CN111506627A - Target behavior clustering method and system - Google Patents
Target behavior clustering method and system Download PDFInfo
- Publication number
- CN111506627A CN111506627A CN202010317163.8A CN202010317163A CN111506627A CN 111506627 A CN111506627 A CN 111506627A CN 202010317163 A CN202010317163 A CN 202010317163A CN 111506627 A CN111506627 A CN 111506627A
- Authority
- CN
- China
- Prior art keywords
- vector
- target
- value
- order difference
- behavior
- 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
Images
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
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2462—Approximate or statistical queries
-
- 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
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Probability & Statistics with Applications (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Remote Sensing (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Radar Systems Or Details Thereof (AREA)
- Image Analysis (AREA)
- Traffic Control Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a target behavior clustering method and a target behavior clustering system. The process is as follows: A. extracting a track data packet to be clustered of a target; b to D are respectively executed for each track data packet: B. analyzing vehicle motion parameters of preset dimensions from the track data packet, and carrying out vectorization processing on the vehicle motion parameters of all dimensions; C. constructing a feature set of the target behavior; D. constructing a target behavior entity description; E. constructing a similarity matrix according to the similarity among the target behavior entity descriptions; F. clustering the target behaviors according to the similarity matrix; G. and calculating the central point and the related statistics of each target behavior cluster and storing the central point and the related statistics in a correlation manner. The invention describes the behavior entity of each target by extracting the feature set of the target (single or group) behavior, and combines the characteristics of the target, so that the behavior description is more targeted, the definition of the target behavior is more definite, and the cluster division is more accurate.
Description
Technical Field
The invention relates to the field of vehicle networking, in particular to a method and a system for clustering behaviors of single or group users.
Background
For massive user trajectory data of a time sequence, the group behavior of a user is abstracted from the massive user trajectory data, and the massive data is required to be clustered when the user population behavior is applied to a real-time collision detection system. And for the user/group behaviors based on the time series trajectory data, the user/group behaviors have certain personalized characteristics, and especially under a specific scene, the user/group behaviors of different clusters have very large difference. The existing clustering method is mostly considered from the direction of universality, namely, various driving behavior data are subjected to clustering analysis as historical data of the same type, so that the group behaviors with special scenes or personalized characteristics are assimilated, the finally obtained cluster personalization is not obvious, and the target behavior analysis of the special scenes is directly lack of high-pertinence basis. The direct expression is that the collision detection with a driving behavior having a special driving habit or a vehicle condition different from that of many vehicles is erroneously determined.
Based on the above background, the clustering of behaviors by common distance measurement functions such as euclidean distance used in the conventional clustering method cannot achieve the best effect, and an ideal result cannot be achieved for the clustering of driving behaviors.
Disclosure of Invention
The invention aims to: aiming at the existing problems, a target behavior clustering method is provided. The problems that in an existing clustering method, special driving scenes are not considered in a targeted mode, clustering results are not strong in targeted mode, and class cluster division is not accurate enough are solved.
The technical scheme adopted by the invention is as follows:
a method of clustering target behaviors, comprising the steps of:
A. extracting a track data packet to be clustered of a target;
b to D are respectively executed for each track data packet:
B. analyzing vehicle motion parameters of preset dimensions from the track data packet, and carrying out vectorization processing on the vehicle motion parameters of all dimensions;
C. performing first preprocessing on the vectors of all dimensions to respectively obtain the features of all dimensions, and constructing a feature set of a target behavior according to the features of all dimensions;
D. performing second preprocessing on parameters of all dimensions in the feature set to construct a target behavior entity description;
E. constructing a similarity matrix according to the similarity among the target behavior entity descriptions;
F. clustering the target behaviors according to the similarity matrix;
G. and calculating the central point and the related statistics of each target behavior cluster, and performing associated storage on the central point and the related statistics of each target behavior cluster.
According to the clustering method, the characteristic set of the target (single or group) behaviors is extracted to describe the behavior entity of each target, and the characteristics of the targets are combined, so that the behavior description is more targeted, the definition of the target behaviors is more clear, and the cluster division is more accurate. For the application of occasions such as collision detection and the like, personalized factors such as the driving behavior, the vehicle condition and the like of a target are considered on the detection probability, so that the basis in the detection is more targeted, and the condition of false detection is not easy to occur. It should be noted that, for a single user, the parsed data packet is a historical track data packet of the user, which has a historical concept; for a user population, the trajectory data packet is a trajectory data packet of a plurality of users in the same time period, and the concept of history is not available.
Further, the step B includes:
for the targets of the user category, the vehicle motion parameters analyzed from the track data packet comprise three speed, three acceleration axes and three angular velocity axes, and the corresponding vectorization processing is carried out to obtain a speed vector, a three acceleration axis vector and a three angular velocity axis vector;
for the target of a single user, the vehicle motion parameters analyzed from the track data packet include speed, three axes of acceleration, three axes of angular velocity and alarm state, and the corresponding vectorization processing is performed to obtain a speed vector, a three axes of acceleration, a three axes of angular velocity and an alarm vector.
Further, in the step C, the first preprocessing includes:
for the targets of the user category, the first preprocessing process includes:
calculating a non-zero velocity mean value and a first-order difference minimum value according to the velocity vector;
respectively correcting three acceleration shafts, and respectively calculating the maximum value of a first-order difference absolute value and the median of the first-order difference absolute value of each vector in the acceleration three-shaft vector based on the corrected three acceleration shafts;
respectively calculating the maximum value of the first-order difference absolute value and the median of the first-order difference absolute value of each vector in the angular velocity triaxial vectors, and calculating the maximum value of the maximum values of the first-order difference absolute values of each angular velocity triaxial vector;
for the target of a single user, the first preprocessing process comprises the following steps:
calculating a non-zero velocity mean value and a first-order difference minimum value according to the velocity vector;
respectively correcting three acceleration shafts, and respectively calculating the maximum value of the first-order difference absolute value of each vector in the acceleration three-shaft vector based on the corrected three acceleration shafts;
respectively calculating the maximum value of the first-order difference absolute value of each vector in the angular velocity triaxial vectors, and calculating the maximum value of the maximum values of the first-order difference absolute values of each angular velocity triaxial vector;
the alarm vector is converted into a subcode.
Further, in the step D, the second preprocessing includes:
for the targets of the user category, the second preprocessing process includes:
respectively performing sub-coding after binning on the non-zero velocity mean value, the first-order difference absolute value median of each vector in the acceleration triaxial vector and the first-order difference absolute value median of each angular velocity triaxial vector, constructing a second mean value on the basis of the non-zero velocity mean value and the velocity vector first-order difference minimum value, and respectively performing mean processing on the maximum value of the first-order difference absolute values of each angular velocity triaxial vector and the maximum value of the first-order difference absolute values of each vector in the acceleration triaxial vector on the basis of corresponding threshold values;
for the target of the single user, the second preprocessing process includes:
and respectively carrying out average processing on the non-zero speed average value of the speed vector, the maximum value of the first-order difference absolute value of each vector in the acceleration three-axis vector and the maximum value of the first-order difference absolute value of each angular speed three-axis vector based on corresponding threshold values, and constructing a second average value based on the non-zero speed average value and the speed vector first-order difference minimum value.
Further, the step E specifically includes: and calculating the similar distance between every two target behavior entity descriptions according to the specific parameters in the target behavior entity descriptions, and constructing a similarity matrix according to each calculated similar distance.
The above selected parameters are necessarily related to the characteristics of the target behavior habit. The similarity judgment is carried out according to the specific parameters, the judgment result is more accurate, and the influence of some irrelevant parameters on the comparison result can be effectively avoided while the calculation resources are efficiently utilized.
Further, for the targets of the user category, the parameters participating in the similarity calculation include: sub-coding vectors after binning of a non-zero velocity mean value, a first-order difference absolute value median of each vector in an acceleration three-axis vector and a first-order difference absolute value median of each angular velocity three-axis vector;
for the goals of a single user, the parameters involved in the similarity calculation include: the non-zero speed mean value and the maximum value of the first-order difference absolute value of each angular speed triaxial vector are based on the mean value of the corresponding threshold value, and the second mean value and the alarm vector are sub-coded.
Further, the method for calculating the similar distance includes:
for the targets of the user species group, the similar distance calculation method comprises the following steps:
wherein, Oact-m、Oact-nRespectively two compared target entity description subsets, wherein the target entity description subsets are formed by parameters participating in similarity calculation;
for the target of a single user, the similar distance calculation method is as follows:
wherein, Om、OnRespectively two compared target entity description subsets, which are formed by parameters participating in similarity calculation, VFavgThe mean value being a non-zero mean value of the velocity based on a corresponding threshold value, Δ VFminIs a second mean value, Δ AFgAnd the maximum value of the maximum values of the first-order difference absolute values of all the angular velocity triaxial vectors is based on the mean value of the corresponding threshold, and Cr is the sub-code of the alarm vector.
Compared with common similarity comparison methods such as Euclidean distance and the like, the similarity comparison parameters and the similarity comparison method can obtain higher contour coefficients, so that the clustering effect is more prominent.
Further, the second average value is calculated by:
where Δ VFmin is the second mean, Vavg is the velocity non-zero velocity mean, and Δ Vmin is the velocity vector first order difference minimum.
Further, in the step G, the process of associating and storing the central point and the related statistics of each target behavior class cluster includes: and constructing corresponding target behavior cluster vectors according to the calculated central point and related statistics of each target behavior cluster, and writing each target behavior cluster vector into a target behavior database to finish storage. The target behaviors are stored in a vector form, the relevance between related parameters is stronger, and meanwhile, a storage network is simpler, and the query and extraction of data are facilitated.
In order to solve all or part of the problems, the invention also provides a target behavior clustering system which operates the target behavior clustering method.
In summary, due to the adoption of the technical scheme, the invention has the beneficial effects that:
1. the clustering method provided by the invention pertinently extracts the feature set of the target behavior, fully considers the personalized features of the target, and is used for describing the entity of the population or the behavior of a single user, so that the definition of the target behavior is more definite, and the described scene is more specific. Under a specific application scene, the clustering method provided by the invention is more referential to the judgment of the driving behavior.
2. Compared with the conventional judgment methods such as Euclidean distance and the like, the similarity judgment parameters and method designed by the invention are more in line with the driving behavior characteristics of various clusters, and the description of the behavior characteristics among the categories is more independent and definite.
Drawings
The invention will now be described, by way of example, with reference to the accompanying drawings, in which:
FIG. 1 is a general flow chart of the objective behavior clustering method of the present invention.
Fig. 2 is a flow chart of a user population behavior clustering method.
FIG. 3 is a flow chart of a user historical driving behavior clustering method.
Detailed Description
All of the features disclosed in this specification, or all of the steps in any method or process so disclosed, may be combined in any combination, except combinations of features and/or steps that are mutually exclusive.
Any feature disclosed in this specification (including any accompanying claims, abstract) may be replaced by alternative features serving equivalent or similar purposes, unless expressly stated otherwise. That is, unless expressly stated otherwise, each feature is only an example of a generic series of equivalent or similar features.
Example one
The embodiment discloses a target behavior clustering method, as shown in fig. 1, including the following processes:
A. and extracting a track data packet to be clustered of the target.
For the targets of the user category, the track data packets to be clustered are the track data packets of all users at the same time. For the target of a single user, the track data packets to be clustered are all historical track data packets of the user.
B to D are respectively executed for each track data packet:
B. and analyzing the vehicle motion parameters of the preset dimensions from the track data packet, and carrying out vectorization processing on the vehicle motion parameters of each dimension.
For the targets of the user category, the vehicle motion parameters analyzed from the trajectory data packet include three axes of speed, acceleration and three axes of angular velocity, and the corresponding vectorization processing is performed to obtain a speed vector, a three-axis vector of acceleration and a three-axis vector of angular velocity.
For the target of a single user, the vehicle motion parameters analyzed from the track data packet include speed, three axes of acceleration, three axes of angular velocity and alarm state, and the corresponding vectorization processing is performed to obtain a speed vector, a three axes of acceleration, a three axes of angular velocity and an alarm vector.
C. And performing first preprocessing on the vectors of all dimensions to respectively obtain the features of all dimensions, and constructing a feature set of the target behavior according to the features of all dimensions.
For the targets of the user category, the first preprocessing process includes:
calculating a non-zero velocity mean value and a first-order difference minimum value according to the velocity vector;
respectively correcting three acceleration shafts, and respectively calculating the maximum value of a first-order difference absolute value and the median of the first-order difference absolute value of each vector in the acceleration three-shaft vector based on the corrected three acceleration shafts;
and respectively calculating the maximum value of the first-order difference absolute value and the median of the first-order difference absolute value of each vector in the angular velocity triaxial vectors, and calculating the maximum value of the maximum values of the first-order difference absolute values of each angular velocity triaxial vector.
For the target of a single user, the first preprocessing process comprises the following steps:
calculating a non-zero velocity mean value and a first-order difference minimum value according to the velocity vector;
respectively correcting three acceleration shafts, and respectively calculating the maximum value of the first-order difference absolute value of each vector in the acceleration three-shaft vector based on the corrected three acceleration shafts;
respectively calculating the maximum value of the first-order difference absolute value of each vector in the angular velocity triaxial vectors, and calculating the maximum value of the maximum values of the first-order difference absolute values of each angular velocity triaxial vector;
the alarm vector is converted into a subcode.
D. And performing second preprocessing on the parameters of each dimension in the feature set to construct a target behavior entity description.
For the targets of the user category, the second preprocessing process includes:
the sub-coding after binning is carried out on the non-zero velocity mean value, the first-order difference absolute value median of each vector in the acceleration triaxial vector and the first-order difference absolute value median of each angular velocity triaxial vector respectively, a second mean value is constructed on the basis of the non-zero velocity mean value and the velocity vector first-order difference minimum value, and the mean value processing is carried out on the maximum value of the first-order difference absolute values of each angular velocity triaxial vector and the maximum value of the first-order difference absolute values of each vector in the acceleration triaxial vector respectively on the basis of corresponding threshold values.
For the target of the single user, the second preprocessing process includes:
and respectively carrying out average processing on the non-zero speed average value of the speed vector, the maximum value of the first-order difference absolute value of each vector in the acceleration three-axis vector and the maximum value of the first-order difference absolute value of each angular speed three-axis vector based on corresponding threshold values, and constructing a second average value based on the non-zero speed average value and the speed vector first-order difference minimum value.
E. And constructing a similarity matrix according to the similarity among the target behavior entity descriptions.
And calculating the similar distance between every two target behavior entity descriptions according to the specific parameters in the target behavior entity descriptions, and constructing a similarity matrix according to each calculated similar distance.
For the targets of the user category, the parameters participating in the similarity calculation include: and the sub-coding vectors are subjected to binning by using the non-zero velocity mean value, the first-order difference absolute value median of each vector in the acceleration triaxial vector and the first-order difference absolute value median of each angular velocity triaxial vector.
For the goals of a single user, the parameters involved in the similarity calculation include: the non-zero speed mean value and the maximum value of the first-order difference absolute value of each angular speed triaxial vector are based on the mean value of the corresponding threshold value, and the second mean value and the alarm vector are sub-coded.
F. And clustering the target behaviors according to the similarity matrix.
And (3) clustering the target behaviors by using a kmeans + + algorithm according to the similarity matrix, and searching the parameters through a grid to obtain a plurality of categories of target behavior clusters with optimal targets.
G. And calculating the central point and the related statistics of each target behavior cluster, and performing associated storage on the central point and the related statistics of each target behavior cluster.
And calculating the central point and the related statistics of each target behavior cluster, and constructing a corresponding target behavior cluster vector. And writing the target behavior cluster vectors into a target behavior database for subsequent use.
Example two
In this embodiment, taking the user category target as an example, a process of constructing a target behavior entity description is disclosed (i.e., the above steps B to D):
the following describes the user population behavior entity description construction process in detail.
1. Obtaining track data packets uploaded by each equipment terminal at the same time, respectively analyzing track segments with time intervals of S seconds and lengths of N from each track data packet, and extracting speed, three axes of acceleration and three axes of angular velocity in each point data. Constituting velocity vector V, acceleration triaxial vector X, Y, Z, angular velocity triaxial vector H, T, K.
2. Calculating the characteristics of the velocity, acceleration and angular velocity dimensions respectively:
calculating the mean value V of non-0 velocities from the velocity vector VavgFirst order difference minimum value Δ Vmin。
According to the component correction method of the triaxial acceleration coordinate system, correcting X, Y, Z to obtain Xx,Yy,Zz. Calculating XxFirst order difference X ofx', to obtain a first order difference absolute value | Xx' |, finding the maximum value Δ Xx=max(|Xx' |). The maximum value delta Y of the first-order difference absolute value of the acceleration of the other two axes is obtained by the same methody、ΔZz. In addition, a first order difference absolute value | X is obtainedxMedian of' | Δ Xx-median=median(|Xx' |) and the median delta Y of the first-order difference absolute value of the acceleration of the other two axes is calculated by the same methody-median、ΔZz-median。
The maximum values of the first-order difference absolute values delta H, delta T and delta K are obtained by taking the first-order difference absolute values | H ' |, | T ' | and | K ' | according to the angular velocity triaxial vector, and the maximum value delta A in the maximum values of the first-order difference absolute values of the angular velocity triaxial is further obtainedgMax (Δ H, Δ T, Δ K), and the median Δ a of the first order difference absolute values is foundg-median=median(|H’|,|T’|,|K’|)。
Finally forming a characteristic set F for constructing user population behaviorsact=[Vavg,ΔVmin,ΔXx,ΔYy,ΔZz,ΔXx-median,ΔYy-median,ΔZz-median,ΔAg,ΔAg-median]。
3. From feature set FactAnd constructing an entity description O of the user population behavioract=[Vbox,ΔVFmin,ΔXbox,ΔYbox,ΔZbox,ΔXFx,ΔYFy,ΔZFz,ΔAbox,ΔAFg]Wherein:
Vboxis the mean value of velocity VavgAnd (5) sub-coding the vector after the box separation. If the speed binning result has N bins, the sub-coding vector of the speed binning is a vector with the length N and is initialized to 0. VavgIf the result of the box separation is No. M, then V is setboxIs marked as 1.
ΔXbox,ΔYbox,ΔZboxAre respectively Delta Xx-median,ΔYy-median,ΔZz-medianSub-coding vectors of the binned result.
ΔAboxIs Δ Ag-medianSub-coding vectors of the binned result.
ΔAFg=ΔAg/Ag0,Ag0Is a preset threshold.
ΔXFx=ΔXx/G0,G0Is a preset threshold.
ΔYFy=ΔYy/G0,G0Is a preset threshold.
ΔZFz=ΔZz/G0,G0Is a preset threshold.
EXAMPLE III
The present embodiment discloses a process for constructing a similarity matrix according to the similarity between descriptions of target behavior entities for targets of user category groups based on the second embodiment (i.e. the step E):
for the user population as the target, the characteristics participating in the similarity calculation are entity description vector subset [ Vbox,ΔXbox,ΔYbox,ΔZbox,ΔAbox]Let the subset of the description vectors of the group entities in a track packet be Oact-m=[Vbox-m,ΔXbox-m,ΔYbox-m,ΔZbox-m,ΔAbox-m]The subset of the population entity description of one track packet is Oact-n=[Vbox-n,ΔXbox-n,ΔYbox-n,ΔZbox-n,ΔAbox-n]The distance function between them is:
finally, calculating to obtain a population behavior similarity matrix of N track data packets of all users as follows:
example four
The present embodiment discloses a process for constructing a description of a target behavior entity (i.e., the above steps B to D), taking a target of a single user as an example:
1. and analyzing track sections with the time interval of S seconds and the length of N from all the historical track data packets uploaded by the equipment terminal respectively, and extracting the speed, the three axes of acceleration, the three axes of angular velocity and the alarm state in each point data. The speed vector V, the acceleration triaxial vector X, Y, Z, the angular velocity triaxial vector H, T, K and the alarm vector C are formed.
2. Respectively calculating the characteristics of speed, acceleration, angular velocity and alarm dimension:
calculating the mean value V of non-0 velocities from the velocity vector VavgFirst order difference minimum value Δ Vmin。
According to the component correction method of the triaxial acceleration coordinate system, correcting X, Y, Z to obtain Xx,Yy,Zz. Calculating XxFirst order difference X ofx', to obtain a first order difference absolute value | Xx' |, finding the maximum value Δ Xx=max(|Xx' |). The maximum value delta Y of the first-order difference absolute value of the acceleration of the other two axes is obtained by the same methody、ΔZz。
The maximum values of the first-order difference absolute values delta H, delta T and delta K are obtained by taking the first-order difference absolute values | H ' |, | T ' | and | K ' | according to the angular velocity triaxial vector, and the maximum value delta A in the maximum values of the first-order difference absolute values of the angular velocity triaxial is further obtainedg=max(ΔH,ΔT,ΔK)。
And converting the alarm vector C into a subcode Cr.
Finally forming a characteristic set F for constructing the driving behavior of the useract=[Vavg,ΔVmin,ΔXx,ΔYy,ΔZz,ΔAg,Cr]。
3. From feature set FactBuilding an entity description O of the driving behavior of the useract=[VFavg,ΔVFmin,ΔXFx,ΔYFy,ΔZFz,ΔAFg,Cr]Wherein:
VFavg=Vavg/V0,V0is a preset threshold.
ΔAFg=ΔAg/Ag0,Ag0Is a preset threshold.
ΔXFx=ΔXx/G0,G0Is a preset threshold.
ΔYFy=ΔYy/G0,G0Is a preset threshold.
ΔZFz=ΔZz/G0,G0Is a preset threshold.
EXAMPLE five
Based on the fourth embodiment, the present embodiment discloses a process for constructing a similarity matrix according to the similarity between descriptions of target behavior entities by using a target of a single user (i.e., the step E):
the features participating in similarity calculation are entity description vector subsets VFavg,ΔVFmin,ΔAFg,Cr]Let the entity description vector subset of a trace packet be Om=[VFavg-m,ΔVFmin-m,ΔAFg-m,Crm]The entity description vector subset of the other track data packet is On=[VFavg-n,ΔVFmin-n,ΔAFg-n,Crn]The distance between them is:
finally, the historical behavior similarity matrix of a single user with N historical track data packets is obtained through calculation:
EXAMPLE seven
The embodiment discloses a process of storing the center point of the target behavior cluster and the related statistic in an associated manner (i.e. step G above).
For the user species group target, based on the second embodiment, the central point and the related statistics of each group behavior cluster are calculated, and a binned group behavior cluster vector C is composed of [ CID, count,<terminal,icount>,CVbox,ΔCXbox,ΔCYbox,ΔCZbox,ΔCAbox,Qv,Qx,Qy,Qz,Qa,QDv,QDx,QDy,QDz,QDa]. Wherein:
the CID identifies the population class.
count is the total packet number of the group track packet
< terminal, icount > is a two-tuple of N devices in the population < device number, total trace package of the device >
CVboxAll entity descriptions of group behavior for all entities belonging to this categoryboxThe bitwise and operation of the vector results.
ΔCXboxΔ X in all entity descriptions for group behavior belonging to this categoryboxThe bitwise and operation of the vector results.
ΔCYboxΔ Y in all entity descriptions for group behavior belonging to this categoryboxThe bitwise and operation of the vector results.
ΔCZboxΔ Z in all entity descriptions for group behavior belonging to this categoryboxThe bitwise and operation of the vector results.
ΔCAboxΔ A in all entity descriptions for group behavior belonging to this categoryboxThe bitwise and operation of the vector results.
QvΔ VF in all entity descriptions for group behaviors belonging to this classminBy default, its 80% quantile is taken.
QxΔ FX in all entity descriptions for behavior of a population belonging to this classxBy default, its 80% quantile is taken.
QyΔ FY in all entity descriptions for group behavior belonging to this categoryyBy default, its 80% quantile is taken.
QzΔ FZ in all entity descriptions for group behavior belonging to this categoryzBy default, its 80% quantile is taken.
QaΔ AF in all entity descriptions for group behavior belonging to this categorygBy default, its 80% quantile is taken.
QDvΔ VF in all entity descriptions for group behaviors belonging to this classminThe default is 75% quantile-25% quantile.
QDxΔ FX in all entity descriptions for behavior of a population belonging to this classxThe default is 75% quantile-25% quantile.
QDyDescribe Δ FY for all entities belonging to this category's historical behavioryThe default is 75% -25% quantile.
QDzDescribing Δ FZ for all entities belonging to this category's historical behaviorzThe default is 75% -25% quantile.
QDaΔ AF in all entity descriptions for group behavior belonging to this categorygThe default is 75% quantile-25% quantile.
And then writing the cluster vector C of the group behavior into a user group behavior database.
For the target of a single user, based on the fourth embodiment, the central point and the related statistics of each historical behavior class cluster are calculated, and a binned historical behavior class cluster vector C ═ Terminal, CID, CVF is formedavg,ΔCVFmin,ΔCAFg,CCr,Qx,Qy,Qz,QDx,QDy,QDz]
Terminal is the device number.
The CID is the device historical behavior category identification.
CVFavgDescribing VF in a historical behavioral entity for belonging to that categoryavgIs measured.
ΔCVFminFor Δ VF in historical behavioral entity descriptions belonging to that categoryminIs measured.
ΔCAFgFor Δ AF in description of historical behavioral entities belonging to this categorygIs measured.
CCr is the result of a bitwise and of Cr in the historical behavioral entity description belonging to this category.
QxFor Δ FX in historical behavioral entity description belonging to this categoryxBy default, its 80% quantile is taken.
QyFor Δ FY in historical behavioral entity description belonging to this categoryyBy default, its 80% quantile is taken.
QzFor Δ FZ in historical behavioral entity description belonging to this categoryzBy default, its 80% quantile is taken.
QDxFor Δ FX in historical behavioral entity description belonging to this categoryxThe default is 75% quantile-25% quantile.
QDyFor Δ FY in historical behavioral entity description belonging to this categoryyThe default is 75% -25% quantile.
QDzFor Δ FZ in historical behavioral entity description belonging to this categoryzThe default is 75% -25% quantile.
And writing the historical behavior cluster vector C into a user historical behavior database.
Example eight
The embodiment discloses a user population behavior clustering method, as shown in fig. 2, including the following steps:
and S001, extracting the track data packets of all the users at the same time from the historical database.
And S002, analyzing track segments with the time interval of S seconds and the length of N from each track data packet, and extracting the speed, three axes of acceleration and three axes of angular velocity in each point data packet. Constituting velocity vector V, acceleration triaxial vector X, Y, Z, angular velocity triaxial vector H, T, K.
And S003, calculating corresponding data characteristics according to the vectors.
Calculating the mean value V of non-0 velocities from the velocity vector VavgFirst order difference minimum value Δ Vmin
According to the component correction method of the triaxial acceleration coordinate system, correcting X, Y, Z to obtain Xx,Yy,Zz. Calculating XxFirst order difference | Xx' |, obtaining a first order difference absolute value, and solving a maximum value delta Xx=max(|Xx' |). Similarly, calculate Δ Yy,ΔZz. In addition, a first order difference absolute value | X is obtainedxMedian of' | Δ Xx-median=median(|Xx' |) and calculating Δ Y in the same mannery-median,ΔZz-median。
The maximum values of the first-order difference absolute values delta H, delta T and delta K are obtained by taking the first-order difference absolute values | H ' |, | T ' | and | K ' | and further the maximum value delta A of the three axes of the angular velocity according to the three-axis vector of the angular velocitygMax (Δ H, Δ T, Δ K), and the median Δ a of the first order difference absolute values is foundg-median=median(|H’|,|T’|,|K’|)。
Finally forming a characteristic set F for constructing user population behaviorsact=[Vavg,ΔVmin,ΔXx,ΔYy,ΔZz,ΔXx-median,ΔYy-median,ΔZz-median,ΔAg,ΔAg-median]。
S004. the feature set FactAnd constructing an entity description O of the user population behavioract=[Vbox,ΔVFmin,ΔXbox,ΔYbox,ΔZbox,ΔXFx,ΔYFy,ΔZFz,ΔAbox,ΔAFg]Wherein:
Vboxis the mean value of velocity VavgAnd (5) sub-coding the vector after the box separation. The speed is divided by making the speed division result have N boxesThe sub-coded vector of a bin is a vector of length N, initialized to 0. VavgIf the result of the box separation is No. M, then V is setboxIs marked as 1.
ΔVFmin=(Vavg-ΔVmin)/VavgIf V isavg=0ΔVFmin=-1。
ΔXbox,ΔYbox,ΔZboxAre respectively Delta Xx-median,ΔYy-median,ΔZz-medianSub-coding vectors of the binned result.
ΔAboxIs Δ Ag-medianSub-coding vectors of the binned result.
ΔAFg=ΔAg/Ag0,Ag0To preset the threshold, the default value is 1800/s。
ΔXFx=ΔXx/G0,G0The default value is 980mg for the preset threshold.
ΔYFy=ΔYy/G0,G0The default value is 980mg for the preset threshold.
ΔZFz=ΔZz/G0,G0The default value is 980mg for the preset threshold.
And S005, calculating the driving behavior similarity between every two track data packets of all the users according to the user-defined group behavior similarity function, and generating a user group behavior similarity matrix.
Features participating in similarity calculation are entity description vector subsets Vbox,ΔXbox,ΔYbox,ΔZbox,ΔAbox]Let the subset of the description vectors of the group entities in a track packet be Oact-m=[Vbox-m,ΔXbox-m,ΔYbox-m,ΔZbox-m,ΔAbox-m]The subset of the population entity description of one track packet is Oact-n=[Vbox-n,ΔXbox-n,ΔYbox-n,ΔZbox-n,ΔAbox-n]。
The distance between them is:
finally, calculating to obtain a population behavior similarity matrix of N track data packets of all users as follows:
and S006, clustering the population behaviors of the user by using a kmeans + + algorithm according to the population behavior similarity matrix, and searching for the reference through a grid to obtain the optimal M categories of population behavior clusters of the user.
S007, calculating the central point and the related statistics of each cluster behavior, and composing the warehousing cluster behavior cluster vector C ═ CID, count,<terminal,icount>,CVbox,ΔCXbox,ΔCYbox,ΔCZbox,ΔCAbox,Qv,Qx,Qy,Qz,Qa,QDv,QDx,QDy,QDz,QDa]。
the CID identifies the population class.
count is the total packet number of the group track packet
< terminal, icount > is a two-tuple of N devices in the population < device number, total trace package of the device >
CVboxAll entity descriptions of group behavior for all entities belonging to this categoryboxThe bitwise and operation of the vector results.
ΔCXboxΔ X in all entity descriptions for group behavior belonging to this categoryboxThe bitwise and operation of the vector results.
ΔCYboxΔ Y in all entity descriptions for group behavior belonging to this categoryboxThe bitwise and operation of the vector results.
ΔCZboxΔ Z in all entity descriptions for group behavior belonging to this categoryboxThe bitwise and operation of the vector results.
ΔCAboxΔ A in all entity descriptions for group behavior belonging to this categoryboxThe bitwise and operation of the vector results.
QvΔ VF in all entity descriptions for group behaviors belonging to this classminBy default, its 80% quantile is taken.
QxΔ FX in all entity descriptions for behavior of a population belonging to this classxBy default, its 80% quantile is taken.
QyΔ FY in all entity descriptions for group behavior belonging to this categoryyBy default, its 80% quantile is taken.
QzΔ FZ in all entity descriptions for group behavior belonging to this categoryzBy default, its 80% quantile is taken.
QaΔ AF in all entity descriptions for group behavior belonging to this categorygBy default, its 80% quantile is taken.
QDvΔ VF in all entity descriptions for group behaviors belonging to this classminThe default is 75% quantile-25% quantile.
QDxΔ FX in all entity descriptions for behavior of a population belonging to this classxThe default is 75% quantile-25% quantile.
QDyDescribe Δ FY for all entities belonging to this category's historical behavioryThe default is 75% -25% quantile.
QDzDescribing Δ FZ for all entities belonging to this category's historical behaviorzThe default is 75% -25% quantile.
QDaΔ AF in all entity descriptions for group behavior belonging to this categorygThe default is 75% quantile-25% quantile.
And S008, writing the result of the S007 into a user population behavior database.
Example nine
The embodiment discloses a user driving behavior clustering method, as shown in fig. 3, including the following steps:
and S001, extracting all track data packets of a single user from the historical database.
And S002, analyzing track sections with the time interval of S seconds and the length of N from each track data packet, and extracting the speed, the three axes of acceleration, the three axes of angular velocity and the alarm state in each point data. The speed vector V, the acceleration triaxial vector X, Y, Z, the angular velocity triaxial vector H, T, K and the alarm vector C are formed.
And S003, calculating the corresponding data characteristics of each track data packet.
Calculating the mean value V of non-0 velocities from the velocity vector VavgFirst order difference minimum value Δ Vmin
According to the component correction method of the triaxial acceleration coordinate system, correcting X, Y, Z to obtain Xx,Yy,Zz. Calculating XxFirst order difference X ofx', obtaining the maximum value of the first order difference absolute value DeltaXx=max(|Xx' |). Calculated in the same way, Δ Yy,ΔZz。
The first order difference is taken from the angular velocity three-axis vector in the same way, the maximum value delta H, delta T and delta K of the absolute value of the first order difference are solved, and the maximum value delta A of the angular velocity three-axis is further solvedg=max(ΔH,ΔT,ΔK)。
And converting the alarm vector C into a subcode Cr.
Finally forming a characteristic set F for constructing the driving behavior of the useract=[Vavg,ΔVmin,ΔXx,ΔYy,ΔZz,ΔAg,Cr]。
S004. the feature set FactBuilding an entity description O of the driving behavior of each track packet of the user historyact=[VFavg,ΔVFmin,ΔXFx,ΔYFy,ΔZFz,ΔAFg,Cr]Wherein:
VFavg=Vavg/V0,V0for the preset threshold, the default value is 15 km/h.
ΔVFmin=(Vavg-ΔVmin)/VavgIf V isavg=0ΔVFmin=-1。
ΔAFg=ΔAg/Ag0,Ag0To preset the threshold, the default value is 1800/s。
ΔXFx=ΔXx/G0,G0The default value is 980mg for the preset threshold.
ΔYFy=ΔYy/G0,G0The default value is 980mg for the preset threshold.
ΔZFz=ΔZz/G0,G0The default value is 980mg for the preset threshold.
And S005, calculating the driving behavior similarity between every two historical track data packets of the user according to the user-defined driving behavior similarity function, and generating a user driving behavior similarity matrix.
The features participating in similarity calculation are entity description vector subsets VFavg,ΔVFmin,ΔAFg,Cr]Let the entity description vector subset of a trace packet be Om=[VFavg-m,ΔVFmin-m,ΔAFg-m,Crm]The entity description vector subset of the other track data packet is On=[VFavg-n,ΔVFmin-n,ΔAFg-n,Crn]The distance between them is:
finally, the historical behavior similarity matrix of a single user with N historical track data packets is obtained through calculation:
and S006, clustering the historical behaviors of the user by using a kmeans + + algorithm according to the historical behavior similarity matrix, and searching the reference through a grid to obtain the optimal M categories of historical behavior clusters of the user.
S007, calculating the central point and the related statistics of each historical behavior cluster, and forming a warehousing historical behavior cluster vector C ═ Terminal, CID, CVFavg,ΔCVFmin,ΔCAFg,CCr,Qx,Qy,Qz,QDx,QDy,QDz]
Terminal is the device number.
The CID is the device historical behavior category identification.
CVFavgDescribing VF in a historical behavioral entity for belonging to that categoryavgIs measured.
ΔCVFminFor Δ VF in historical behavioral entity descriptions belonging to that categoryminIs measured.
ΔCAFgFor Δ AF in description of historical behavioral entities belonging to this categorygIs measured.
CCr is the result of a bitwise and of Cr in the historical behavioral entity description belonging to this category.
QxFor Δ FX in historical behavioral entity description belonging to this categoryxBy default, its 80% quantile is taken.
QyFor Δ FY in historical behavioral entity description belonging to this categoryyBy default, its 80% quantile is taken.
QzFor Δ FZ in historical behavioral entity description belonging to this categoryzBy default, its 80% quantile is taken.
QDxFor Δ FX in historical behavioral entity description belonging to this categoryxThe default is 75% quantile-25% quantile.
QDyTo belong to this categoryΔ FY in the description of historical behavioral entitiesyThe default is 75% -25% quantile.
QDzFor Δ FZ in historical behavioral entity description belonging to this categoryzThe default is 75% -25% quantile.
And S008, writing the result of the S007 into a user historical behavior database.
Example ten
The embodiment discloses a target behavior clustering system which operates the clustering method in any one of the embodiments.
The invention is not limited to the foregoing embodiments. The invention extends to any novel feature or any novel combination of features disclosed in this specification and any novel method or process steps or any novel combination of features disclosed.
Claims (10)
1. A target behavior clustering method is characterized by comprising the following steps:
A. extracting a track data packet to be clustered of a target;
b to D are respectively executed for each track data packet:
B. analyzing vehicle motion parameters of preset dimensions from the track data packet, and carrying out vectorization processing on the vehicle motion parameters of all dimensions;
C. performing first preprocessing on the vectors of all dimensions to respectively obtain the features of all dimensions, and constructing a feature set of a target behavior according to the features of all dimensions;
D. performing second preprocessing on parameters of all dimensions in the feature set to construct a target behavior entity description;
E. constructing a similarity matrix according to the similarity among the target behavior entity descriptions;
F. clustering the target behaviors according to the similarity matrix;
G. and calculating the central point and the related statistics of each target behavior cluster, and performing associated storage on the central point and the related statistics of each target behavior cluster.
2. The method for clustering target behaviors of claim 1, wherein the step B comprises:
for the targets of the user category, the vehicle motion parameters analyzed from the track data packet comprise three speed, three acceleration axes and three angular velocity axes, and the corresponding vectorization processing is carried out to obtain a speed vector, a three acceleration axis vector and a three angular velocity axis vector;
for the target of a single user, the vehicle motion parameters analyzed from the track data packet include speed, three axes of acceleration, three axes of angular velocity and alarm state, and the corresponding vectorization processing is performed to obtain a speed vector, a three axes of acceleration, a three axes of angular velocity and an alarm vector.
3. The method for clustering target behaviors as claimed in claim 2, wherein in the step C, the first preprocessing includes:
for the targets of the user category, the first preprocessing process includes:
calculating a non-zero velocity mean value and a first-order difference minimum value according to the velocity vector;
respectively correcting three acceleration shafts, and respectively calculating the maximum value of a first-order difference absolute value and the median of the first-order difference absolute value of each vector in the acceleration three-shaft vector based on the corrected three acceleration shafts;
respectively calculating the maximum value of the first-order difference absolute value and the median of the first-order difference absolute value of each vector in the angular velocity triaxial vectors, and calculating the maximum value of the maximum values of the first-order difference absolute values of each angular velocity triaxial vector;
for the target of a single user, the first preprocessing process comprises the following steps:
calculating a non-zero velocity mean value and a first-order difference minimum value according to the velocity vector;
respectively correcting three acceleration shafts, and respectively calculating the maximum value of the first-order difference absolute value of each vector in the acceleration three-shaft vector based on the corrected three acceleration shafts;
respectively calculating the maximum value of the first-order difference absolute value of each vector in the angular velocity triaxial vectors, and calculating the maximum value of the maximum values of the first-order difference absolute values of each angular velocity triaxial vector;
the alarm vector is converted into a subcode.
4. The method for clustering target behaviors as claimed in claim 3, wherein in the step D, the second preprocessing comprises:
for the targets of the user category, the second preprocessing process includes:
respectively performing sub-coding after binning on the non-zero velocity mean value, the first-order difference absolute value median of each vector in the acceleration triaxial vector and the first-order difference absolute value median of each angular velocity triaxial vector, constructing a second mean value on the basis of the non-zero velocity mean value and the velocity vector first-order difference minimum value, and respectively performing mean processing on the maximum value of the first-order difference absolute values of each angular velocity triaxial vector and the maximum value of the first-order difference absolute values of each vector in the acceleration triaxial vector on the basis of corresponding threshold values;
for the target of the single user, the second preprocessing process includes:
and respectively carrying out average processing on the non-zero speed average value of the speed vector, the maximum value of the first-order difference absolute value of each vector in the acceleration three-axis vector and the maximum value of the first-order difference absolute value of each angular speed three-axis vector based on corresponding threshold values, and constructing a second average value based on the non-zero speed average value and the speed vector first-order difference minimum value.
5. The method for clustering target behaviors according to any one of claims 1 to 4, wherein the step E specifically comprises: and calculating the similar distance between every two target behavior entity descriptions according to the specific parameters in the target behavior entity descriptions, and constructing a similarity matrix according to each calculated similar distance.
6. The method for clustering target behaviors of claim 5, wherein the parameters involved in the similarity calculation for the targets of the user classes comprise: sub-coding vectors after binning of a non-zero velocity mean value, a first-order difference absolute value median of each vector in an acceleration three-axis vector and a first-order difference absolute value median of each angular velocity three-axis vector;
for the goals of a single user, the parameters involved in the similarity calculation include: the non-zero speed mean value and the maximum value of the first-order difference absolute value of each angular speed triaxial vector are based on the mean value of the corresponding threshold value, and the second mean value and the alarm vector are sub-coded.
7. The method for clustering target behaviors as claimed in claim 6, wherein the similarity distance is calculated by:
for the targets of the user species group, the similar distance calculation method comprises the following steps:
wherein, Oact-m、Oact-nRespectively two compared target entity description subsets, wherein the target entity description subsets are formed by parameters participating in similarity calculation;
for the target of a single user, the similar distance calculation method is as follows:
wherein, Om、OnRespectively two compared target entity description subsets, which are formed by parameters participating in similarity calculation, VFavgThe mean value being a non-zero mean value of the velocity based on a corresponding threshold value, Δ VFminIs a second mean value, Δ AFgAnd the maximum value of the maximum values of the first-order difference absolute values of all the angular velocity triaxial vectors is based on the mean value of the corresponding threshold, and Cr is the sub-code of the alarm vector.
9. The method for clustering target behaviors according to claim 1, wherein in the step G, the process of storing the center points and the related statistics of each target behavior class in association comprises: and constructing corresponding target behavior cluster vectors according to the calculated central point and related statistics of each target behavior cluster, and writing each target behavior cluster vector into a target behavior database to finish storage.
10. An object behavior clustering system, characterized in that it operates the object behavior clustering method according to any one of claims 1 to 9.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010317163.8A CN111506627B (en) | 2020-04-21 | 2020-04-21 | Target behavior clustering method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010317163.8A CN111506627B (en) | 2020-04-21 | 2020-04-21 | Target behavior clustering method and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111506627A true CN111506627A (en) | 2020-08-07 |
CN111506627B CN111506627B (en) | 2023-05-30 |
Family
ID=71864895
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010317163.8A Active CN111506627B (en) | 2020-04-21 | 2020-04-21 | Target behavior clustering method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111506627B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116562598A (en) * | 2023-07-07 | 2023-08-08 | 成都花娃网络科技有限公司 | Distribution scheduling method, device and storage medium |
Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006021641A (en) * | 2004-07-08 | 2006-01-26 | Nissan Motor Co Ltd | Vehicle periphery display device and its display method |
CN103218906A (en) * | 2013-04-23 | 2013-07-24 | 中国科学院深圳先进技术研究院 | Falling data acquiring and analyzing platform |
CN103729550A (en) * | 2013-12-18 | 2014-04-16 | 河海大学 | Multi-model integrated flood forecasting method based on propagation time clustering analysis |
CN105681089A (en) * | 2016-01-26 | 2016-06-15 | 上海晶赞科技发展有限公司 | Network user behavior clustering method, device and terminal |
CN106778532A (en) * | 2016-11-28 | 2017-05-31 | 合肥工业大学 | Based on the driving posture feature sorting technique for removing differentiation size parameter |
CN106815312A (en) * | 2016-12-21 | 2017-06-09 | 东软集团股份有限公司 | A kind of driver's evaluation method and device |
CN107585164A (en) * | 2017-09-04 | 2018-01-16 | 交通运输部公路科学研究所 | A kind of method and device for the driver that classifies |
CN108280415A (en) * | 2018-01-17 | 2018-07-13 | 武汉理工大学 | Driving behavior recognition methods based on intelligent mobile terminal |
CN108470146A (en) * | 2018-02-11 | 2018-08-31 | 西南电子技术研究所(中国电子科技集团公司第十研究所) | The similar flight path recognition methods of classical flight path |
CN109101567A (en) * | 2018-07-17 | 2018-12-28 | 杭州电子科技大学 | A kind of distributed text approximate KNN semantic search calculation method |
CN109784970A (en) * | 2018-12-13 | 2019-05-21 | 交控科技股份有限公司 | It is a kind of to be ridden the service recommendation method and device of data based on AFC passenger |
CN110490264A (en) * | 2019-08-23 | 2019-11-22 | 中国民航大学 | Multidimensional distance cluster method for detecting abnormality and system based on time series |
CN110825826A (en) * | 2019-11-07 | 2020-02-21 | 深圳大学 | Clustering calculation method, device, terminal and storage medium |
-
2020
- 2020-04-21 CN CN202010317163.8A patent/CN111506627B/en active Active
Patent Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006021641A (en) * | 2004-07-08 | 2006-01-26 | Nissan Motor Co Ltd | Vehicle periphery display device and its display method |
CN103218906A (en) * | 2013-04-23 | 2013-07-24 | 中国科学院深圳先进技术研究院 | Falling data acquiring and analyzing platform |
CN103729550A (en) * | 2013-12-18 | 2014-04-16 | 河海大学 | Multi-model integrated flood forecasting method based on propagation time clustering analysis |
CN105681089A (en) * | 2016-01-26 | 2016-06-15 | 上海晶赞科技发展有限公司 | Network user behavior clustering method, device and terminal |
CN106778532A (en) * | 2016-11-28 | 2017-05-31 | 合肥工业大学 | Based on the driving posture feature sorting technique for removing differentiation size parameter |
CN106815312A (en) * | 2016-12-21 | 2017-06-09 | 东软集团股份有限公司 | A kind of driver's evaluation method and device |
CN107585164A (en) * | 2017-09-04 | 2018-01-16 | 交通运输部公路科学研究所 | A kind of method and device for the driver that classifies |
CN108280415A (en) * | 2018-01-17 | 2018-07-13 | 武汉理工大学 | Driving behavior recognition methods based on intelligent mobile terminal |
CN108470146A (en) * | 2018-02-11 | 2018-08-31 | 西南电子技术研究所(中国电子科技集团公司第十研究所) | The similar flight path recognition methods of classical flight path |
CN109101567A (en) * | 2018-07-17 | 2018-12-28 | 杭州电子科技大学 | A kind of distributed text approximate KNN semantic search calculation method |
CN109784970A (en) * | 2018-12-13 | 2019-05-21 | 交控科技股份有限公司 | It is a kind of to be ridden the service recommendation method and device of data based on AFC passenger |
CN110490264A (en) * | 2019-08-23 | 2019-11-22 | 中国民航大学 | Multidimensional distance cluster method for detecting abnormality and system based on time series |
CN110825826A (en) * | 2019-11-07 | 2020-02-21 | 深圳大学 | Clustering calculation method, device, terminal and storage medium |
Non-Patent Citations (4)
Title |
---|
THOMAS BRANDMEIER: "Radar-Based Analysis of Pedestrian Micro-Doppler Signatures Using Motion Capture Sensors" * |
张淼;刘明周;扈静;刘正琼;钱佩伦;: "基于K-均值聚类的汽车驾驶姿态偏好研究" * |
李经纬;赵治国;沈沛鸿;郭秋伊;: "驾驶风格K-means聚类与识别方法研究" * |
黄潇婷;张晓珊;赵莹;: "大陆游客境外旅游景区内时空行为模式研究――以香港海洋公园为例", 资源科学 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116562598A (en) * | 2023-07-07 | 2023-08-08 | 成都花娃网络科技有限公司 | Distribution scheduling method, device and storage medium |
CN116562598B (en) * | 2023-07-07 | 2023-09-19 | 成都花娃网络科技有限公司 | Distribution scheduling method, device and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN111506627B (en) | 2023-05-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109194612B (en) | Network attack detection method based on deep belief network and SVM | |
CN105760888B (en) | A kind of neighborhood rough set integrated learning approach based on hierarchical cluster attribute | |
CN104951764B (en) | Hot-short Activity recognition method based on secondary spectral clustering and HMM-RF mixed models | |
CN108900432A (en) | A kind of perception of content method based on network Flow Behavior | |
CN104331523B (en) | A kind of question sentence search method based on conceptual object model | |
CN109490838A (en) | A kind of Recognition Method of Radar Emitters of data base-oriented incompleteness | |
CN109218223A (en) | A kind of robustness net flow assorted method and system based on Active Learning | |
CN111506627A (en) | Target behavior clustering method and system | |
CN111311702A (en) | Image generation and identification module and method based on BlockGAN | |
CN112788038A (en) | Method for distinguishing DDoS attack and elephant flow based on PCA and random forest | |
CN111314910A (en) | Novel wireless sensor network abnormal data detection method for mapping isolation forest | |
Shafronenko et al. | Fuzzy clustering of distorted observations based on optimal expansion using partial distances | |
Ashok Kumar et al. | A transfer learning framework for traffic video using neuro-fuzzy approach | |
CN111506692B (en) | Collision detection method based on target behaviors | |
CN110769259A (en) | Image data compression method for tracking track content of video target | |
CN108664848B (en) | Image target identification method and device | |
CN113343079A (en) | Attack detection robust recommendation method based on random forest and target item identification | |
CN112383488A (en) | Content identification method suitable for encrypted and non-encrypted data streams | |
Kopp et al. | Interpreting and clustering outliers with sapling random forests | |
CN115348198B (en) | Unknown encryption protocol identification and classification method, device and medium based on feature retrieval | |
JPH10143613A (en) | Pattern recognition method | |
Li et al. | A supervised clustering algorithm for computer intrusion detection | |
CN114553790A (en) | Multi-mode feature-based small sample learning Internet of things traffic classification method and system | |
CN111814909A (en) | Information processing method based on network live broadcast and online e-commerce delivery and cloud server | |
Acevedo-Rodríguez et al. | Clustering of trajectories in video surveillance using growing neural gas |
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 |