vehicle, track separation is carried out to its carrying track, carries out specifically carrying out operation when track is separated for Current vehicle as follows:
(3-1) will divide current carrying track according to the position of time step point, obtain the single running orbit that each vehicle is corresponding;
(3-2) for any one single running orbit, present single running orbit divides by the position according to space leaping point, obtains corresponding sub-running orbit;
In the present embodiment, the position of time step point is determined by the following method:
The interval of the timestamp of adjacent two satellite location datas is greater than the time threshold (in the present embodiment, time threshold gets 15 minutes) of setting, then think life period jump between these two adjacent satellite location datas.
In the present embodiment, the position of definition space jump is determined by the following method:
The space Euclidean distance (i.e. Euclidean distance) of adjacent two satellite location datas is greater than the capacity-threshold (getting 2.5km in the present embodiment) of setting, then think Existential Space jump between these two adjacent satellite location datas.
(3-3) carry out correction according to travelling speed to each sub-running orbit to obtain revising sub-running orbit, specific as follows:
Calculate the travelling speed of vehicle between satellite location data that in sub-running orbit, any two sequential are adjacent, if travelling speed is greater than default threshold speed, then think that wherein the posterior satellite location data of sequential is abnormal, and the satellite location data of this exception is deleted from sub-running orbit; Otherwise, inoperation.
The threshold speed S_max preset when antithetical phrase running orbit carries out speed judgement in the present embodiment is 60m/s.
Wherein, step (3-2) and (3-3) are preferably, can improve precision of prediction further.
The object that track is separated is that such historical track is divided into multiple single running orbit.Single running orbit refers to that vehicle on purpose moves to the movement locus the process in another place from the three unities.Single running orbit excavates the basis of motor pattern after being, because motor pattern reflection is the exercise habit of vehicle when completing a single running orbit and routing preference.Track is separated by identifying that the mode of burble point realizes, and burble point comprises time step point, space leaping point, velocity sag point, long-time dwell point (handling well when pre-service) etc.
(4) adopt the Douglas-Peucker algorithm based on high line to carry out track simplification to the sub-running orbit of correction and be simplified running orbit (for ease of subsequent operation, now each predigested running track is numbered, and successively successively each point in predigested running track is numbered according to sequential).
(4-1) two end points revising sub-running orbit are connected into line segment;
(4-2) find between two end points revising sub-running orbit from this line segment distance point farthest, and obtain the high linear distance of this point farthest to this line segment;
If (4-3) this high linear distance is less than set high linear distance threshold value (in the present embodiment, this high linear distance threshold value equals 80m), then give up this point, otherwise, retain this point, and be connected with two end points, obtain two strip line segments;
(4-4) (4-2) ~ (4-3) step is performed respectively to the two strip line segments obtained, until have new retention point to produce (namely not producing new sub-line segment) in (4-3) step.
Because the time interval of object report position information is shorter, the single running orbit that track obtains after being separated contains the record of a lot of redundancy.Such as will report when certain object stops in certain position for a long time the positional information that a series of longitude and latitude is identical continuously, by the multiple point on same straight line of report during certain object straight-line travelling.The record of these redundancies may be nonsensical or can be inferred by the mode of linear fit by other record.The object that track simplifies removes these redundant recordings in single running orbit exactly, only retains the key point in track.
(5) predigested running track corresponding for all vehicles is carried out line segment cluster
In the present embodiment, clustering object is simplify through above-mentioned steps the predigested running track obtained, and clustering method is as follows:
(5-1) (adjacent two points form a line segment each predigested running track to be split as some line segments, such as: ten points are removable is divided into 9 line segments), and each line segment is marked, mark comprises the numbering of predigested running track belonging to it, and this line segment is which the bar line segment in predigested running track belonging to it.
That the numbering of starting point in predigested running track of the line segment which the bar line segment in affiliated predigested running track obtains according to this fractionation is determined in the present embodiment, for ease of realizing, directly can represent that this line segment is which the bar line segment in predigested running track belonging to it with the numbering of starting point in predigested running track.
(5-2) for each line segment, find from its nearest line segment, and calculate the distance between these two line segments.
For any two line segment L
iwith line segment L
j, distance d is therebetween by following formulae discovery:
d＝d
⊥+d
||+d
e
And:
d
||=Max (l
|| 1, l
|| 2), d
θ=|| L
f|| * sin (θ),
Wherein, || L
f|| be line segment L
jlength,
L
⊥ 1, l
⊥ 2be respectively line segment L
jtwo end points to line segment L
ilength,
L
|| 1, l
|| 2be respectively line segment L
itwo end points to the distance of the vertical line nearest apart from it, described vertical line comprised line segment L
jtwo end points to line segment L
itwo vertical lines,
θ is line segment L
iwith line segment L
jangle.
Below in conjunction with Fig. 2 with line segment L
iwith line segment L
jbetween distance be the distance definition that example is described between two line segments: as shown in Figure 2, line segment L
iwith line segment L
j, their end points is s respectively
i, s
i, s
j, e
j, p
s, p
ebe respectively s
j, e
jto line segment L
jdo the intersection point that vertical line obtains, l
⊥ 1, l
⊥ 2the length of these two vertical lines respectively, l
|| 1, l
|| 2s respectively
ito p
sand s
jto p
edistance.θ was s
jdo and line segment L
iparallel parallel lines and line segment L
jangle, d
θfor crossing s
jdo and line segment L
iparallel parallel lines and l
⊥ 2the intersection point of line segment is to e
jdistance.
Line segment L
iwith line segment L
jbetween distance d be made up of three parts: d
⊥, d
||, d
θ, d=d
⊥+ d
||+ d
θ, wherein:
d
||＝Max(l
||1，l
||2)，d
θ＝||L
f||*sin(θ)。
If distance (5-3) is therebetween less than default line segment distance threshold (getting 150m in the present embodiment), two tracks are gathered for same class, and calculate the representative line segment (being defined as follows) of this class, otherwise, using current line segment separately as a group;
(5-4) calculate the distance (between cluster, distance definition is as follows) of any two groups, equally, what be less than default class distance threshold (getting 150m in the present embodiment) is classified as a class, is greater than, does not process; Circulation step (5-4) is until stop when cluster result no longer changes;
Distance between two clusters: the distance between the representative line segment of two clusters (i.e. class) is the distance between cluster.Wherein, the representative line segment of each class obtains by the following method:
Be averaging as starting point using the position of the starting point of all line segments in such, the position of the terminal of all line segments is averaging the position as terminal, and the obtained line segment that starting point is connected with terminal is this cluster representative line segment.
For any line segment, the starting point of line segment and terminal are determined according to the acquisition time of satellite location data corresponding to two end points, and sequential is preceding is starting point, and sequential is posterior is terminal.
(5-5) the line segment quantity in each cluster is added up: when quantity is less than amount threshold (getting 15 in the present embodiment), then give up such; Otherwise, retain, and then obtain final cluster result.
Each simplification track that previous step exports can be expressed as end to end directed line segment, and line segment cluster refers to classifies these line segments, and the line segment by closely similar (line segment head and the tail end points is close on geographic position) flocks together.The object of line segment cluster has two: 1) filtering object is by the lower section of frequency; 2) find out the track through each section, excavate for operational mode and prepare.Each line segment cluster contains the closely similar line segment (these attributes comprise position, direction and length) of one group of each attribute, and therefore the line segment of each intra-cluster can represent line segment to represent with one.
(6) carry out motor pattern excavation according to cluster result to all vehicles, concrete grammar is as follows:
(6-1) distance between any two classes is calculated, the logarithm that two classes of splicing distance threshold (getting 1000 in the present embodiment) judge wherein sliceable line segment is less than for distance, if logarithm is greater than logarithmic threshold (getting 3 in the present embodiment), then think that these two classes are adjacent, otherwise, non-conterminous;
(6-2), after adjacencies is determined, the neighbouring relations between poly-according to each, represent the connected relation between cluster (i.e. class) with connected graph.
The connected graph of the present embodiment is a digraph.Figure 3 shows that example, the node in connected graph is line segment cluster, while represent the annexation between cluster.The condition that there is the limit of Ci → Cj is that Ci represents the terminal of track and Cj represents the starting point of track closely, wherein, and i=1,2,3,4; J=1,2,3,4.
(6-3) corresponding frequent pattern tree (fp tree) is generated according to this connected graph, to represent the frequent movement locus of vehicle.
Frequent movement locus in the present embodiment: according to the proximity relations of each cluster, by adjacent cluster splicing generation motor pattern, connects the representative line segment of cluster corresponding for this motor pattern according to neighbouring relations and namely obtains frequent movement locus.
The root node of scheme-tree (i.e. frequent pattern tree (fp tree)) is only for preserving the list of child node, its child node (namely the degree of depth is the node of 1) is corresponding with the cluster result (i.e. class) of line segment cluster, is that the node of 1 is root subtree saves with all frequent movement locus of line segment cluster corresponding to this node with the degree of depth.Frequent pattern tree (fp tree) as shown in Figure 4, this scheme-tree has 9 nodes, be respectively N0, N1, N2, N3, N4, N5, N6, N7, N8, these 9 nodes are corresponding four clusters (being respectively C1, C2, C3 and C4) respectively, can find out, be that the subtree of root saves with these three frequent movement locus of C4-C2-C1, C2-C1 and C3-C1 of C1 ending with N1.
Each non-root node of scheme-tree comprises cluster and support two attributes, the line segment cluster of cluster attribute representation node association, support attribute representation is the quantity of the support value of the frequent movement locus of the path representative of the node of 1, the vehicle single running orbit of the actual node for being 1 from this node to the degree of depth from this node to the degree of depth.Such as the support value of this frequent movement locus of C4-C2-C1 is just kept in node N8, and the support value of C2-C1 to be kept in N5 etc.
According to definition, support attribute representation is the support value of the frequent movement locus of the path representative of the node of 1 from this node to the degree of depth, so in like manner, 30 in C1 represents that C1 nodes are to the degree of depth the be 1 support value of frequent movement locus of node of (be exactly it oneself).
The node that each degree of depth that it should be noted that in scheme-tree is greater than 1 is the path of the father node of 1 to its degree of depth is a motor pattern, and all motor patterns exist all in this form in this tree.As the C4-C2 pattern in figure, although C2 and C4 also exists with set membership in N1 subtree, they do not form a motor pattern.Its support also cannot calculate in N1 subtree, but is kept in N2 subtree.
Frequent movement locus refers to that vehicle is by the higher section track of frequency, its reflection be the characteristics of motion and the exercise habit of vehicle.It also can regard the common sub-trajectory of one group of historical track as, but directly comparing historical track, to go to find common sub-trajectory complexity too high, and therefore we adopt the method extending to long common sub-trajectory from short common sub-trajectory.The shortest common sub-trajectory is the cluster of line segment, and its length is 1.Length be 2 common sub-trajectory can obtain by connecting two line segment clusters.Two line segment clusters connect into length, and to be the condition of the common sub-trajectory of 2 be: 1) geographic position of two clusters is connected; 2) there is the historical track of some successively by the region at these two cluster places.The common sub-trajectory that can be N by length by similar process extends into the common sub-trajectory that length is N+1, thus excavates long common sub-trajectory, and these sub-trajectories just can as the basis of next step prediction.
(7) prediction is converged
The position of known current time all vehicles when predicting.Be set as follows target of prediction in the present embodiment: predict within the t time, move to the quantity of the vehicle at desired location P point place.
Forecasting process is as follows:
(7-1) be integrated on scheme-tree by the location point (i.e. the position of all vehicles of current time) of P point and given vehicle, merging method is as follows:
(7-11) find from a nearest cluster of P point, calculate the distance of P point to this cluster representative line segment, if this distance is greater than estimation range threshold value (usually get 30 ~ 100, get 50 in this enforcement), then prediction of failure; Otherwise, proceed as follows:
(7-12) find from its nearest cluster respectively to given vehicle location, calculate the distance between vehicle location and this cluster representative line segment, if be greater than estimation range threshold value, then give up this vehicle location point (namely not measuring this vehicle running orbit in the future in advance by existing frequent movement locus), otherwise, make vertical line by this vehicle location point to the representative line segment of nearest cluster, and be the reposition point (projected position namely in frequent movement locus) of this vehicle with intersection point.
(7-2) according to the position relationship of vehicle reposition point and residing cluster representative line segment two end points, the timestamp proportionally calculating this reposition point (supposes that two end points are A (longitude, latitude, time stamp T 1), B (longitude, latitude, time stamp T 2), vehicle reposition point is N (longitude, latitude, time stamp T 3), according to length ratio relation, T3=T1+ (AN/AB) * (T2-T1)).
(7-3) according to the time stamp T 3 of reposition point, with t (given predicted time), this vehicle is obtained when (T3+t) moment based on scheme-tree, the position P ' in frequent track
(7-4) Euclidean distance of P ' to P point is calculated, if be less than threshold value (usually to get 30 ~ 80,50 are got) in the present invention, and the vehicle reposition point N place cluster (7-2) is greater than 3 in the support value of pattern seeds, then predict that vehicle fleet size adds 1, otherwise do not add.
Above-described embodiment has been described in detail technical scheme of the present invention and beneficial effect; be understood that and the foregoing is only most preferred embodiment of the present invention; be not limited to the present invention; all make in spirit of the present invention any amendment, supplement and equivalent to replace, all should be included within protection scope of the present invention.
Claims (10)
1. the vehicle based on Euclidean space converges a Forecasting Methodology, it is characterized in that, comprises the steps:
(1) obtain the carrying running orbit of all vehicles in region to be predicted, and pre-service is carried out to each carrying running orbit;
(2) line segment cluster is carried out to the pre-processed results of described step (1) and obtain cluster result, and carry out motor pattern excavation according to described cluster result;
(3) according to position and the motor pattern Result of each vehicle of current time, the vehicle of desired location in prediction setting-up time, is arrived.
2. converge Forecasting Methodology based on the vehicle of Euclidean space as claimed in claim 1, it is characterized in that, in described step (1), as follows pre-service is carried out to each carrying running orbit:
(1-1) determine the time step point of this carrying running orbit and the position of space leaping point, and with the position of space leaping point, track is carried out to this carrying running orbit according to time step point and be separated;
(1-2) adopt Douglas-Peucker algorithm to carry out track simplification to the result after track separation and obtain pretreated carrying running orbit.
3. converge Forecasting Methodology based on the vehicle of Euclidean space as claimed in claim 2, it is characterized in that, in described step (1-1), track is carried out to carrying running orbit and be separated as follows:
(1-11) corresponding single running orbit is obtained as track separating resulting according to the position of time step point using carrying out track classification to the carrying track of each vehicle;
(1-12) for any one single running orbit, according to the position of space leaping point, division is carried out to present single running orbit and obtain some sub-running orbits as track separating resulting.
4. converge Forecasting Methodology based on the vehicle of Euclidean space as claimed in claim 3, it is characterized in that, carry out also carrying out speed judgement to described sub-running orbit after track separation obtains sub-running orbit to carrying running orbit in described step (1-1):
If the speed in described sub-running orbit between any two continuity points exceedes default threshold speed, then think that point is abnormity point below, and give up this abnormity point, obtain revising sub-running orbit as track separating resulting.
5. converge Forecasting Methodology based on the vehicle of Euclidean space as claimed in claim 2, it is characterized in that, described step (1-2) adopts the Douglas-Peucker algorithm based on high line to carry out track simplification to each track separating resulting.
6., as the vehicle based on Euclidean space in Claims 1 to 5 as described in any one converges Forecasting Methodology, it is characterized in that, described step carries out line segment cluster based on Euclidean space distance to all pre-processed results in (2), comprises the steps:
(2-1) each pre-processed results is divided into some line segments;
(2-2) for any line segment that current pre-processed results is corresponding, in the line segment that other pre-processed results are corresponding, the line segment nearest with it is determined:
If the distance between these two line segments is less than default line segment distance threshold, then these two line segments are gathered for same class;
Otherwise, using this line segment separately as a class;
(2-3) for any two classes, calculate distance therebetween, if described distance is less than default class distance threshold, then the two is merged into a class;
Otherwise, do not process;
(2-4) execution step (2-3) is returned until stop when categorical measure is constant;
(2-5) add up the line segment quantity in each classification, give up the class of the line segment amount threshold preset that line segment quantity is less than.
7. as claimed in claim 6 converge Forecasting Methodology based on the vehicle of Euclidean space, it is characterized in that, it is as follows to carry out motor pattern mining process in described step (2):
(S1) for any two classes, the distance between any two classes is calculated, and according to distance and when the annexation of the first two class middle conductor judges the proximity relations between the first two class;
(S2) build frequent pattern tree (fp tree) according to the proximity relations of all classes, in described frequent pattern tree (fp tree), root node is for preserving the list of the child node be connected with this root node, its child node and line segment cluster one_to_one corresponding; In described frequent pattern tree (fp tree), each non-root node comprises cluster and support two attributes, the class that this node of cluster attribute representation is corresponding, and Support value represents from this node to the degree of depth quantity of the vehicle operating track of the node being 1.
8. converge Forecasting Methodology based on the vehicle of Euclidean space as claimed in claim 7, it is characterized in that, described step (S1) judges that the method for the proximity relations between the first two class is specific as follows:
If the distance between the first two class is less than default splicing distance threshold, then according to the logarithm of each line segment line segment that position judgment is wherein sliceable in its each pre-processed results of originating, if sliceable logarithm is greater than default logarithmic threshold, then think that these two classes are adjacent, otherwise, non-conterminous.
9. converge Forecasting Methodology based on the vehicle of Euclidean space as claimed in claim 8, it is characterized in that, the distance between any two classes calculates by the following method:
Determine the representative line segment of each class, the distance between using the distance between the representative line segment of each class as two classes.
10. converge Forecasting Methodology based on the vehicle of Euclidean space as claimed in claim 9, it is characterized in that, proceed as follows for any vehicle in described step (3):
(3-1) position of this vehicle of current time is integrated on described frequent pattern tree (fp tree), and calculates the projected position of this vehicle of current time in frequent movement locus and corresponding timestamp according to merger result;
(3-2) this vehicle position residing after Preset Time is calculated as predicted position according to projected position in frequent movement locus of the position of this vehicle of current time and timestamp;
(3-3) distance between computational prediction position and desired location, if be less than threshold value and the Support value of the node corresponding in frequent pattern tree (fp tree) of this projected position is greater than 3, then think that this vehicle can arrive desired location through setting-up time, otherwise, think that this vehicle can not arrive desired location through setting-up time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510310640.7A CN104952248B (en) | 2015-06-08 | 2015-06-08 | A kind of vehicle convergence Forecasting Methodology based on Euclidean space |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510310640.7A CN104952248B (en) | 2015-06-08 | 2015-06-08 | A kind of vehicle convergence Forecasting Methodology based on Euclidean space |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104952248A true CN104952248A (en) | 2015-09-30 |
| CN104952248B CN104952248B (en) | 2017-09-12 |
Family
ID=54166868
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510310640.7A Active CN104952248B (en) | 2015-06-08 | 2015-06-08 | A kind of vehicle convergence Forecasting Methodology based on Euclidean space |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104952248B (en) |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106652440A (en) * | 2015-10-30 | 2017-05-10 | 杭州海康威视数字技术股份有限公司 | Method and apparatus for determining frequent activity area of vehicle |
| CN106846538A (en) * | 2015-12-04 | 2017-06-13 | 杭州海康威视数字技术股份有限公司 | Cross car record treating method and apparatus |
| CN106899306A (en) * | 2017-02-20 | 2017-06-27 | 武汉大学 | A kind of track of vehicle line data compression method of holding moving characteristic |
| CN107256631A (en) * | 2017-08-08 | 2017-10-17 | 南京英斯特网络科技有限公司 | A kind of track of vehicle data aggregate operation method |
| CN108093361A (en) * | 2017-12-08 | 2018-05-29 | 华中科技大学 | A kind of method for positioning user and system based on network signaling data analysis |
| CN108109369A (en) * | 2018-02-06 | 2018-06-01 | 深圳市物语智联科技有限公司 | A kind of vehicle in use based on driving trace and non-vehicle in use identification measure of supervision |
| CN108120991A (en) * | 2017-12-06 | 2018-06-05 | 上海评驾科技有限公司 | A kind of wheelpath optimization method |
| CN108257386A (en) * | 2016-12-29 | 2018-07-06 | 杭州海康威视数字技术股份有限公司 | Driving trace acquisition methods and device |
| CN108303100A (en) * | 2017-12-25 | 2018-07-20 | 厦门市美亚柏科信息股份有限公司 | Focus point analysis method and computer readable storage medium |
| CN109816170A (en) * | 2019-01-25 | 2019-05-28 | 湖北大学 | A kind of taxi waiting time prediction technique and system excavated based on track |
| CN110148298A (en) * | 2019-06-24 | 2019-08-20 | 重庆大学 | Private car rule travel behaviour based on motor vehicle electronic mark data finds method |
| CN111310070A (en) * | 2019-12-20 | 2020-06-19 | 东软集团股份有限公司 | Method and device for determining frequent trips, storage medium and electronic equipment |
| CN112116806A (en) * | 2020-08-12 | 2020-12-22 | 深圳技术大学 | Traffic flow characteristic extraction method and system |
| CN112652161A (en) * | 2019-10-12 | 2021-04-13 | 阿里巴巴集团控股有限公司 | Method and device for processing traffic flow path distribution information and electronic equipment |
| CN114187489A (en) * | 2021-12-14 | 2022-03-15 | 中国平安财产保险股份有限公司 | Vehicle abnormal driving risk detection method and device, electronic equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040158398A1 (en) * | 2002-12-06 | 2004-08-12 | International Business Machines Corporation | Compressing location data of moving objects |
| CN102629297A (en) * | 2012-03-06 | 2012-08-08 | 北京建筑工程学院 | Traveler activity rule analysis method based on stroke recognition |
| CN103020284A (en) * | 2012-12-28 | 2013-04-03 | 刘建勋 | Method for recommending taxi pickup point based on time-space clustering |
| CN103854472A (en) * | 2012-12-05 | 2014-06-11 | 深圳先进技术研究院 | Taxi cloud-intelligent scheduling method and system |
| KR101413505B1 (en) * | 2009-04-22 | 2014-07-01 | 인릭스, 인코퍼레이티드 | Predicting method and device of expected road traffic conditions based on historical and current data |
| CN104462190A (en) * | 2014-10-24 | 2015-03-25 | 中国电子科技集团公司第二十八研究所 | On-line position prediction method based on mass of space trajectory excavation |
-
2015
- 2015-06-08 CN CN201510310640.7A patent/CN104952248B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040158398A1 (en) * | 2002-12-06 | 2004-08-12 | International Business Machines Corporation | Compressing location data of moving objects |
| KR101413505B1 (en) * | 2009-04-22 | 2014-07-01 | 인릭스, 인코퍼레이티드 | Predicting method and device of expected road traffic conditions based on historical and current data |
| CN102629297A (en) * | 2012-03-06 | 2012-08-08 | 北京建筑工程学院 | Traveler activity rule analysis method based on stroke recognition |
| CN103854472A (en) * | 2012-12-05 | 2014-06-11 | 深圳先进技术研究院 | Taxi cloud-intelligent scheduling method and system |
| CN103020284A (en) * | 2012-12-28 | 2013-04-03 | 刘建勋 | Method for recommending taxi pickup point based on time-space clustering |
| CN104462190A (en) * | 2014-10-24 | 2015-03-25 | 中国电子科技集团公司第二十八研究所 | On-line position prediction method based on mass of space trajectory excavation |
Non-Patent Citations (5)
| Title |
|---|
| HOYOUNG JEUNG 等: ",A Hyrid Predi*cti"on Model or Moving Objects", 《ICDE》 * |
| NIKOS MAMOULIS 等: "Mining, Indexing, and Querying Historical Spatiotemporal Data", 《TENTH ACM SIGKDD INTERNATIONAL CONFERENCE ON KNOWLEDGE DISCOVERY & DATA MINING》 * |
| 张延玲 等: "移动对象子轨迹段分割与聚类算法", 《计算机工程与应用》 * |
| 王亮 等: "基于多尺度空间划分与路网建模的城市移动轨迹模式挖掘", 《自动化学报》 * |
| 郭黎敏 等: "基于路网的不确定性轨迹预测", 《计算机研究与发展》 * |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106652440A (en) * | 2015-10-30 | 2017-05-10 | 杭州海康威视数字技术股份有限公司 | Method and apparatus for determining frequent activity area of vehicle |
| CN106652440B (en) * | 2015-10-30 | 2019-05-21 | 杭州海康威视数字技术股份有限公司 | A kind of determination method and device in the frequent activities region of vehicle |
| CN106846538B (en) * | 2015-12-04 | 2019-12-03 | 杭州海康威视数字技术股份有限公司 | Cross vehicle record treating method and apparatus |
| CN106846538A (en) * | 2015-12-04 | 2017-06-13 | 杭州海康威视数字技术股份有限公司 | Cross car record treating method and apparatus |
| US10810870B2 (en) | 2015-12-04 | 2020-10-20 | Hangzhou Hikvision Digital Technology Co., Ltd. | Method of processing passage record and device |
| CN108257386A (en) * | 2016-12-29 | 2018-07-06 | 杭州海康威视数字技术股份有限公司 | Driving trace acquisition methods and device |
| CN106899306A (en) * | 2017-02-20 | 2017-06-27 | 武汉大学 | A kind of track of vehicle line data compression method of holding moving characteristic |
| CN107256631A (en) * | 2017-08-08 | 2017-10-17 | 南京英斯特网络科技有限公司 | A kind of track of vehicle data aggregate operation method |
| CN108120991A (en) * | 2017-12-06 | 2018-06-05 | 上海评驾科技有限公司 | A kind of wheelpath optimization method |
| CN108093361A (en) * | 2017-12-08 | 2018-05-29 | 华中科技大学 | A kind of method for positioning user and system based on network signaling data analysis |
| CN108093361B (en) * | 2017-12-08 | 2020-05-19 | 华中科技大学 | User positioning method and system based on network signaling data analysis |
| CN108303100A (en) * | 2017-12-25 | 2018-07-20 | 厦门市美亚柏科信息股份有限公司 | Focus point analysis method and computer readable storage medium |
| CN108109369A (en) * | 2018-02-06 | 2018-06-01 | 深圳市物语智联科技有限公司 | A kind of vehicle in use based on driving trace and non-vehicle in use identification measure of supervision |
| CN109816170A (en) * | 2019-01-25 | 2019-05-28 | 湖北大学 | A kind of taxi waiting time prediction technique and system excavated based on track |
| CN110148298A (en) * | 2019-06-24 | 2019-08-20 | 重庆大学 | Private car rule travel behaviour based on motor vehicle electronic mark data finds method |
| CN110148298B (en) * | 2019-06-24 | 2022-03-18 | 重庆大学 | Private car regular travel behavior discovery method based on motor vehicle electronic identification data |
| CN112652161A (en) * | 2019-10-12 | 2021-04-13 | 阿里巴巴集团控股有限公司 | Method and device for processing traffic flow path distribution information and electronic equipment |
| CN112652161B (en) * | 2019-10-12 | 2023-01-06 | 阿里巴巴集团控股有限公司 | Method and device for processing traffic flow path distribution information and electronic equipment |
| CN111310070A (en) * | 2019-12-20 | 2020-06-19 | 东软集团股份有限公司 | Method and device for determining frequent trips, storage medium and electronic equipment |
| CN111310070B (en) * | 2019-12-20 | 2024-03-08 | 东软集团股份有限公司 | Method and device for determining frequent trips, storage medium and electronic equipment |
| CN112116806A (en) * | 2020-08-12 | 2020-12-22 | 深圳技术大学 | Traffic flow characteristic extraction method and system |
| CN114187489A (en) * | 2021-12-14 | 2022-03-15 | 中国平安财产保险股份有限公司 | Vehicle abnormal driving risk detection method and device, electronic equipment and storage medium |
| CN114187489B (en) * | 2021-12-14 | 2024-04-30 | 中国平安财产保险股份有限公司 | Method and device for detecting abnormal driving risk of vehicle, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104952248B (en) | 2017-09-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104952248A (en) | Automobile convergence predicting method based on Euclidean space | |
| CN104966129A (en) | Method for separating vehicle running track | |
| CN112789619B (en) | Simulation scene construction method, simulation method and device | |
| CN100357987C (en) | Method for obtaining average speed of city rode traffic low region | |
| Liang et al. | Fuel-saving potentials of platooning evaluated through sparse heavy-duty vehicle position data | |
| CN106840175B (en) | Method and device for matching vehicle running track with road network | |
| CN107784084B (en) | Road network generation method and system based on vehicle positioning data | |
| CN111845773B (en) | Automatic driving vehicle micro-decision-making method based on reinforcement learning | |
| CN107958302B (en) | Empirical path planning method based on virtual topology traffic network | |
| WO2018108559A2 (en) | Storing speed information for predicting a future speed trajectory | |
| CN104951655B (en) | A kind of vehicle sport mode method for digging based on frequent pattern tree (fp tree) | |
| CN110836675B (en) | Decision tree-based automatic driving search decision method | |
| CN108489500A (en) | A kind of global path planning method and system based on Energy Consumption Economy | |
| CN106251642A (en) | A kind of public transport road based on real-time bus gps data chain speed calculation method | |
| CN112509356A (en) | Vehicle driving route generation method and system | |
| CN110232335A (en) | Driving Scene classification method and electronic equipment | |
| Necula | Dynamic traffic flow prediction based on GPS data | |
| Wei et al. | An improved road network partition algorithm for parallel microscopic traffic simulation | |
| DE102013107960B4 (en) | Method for updating a database as well as device and computer program | |
| CN105096590A (en) | Traffic information generation method and device | |
| CN103680185A (en) | Vehicle traveling road level precise division method | |
| US20200173802A1 (en) | Route parameter manager system | |
| CN111033591B (en) | Method and server device for determining the course of a road lane of a road network | |
| WO2020020621A1 (en) | Method and apparatus for range estimation for a vehicle | |
| CN113793500A (en) | Method and system for constructing computable road network of urban traffic road |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |