CN104869105B - A kind of abnormality online recognition method - Google Patents
A kind of abnormality online recognition method Download PDFInfo
- Publication number
- CN104869105B CN104869105B CN201410065370.3A CN201410065370A CN104869105B CN 104869105 B CN104869105 B CN 104869105B CN 201410065370 A CN201410065370 A CN 201410065370A CN 104869105 B CN104869105 B CN 104869105B
- Authority
- CN
- China
- Prior art keywords
- data element
- data
- threshold value
- outlier factor
- boundary
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
A kind of abnormality online recognition method, the potential abnormal point in High Dimensional Data Streams is detected for real-time online.By analyzing the data characteristic of data flow, propose to obtain the Outlier factor value in data flow corresponding to each data with based on the method for angular distribution.With reference to the needs of real-time monitoring data flow, propose that establishing the small-scale dataflow-style based on proper set, boundary set calculates collection, accelerates the arithmetic speed of abnormality online recognition method with this.For the concept drift problem of high amount of traffic, propose to establish the real-time update mechanism of proper set, boundary set, detection accuracy of the abnormality online recognition method in higher dimensional space is ensured with this.Using the method for the present invention, the consumption to time and physical store can be not only greatly reduced, but also can be accurate, in real time the potential abnormal point in on-line checking higher-dimension high amount of traffic, to realize that the assessment of the real-time online of data flow creates condition, so as to enhance the stability of big data application system.
Description
Technical field
The present invention relates to technologies such as data mining, outlier detections, more particularly to a kind of abnormality online recognition side
Method.
Background technology
Outlier detection is one of most important technical method in Data Mining.With the continuous hair of science and technology
Many practical applications such as exhibition, such as e-commerce, network flow monitoring, wireless communication, logistics transportation can all produce sequential, magnanimity
, and vertiginous infinite data flow.Under normal circumstances, mass data flow has the feature such as higher-dimension and concept drift.It is logical
Often, these features greatly hinder the abnormality detection in data flow.Therefore, how from mass data realize to it is dangerous because
Effective excavate of element is a very important problem.
Since the research of outlier detection is risen, some famous research institutions both domestic and external and academic unit are all at this
Numerous studies work has been carried out in a field, and acquires a great achievement.It is summed up mainly by three kinds of method for detecting abnormality, difference
It is to be based on statistics, based on distance, and the method for detecting abnormality based on density.Method for detecting abnormality based on statistics generally requires
The model of given data collection, distributed constant, and expected abnormal point number.However, these parameters be all often be not easy by
Obtain.Preferable effect is had based on the Outlier Detection Algorithm of the distance data set higher to dimension, but is generally required in advance
Relevant parameter is set, and needs frequently to scan whole data set, therefore cannot meet that quick excavate of data flow requires.Base
There is dependence to arest neighbors method in most of algorithm of density, most typically improved with the thought of index data structure
Algorithm performance, but computation complexity is still higher.In addition, with the increase of dimension, in higher dimensional space, data will become
It is more and more sparse.In this case, almost all of data are all abnormal points.Therefore, the method based on density does not apply to yet
In the outlier detection in data flow.
As described above, the abnormality that traditional abnormal point detecting method can not all be adapted in data flow is known online
Not.Therefore there is an urgent need to provide a kind of abnormal state detection method that dynamic can be supported to update, before accuracy of detection is ensured
The consumption that can be reduced to time and physical store is put, so as to fulfill the speed processing and detection in real time to High Dimensional Data Streams.
The content of the invention
For problem present in above-mentioned background, the present invention provides a kind of abnormality online recognition method, to solve
Traditional abnormal point detecting method is not suitable for the problem of abnormality in online recognition data flow.
The step of the technical solution adopted by the present invention, is as follows：
A kind of abnormality online recognition method, the potential abnormal point in High Dimensional Data Streams, bag are detected for real-time online
Include step：
A. the data element in real-time data collection stream, obtains the high dimensional data sample set X containing certain data element, and
High dimensional data sample set X is pre-processed；
B. each data element in set X is analyzed with the Outlier factor formula based on angular distribution, with
This obtains the Outlier factor value of each data element in set X；
C. according to the Outlier factor value of each data element, and the division of the proper set threshold value of setting, boundary set threshold value
All data elements in set X.It is that data element is included into proper set, boundary set, one kind of anomalous concentration.So as to construct
Go out initial proper set, boundary set；
D. the latest data element X (i) in gathered data stream, small-scale dataflow-style meter is established with proper set, boundary set
Calculate collection；
E. latest data element X (i) is analyzed with the Outlier factor formula based on angular distribution, is obtained with this
The Outlier factor value of the data element；
F. according to the Outlier factor value of latest data element, and the proper set threshold value of setting, boundary set threshold value will be newest
Data element X (i) includes one kind in proper set, boundary set.If the data element is abnormal point, abnormal collection is included
O, and exported as abnormal point；
G. whether detection proper set, boundary set overflow in real time.If overflowing, by proper set, boundary set is by first
It is updated into (FIFO) mode is first gone out；
H. step D is jumped to, until having detected all data elements.
Data element in the step A real-time data collection streams, and the data element collected is stored number successively
According to collection X.After the floor data in data set X reaches the upper limit, the data element in data set X is pre-processed.Pretreatment
Arranged in order including the physics to each data element or mathematical feature, professional etiquette of going forward side by side model, simplify processing.
The step B with the Outlier factor formula based on angular distribution to each data element in data set X into
Row operational analysis, the corresponding Outlier factor value of each data element in data set X is obtained with this.Based on the different of angular distribution
Constant factor formula is as follows：
VOA (p)=Var [Θapb]=MOA2(p)-(MOA1(p))2； (1)
Wherein, the VOA (p) in formula (1) represents the Outlier factor value of point P.Θ in formula (2)apbRepresent vector
With vectorThe angle of composition, point p and point a, point b inequalities, n represent the number of data element in data set X.In formula (3)
Θ2 apbRepresent vectorWith vectorBetween form square of angle, n represents the number of data element in data set X.
The step C divides number according to proper set threshold value, boundary set threshold value, and the Outlier factor value of step B acquisitions
According to each data element in collection X.By the Outlier factor value of each data element respectively with proper set threshold value, boundary set threshold
Value is compared.If the Outlier factor value is more than or equal to proper set threshold value, which is included proper set；If the exception
Factor values are less than proper set threshold value and are more than or equal to boundary set threshold value, then the data element are included boundary set；If the exception
Factor values meet to be less than proper set threshold value, and also meet to be less than boundary set threshold value, then the element are included abnormal collection.Come with this
Obtain initial normal sample collection, boundary sample collection.
The step D is that the Outlier factor value for obtaining latest data element X (i) is prepared.It is newest in gathered data stream
Data element X (i), and proper set, boundary set that the data element and step C are obtained form small-scale dataflow-style and calculate
Collection.
The latest data element X that the step E collects step D with the Outlier factor formula based on angular distribution
(i) analyzed, the Outlier factor value of data element X (i) is obtained with this.Analyze the data element X (i) and need to refer to data
Flow pattern calculates the normal point concentrated and boundary point.Outlier factor formula based on angular distribution and the angular distribution in step B are different
Constant factor formula is consistent.
The step F is with according to proper set threshold value, boundary set threshold value, and the exception of the data element X (i) of step E acquisitions
Factor values carry out the ownership of determination data element X (i).By the Outlier factor value of data element X (i) respectively with proper set threshold value, side
Boundary's collection threshold value is compared.If the Outlier factor value is more than or equal to proper set threshold value, which is included proper set；If
The data element is then included boundary set by the Outlier factor value between proper set threshold value and boundary set threshold value；If the exception
Factor values not only meet to be less than proper set threshold value, and also meet to be less than boundary set threshold value, then the data element are included exception
Collection, and exported data element X (i) as abnormal point.The real-time detection to data flow is realized through the above way.
The step G detects established proper set in real time, and whether boundary set overflows.As long as proper set or border
The one of set overflowed, then overflowed by first-in first-out (FIFO) renewal concentrated.Realized with this
Proper set, the real-time update of boundary set solves the problems, such as the concept drift of high amount of traffic, so as to ensure abnormality online recognition side
Detection accuracy of the method in higher dimensional space.
The step H is to realize that necessary condition is created in the real-time detection of abnormality in data flow.Returned to by step H
Step D, circulation step D are realized to the continuous of data flow, in real time detection to step H.
The present invention proposes a kind of brand-new abnormality online recognition method, has the following advantages：
1., can be to the high dimensional data of each in data flow member by using the Outlier factor formula based on angular distribution
Element is analyzed, and can obtain the Outlier factor value of each high dimensional data element exactly.
2. being based on proper set by establishing, the small-scale dataflow-style of boundary set calculates collection, significantly reduces abnormal shape
Consumption of the state recognition methods to time and physical store, is effectively improved place of the abnormality recognition methods to data flow data
Manage speed.To realize that the assessment of the real-time online of data flow creates condition.
3. by the proper set of foundation, boundary set real-time update mechanism, the concept drift of high amount of traffic is efficiently solved
Problem, ensure that recognition capability of the abnormality recognition methods to potential abnormality in high amount of traffic.
Brief description of the drawings
Fig. 1 is the flow chart of implementation steps A of the present invention to step C.
Fig. 2 is the flow chart of implementation steps D of the present invention to step H.
Embodiment
In the following, the embodiment of the present invention is described further with reference to attached drawing.
As shown in Figures 1 and 2, specific implementation process of the invention and operation principle are as follows：
A. the data element in real-time data collection stream, obtains the high dimensional data sample set X containing certain data element, and
High dimensional data sample set X is pre-processed；
B. each data element in set X is analyzed with the Outlier factor formula based on angular distribution, with
This obtains the Outlier factor value of each data element in set X；
C. according to the Outlier factor value of each data element, and the division of the proper set threshold value of setting, boundary set threshold value
All data elements in set X.Data element is included to one kind of proper set, boundary set and anomalous concentration.So as to construct
Go out initial proper set, boundary set；
D. the latest data element X (i) in gathered data stream, small-scale dataflow-style meter is established with proper set, boundary set
Calculate collection；
E. latest data element X (i) is analyzed with the Outlier factor formula based on angular distribution, is obtained with this
The Outlier factor value of the data element；
F. according to the Outlier factor value of latest data element, and the proper set threshold value of setting, boundary set threshold value will be newest
Data element X (i) includes one kind in proper set, boundary set.If the data element is abnormal point, abnormal collection is included
O, and exported as abnormal point；
G. whether detection proper set, boundary set overflow in real time.If overflowing, by proper set, boundary set by first
It is updated into (FIFO) mode is first gone out；
H. step D is jumped to, until having detected all data elements.
By certain sampling period, each data on gathered data stream, acquisition contain a fixed number to step A incessantly
According to the High Dimensional Data Set X of element.Each data element in data set has higher-dimension characteristic, includes the thing of data element
Reason or mathematical feature.By the continuous acquisition of a period of time, and data set X is pre-processed after reaching the upper limit of set X.
Pretreatment includes arranging the physics or mathematical feature of each data element in order, and professional etiquette of going forward side by side model, simplifies processing.
Step B with the Outlier factor formula based on angular distribution to each data element in data acquisition system X into
Row analysis, each data element will obtain Outlier factor value corresponding with itself.Some member in data acquisition system X
Plain X (i) needs to refer to any inequality in set X two point a, b and data element X (i) when being analyzed form a vectorial angle,
And find out all this possibilities.Outlier factor formula based on angular distribution is as follows：
VOA (p)=Var [Θapb]=MOA2(p)-(MOA1(p))2； (1)
Wherein, the VOA (p) in formula (1) represents the Outlier factor value of point P.Θ in formula (2)apbRepresent vector
With vectorThe angle of composition, point p and point a, point b inequalities, n represent the number of data element in data set X.In formula (3)
Θ2 apbRepresent vectorWith vectorBetween form square of angle, n represents the number of data element in data set X.
Step C comes in data set X according to the Outlier factor value of proper set, boundary set threshold value, and data element itself
All data elements divided.By step B, each data element can obtain Outlier factor value corresponding with itself
VOA.The Outlier factor value VOA of each data element is compared with proper set threshold value VOA1, boundary set threshold value VOA2 respectively
Compared with.If Outlier factor value VOA is more than or equal to proper set threshold value VOA1, which is included proper set；If the exception
Factor values VOA is less than proper set threshold value VOA1 and is more than or equal to boundary set threshold value VOA2, then the data element is included border
Collection；If Outlier factor value VOA meets to be less than proper set threshold value VOA1, and also satisfaction is less than boundary set threshold value VOA2, then
The element includes abnormal collection, and is exported as abnormal point.By the above-mentioned means, obtain proper set, the boundary set of most initial.
To speculate that the data element X (i) of subsequent time creates condition with proper set, boundary set in the relative position of higher dimensional space.
Step D by certain sampling period continue gather subsequent time data element X (i), by data element X (i) with
Proper set, boundary set are configured to small-scale dataflow-style and calculate collection.Condition is created for continuous analysis data flow data.
Step E divides the latest data element X (i) collected with the Outlier factor formula based on angular distribution
Analysis, the Outlier factor value of data element X (i) itself is obtained with this.Analyze the data element X (i) and need to refer to small-scale number
The normal point concentrated and boundary point are calculated according to stream type.Outlier factor formula based on angular distribution and the angle point in step B
Cloth Outlier factor formula is consistent.It is limited that the data element number of concentration is calculated due to small-scale dataflow-style, and only to collection
To latest data element X (i) analyzed, therefore the present invention online recognition method have arithmetic speed as quick as thought, ensure
Data flow data is carried out in real time, continuously to assess.
Step F is determined according to the Outlier factor value of proper set threshold value, boundary set threshold value, and data element X (i) itself
The ownership of data element X (i).Herein, by the Outlier factor value VOA of data element X (i) respectively with proper set threshold value VOA1,
Boundary set threshold value VOA2 is compared.If Outlier factor value VOA is more than or equal to proper set threshold value VOA1, the data element
Include proper set；If Outlier factor value VOA is less than proper set threshold value VOA1 and is more than or equal to boundary set threshold value VOA2,
The data element includes boundary set；If Outlier factor value VOA meets to be less than proper set threshold value VOA1, and also meets to be less than side
Boundary collects threshold value VOA2, then the data element is included abnormal collection, and export data element X (i) as abnormal point.Pass through step
The processing of F, abnormality online recognition method realize the real-time detection to data flow data.
Step G detects proper set in real time, boundary set whether there is spillover.If some in proper set, boundary set
At a time overflow, then by update mechanism, it is updated by first-in first-out (FIFO).To proper set, border
The real-time update of collection is to ensure the important method of abnormality online recognition method detection accuracy.By to proper set, border
Collection, which carries out real-time update, can solve the problems, such as the concept drift of high amount of traffic.In addition, only updated just by first-in first-out
Data can be allowed to keep continuity on time and amplitude as far as possible.
Step H causes flow to jump to step D, and the company to data flow data is realized by the circulation of step D to step H
Continuous collection；Realize the foundation that small-scale dataflow-style calculates collection；Realize the uninterrupted analysis to data flow data；Realize
To the real-time update of proper set, boundary set.
Claims (9)
1. a kind of abnormality online recognition method, the potential abnormal point in High Dimensional Data Streams is detected for real-time online, including
Step：
A. the data element in real-time data collection stream, obtains the high dimensional data sample set X containing certain data element, and to height
Dimension data sample set X is pre-processed；
B. each data element in set X is analyzed with the Outlier factor formula based on angular distribution, is come with this
Obtain the Outlier factor value of each data element in set X；
C. according to the Outlier factor value of each data element, and the proper set threshold value of setting, boundary set threshold value division set X
In all data elements, be one kind that data element is included to proper set, boundary set and anomalous concentration, thus construct just
Beginning proper set, boundary set；
D. the latest data element X (i) in gathered data stream, establishes small-scale dataflow-style with proper set, boundary set and calculates collection；
E. latest data element X (i) is analyzed with the Outlier factor formula based on angular distribution, which is obtained with this
According to the Outlier factor value of element；
F. according to the Outlier factor value of latest data element, and the proper set threshold value of setting, boundary set threshold value are by latest data
Element X (i) includes one kind in proper set, boundary set, if the data element is abnormal point, is included abnormal collection O, and
Exported as abnormal point；
G. whether detection proper set, boundary set overflow in real time, if overflowing, proper set, boundary set are pressed advanced elder generation
Go out (FIFO) mode to be updated；
H. step D is jumped to, until having detected all data elements.
A kind of 2. abnormality online recognition method according to claim 1, it is characterised in that：The step A is gathered in real time
Data element in data flow, and the data element collected is stored data set X successively, when being gathered in real time in data set X
Data amount check reach the upper limit after, the data element in data set X is pre-processed, pretreatment include to each data element
Physics or mathematical feature arranged in order, professional etiquette of going forward side by side model, simplify processing.
A kind of 3. abnormality online recognition method according to claim 1, it is characterised in that：The step B is used and is based on
The Outlier factor formula of angular distribution carries out operational analysis to each data element in data set X, and data are obtained with this
Collect the corresponding Outlier factor value of each data element in X.
A kind of 4. abnormality online recognition method according to claim 1, it is characterised in that：The step C is according to normal
Collect threshold value, boundary set threshold value, and the Outlier factor value of step B acquisitions to divide each data element in data set X, will
The Outlier factor value of each data element is compared with proper set threshold value, boundary set threshold value respectively, if the Outlier factor value
More than or equal to proper set threshold value, then the data element is included proper set；If the Outlier factor value be less than proper set threshold value and
More than or equal to boundary set threshold value, then the data element is included boundary set；If the Outlier factor value meets to be less than proper set threshold value,
And also meet to be less than boundary set threshold value, then the data element is included abnormal collection, initial normal sample collection, side are obtained with this
Boundary's sample set.
A kind of 5. abnormality online recognition method according to claim 1, it is characterised in that：The step D is to obtain most
The Outlier factor value of new data element X (i) is prepared, the latest data element X (i) in gathered data stream, and by the data element
Element and proper set, the boundary set of step C acquisitions form small-scale dataflow-style and calculate and collect.
A kind of 6. abnormality online recognition method according to claim 1, it is characterised in that：The step E is used and is based on
The Outlier factor formula of angular distribution analyzes the step D latest data element X (i) collected, and data are obtained with this
The Outlier factor value of element X (i), analyzes the data element X (i) and needs to refer to normal point and border that dataflow-style calculates concentration
Point, the Outlier factor formula based on angular distribution are consistent with the angular distribution Outlier factor formula in step B.
A kind of 7. abnormality online recognition method according to claim 1, it is characterised in that：The step F is according to normal
Collect threshold value, boundary set threshold value, and the Outlier factor value of the data element X (i) of step E acquisitions comes determination data element X's (i)
Ownership, the Outlier factor value of data element X (i) is compared with proper set threshold value, boundary set threshold value respectively, if the exception because
Subvalue is more than or equal to proper set threshold value, then the data element is included proper set；If the Outlier factor value is between proper set threshold value
Between boundary set threshold value, then the data element is included boundary set；If the Outlier factor value not only meets to be less than proper set threshold
Value, and also meet to be less than boundary set threshold value, then the data element is included abnormal collection, and using data element X (i) as abnormal
Point output.
A kind of 8. abnormality online recognition method according to claim 1, it is characterised in that：The step G is detected in real time
Whether established proper set, boundary set overflow, as long as one of in proper set or boundary set is overflowed, then
The set overflowed by first-in first-out (FIFO) renewal, the real-time update of proper set, boundary set, solution are realized with this
The certainly concept drift problem of high amount of traffic, so as to ensure detection accuracy of the abnormality online recognition method in higher dimensional space.
A kind of 9. abnormality online recognition method according to claim 1, it is characterised in that：The step H is to realize number
Necessary condition is created according to the real-time detection of the upper abnormality of stream, step D, circulation step D to step H are returned to by step H, it is real
Now to the continuous of data flow, in real time detection.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410065370.3A CN104869105B (en) | 2014-02-26 | 2014-02-26 | A kind of abnormality online recognition method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410065370.3A CN104869105B (en) | 2014-02-26 | 2014-02-26 | A kind of abnormality online recognition method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104869105A CN104869105A (en) | 2015-08-26 |
| CN104869105B true CN104869105B (en) | 2018-04-24 |
Family
ID=53914630
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410065370.3A Active CN104869105B (en) | 2014-02-26 | 2014-02-26 | A kind of abnormality online recognition method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104869105B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107153931B (en) * | 2016-03-03 | 2020-11-20 | 重庆邮电大学 | Express logistics distribution abnormity detection method |
| CN110311879B (en) * | 2018-03-20 | 2022-02-22 | 重庆邮电大学 | Data flow abnormity identification method based on random projection angle distribution |
| CN110110785B (en) * | 2019-05-05 | 2021-09-14 | 北京印刷学院 | Express logistics process state detection and classification method |
| US11562237B2 (en) * | 2020-04-03 | 2023-01-24 | Micron Technology, Inc. | Processing of overwhelming stimuli in vehicle data recorders |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101848160A (en) * | 2010-05-26 | 2010-09-29 | 钱叶魁 | Method for detecting and classifying all-network flow abnormity on line |
| CN101908065A (en) * | 2010-07-27 | 2010-12-08 | 浙江大学 | On-line attribute abnormal point detecting method for supporting dynamic update |
-
2014
- 2014-02-26 CN CN201410065370.3A patent/CN104869105B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101848160A (en) * | 2010-05-26 | 2010-09-29 | 钱叶魁 | Method for detecting and classifying all-network flow abnormity on line |
| CN101908065A (en) * | 2010-07-27 | 2010-12-08 | 浙江大学 | On-line attribute abnormal point detecting method for supporting dynamic update |
Non-Patent Citations (2)
| Title |
|---|
| A near-linear Time Approximation algorithm for Angle-based Outlier Detection in High-dimensional Data;Ninh Pham 等;《Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining》;20120816;正文第2页右栏倒数第1段至正文第9页左栏倒数第1段，图1 * |
| Angle-Based Outlier Detection in High-dimensional Data;Hans-Peter Kriegel 等;《Proceedings of the 14th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining》;20080827;正文第3页左栏第3部分第1行至第9页左栏最后一段，图1至图2 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104869105A (en) | 2015-08-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104869105B (en) | A kind of abnormality online recognition method | |
| Damineli et al. | Oscillatory signatures underlie growth regimes in Arabidopsis pollen tubes: computational methods to estimate tip location, periodicity, and synchronization in growing cells | |
| Zhang et al. | An integrated algorithm to evaluate flow direction and flow accumulation in flat regions of hydrologically corrected DEMs | |
| US20140164390A1 (en) | Mining trajectory for spatial temporal analytics | |
| CN104750830B (en) | The cycle method for digging of time series data | |
| CN102636486B (en) | Method for automatically detecting three-true-leaf period of cotton | |
| CN109600345A (en) | Abnormal data flow rate testing methods and device | |
| CN104484410A (en) | Data fusion method and system applied to big data system | |
| Zygouras et al. | Towards Detection of Faulty Traffic Sensors in Real-Time. | |
| Song et al. | Real-time anomaly traffic monitoring based on dynamic k-NN cumulative-distance abnormal detection algorithm | |
| CN112132364B (en) | Photovoltaic power station power prediction method, medium and equipment influenced by cloud layer | |
| CN1996044B (en) | Canopy spatial statistics quantitative estimation method based on remote sensing image with fine spatial resolution | |
| CN110930541B (en) | Method for analyzing working condition state of agricultural machine by using GPS information | |
| KR101307337B1 (en) | System and method for Triangle Counting Sampling by using Map-Reduce | |
| Wang et al. | An integrated method for calculating DEM-based RUSLE LS | |
| CN118097961A (en) | Traffic flow data analysis method and device, electronic equipment and storage medium | |
| Pięta et al. | A model of a system for stream data storage and analysis dedicated to sensor networks of embankment monitoring | |
| CN110647935B (en) | Method and device for predicting tree growth trend in power transmission line area | |
| CN106209404A (en) | Analyzing abnormal network flow method and system | |
| CN104598600B (en) | A kind of parallel analysis of digital terrain optimization method based on distributed memory | |
| CN110413622A (en) | A kind of data processing method based on electric power big data platform | |
| JP2012073149A (en) | Wake display device | |
| CN105260249A (en) | Method for extracting calculation intensity features of spatial calculation domain | |
| CN102737095B (en) | Recommendation device, recommendation system, recommendation method, and program | |
| CN109272057A (en) | A kind of clustering method, equipment and the system of damage to crops caused by thunder ice damage |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |