CN108256285A - Flight path exception detecting method and system based on density peaks fast search - Google Patents
Flight path exception detecting method and system based on density peaks fast search Download PDFInfo
- Publication number
- CN108256285A CN108256285A CN201810019779.XA CN201810019779A CN108256285A CN 108256285 A CN108256285 A CN 108256285A CN 201810019779 A CN201810019779 A CN 201810019779A CN 108256285 A CN108256285 A CN 108256285A
- Authority
- CN
- China
- Prior art keywords
- density
- data
- flight
- analysis
- dimension
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/284—Relational databases
- G06F16/285—Clustering or classification
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16Z—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS, NOT OTHERWISE PROVIDED FOR
- G16Z99/00—Subject matter not provided for in other main groups of this subclass
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2216/00—Indexing scheme relating to additional aspects of information retrieval not explicitly covered by G06F16/00 and subgroups
- G06F2216/03—Data mining
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A kind of the flight path exception detecting method and system of improved density peaks fast search, after being pre-processed by the flight record data of flights a large amount of to airline PCA dimension-reduction treatment is carried out using the method for principal component analysis, then density peaks method for fast searching is utilized, abnormal state of flight, the unsupervised approaches for being finally based on density are clustered and analyzed to obtain the flight of abnormal behaviour in itself or caused by pilot with detecting aircraft for analysis.The present invention forms distance and strength information by density peaks method, proposes, using distance divided by density as the decision diagram of horizontal axis, to excavate abnormal flight path.The method and the method based on DBSCAN are compared simultaneously, in the case where not needing to Study first setting, it was demonstrated that the consistency of effect.
Description
Technical field
It is specifically a kind of based on density peaks fast search the present invention relates to the technology that a kind of aeronautical data is excavated
Flight path exception detecting method and system.
Background technology
In the past few decades, airline, planemaker is by survey investigation of aviation accident, analysis, then take improvement with it is pre-
Anti- measure gradually improves aviation safety.This measure is often taken after the accident of fatal crass, for busy aviation
It is not a good behave for transport service.Further to improve aviation safety, aircraft industry is just developing actively, continuously monitors aircraft
Operation and the method for identification risk, so as to take analysis and precautionary measures before disaster generation.Airborne quick access recorder note
In-flight thousands of technical parameters is recorded, including air speed, height, longitude, latitude, the angle of climb, roll angle, engine parameters etc..Also
There is a machine external environment parameter in abundant machine, cabin in machine, pilot operator, the outer wind field, temperature etc. of machine.Most of airlines are
Through implementing flight control quality evaluation project, daily flight operation is analyzed.Previous analysis is all based on overrun testing, just
It is by checking whether specific parameter is more than certain threshold value.But this method applicability is poor, needs for different boats
Line, different aircraft models, the outfit of different equipment are adjusted, heavy workload and the specific professional domain knowledge of needs, and
And need predefined observed quantity.
In recent years, in commercial consumption, cover the municipal intelligent traffic of Train-borne recorder and sensor network, rail traffic neck
Domain, data-driven, data mining technology are widely used.And the research in terms of air transportation is relatively fewer.The exploitations such as lverson
Sensing monitoring system (Inductive Monitoring System) software passed through using semi-supervised learning method it is pretreated
Training set summary data distribution, then during system operation with real-time operation data compare detecting abnormal behaviour, if do not had
Training set, this method will fail.In terms of unsupervised learning, Budalakoti etc. is passed through using Sequence Miner algorithms
Discrete aircraft parameter monitors pilot operator.In order to efficiently use discrete and continuous parameter simultaneously, Srivastav has developed statistics
Frame discretization successive value, on this basis, Das combines centrifugal pump and successive value by kernel function, passes through a kind of support
Vector machine carries out abnormity detecting.Above method all there are one it is common the problem of be exactly need define a codes and standards, master die
Formula has the training set of label in other words, and real flight control is less likely to provide the training set of perfect standardization.
Invention content
The present invention proposes a kind of flight path of improved density peaks fast search for deficiencies of the prior art
Exception detecting method and system, since similar flight has similar on-board data pattern, by same flight different parameters
Time series be transformed into the vector of a higher dimensional space, these vectors are close in superspace.Similar flight is formed
Cluster becomes isolated point there are abnormal flight and is identified.Distance and strength information are formed by density peaks method, proposed
Using distance divided by density as the decision diagram of horizontal axis, abnormal flight path is excavated.The method and the method based on DBSCAN are carried out simultaneously
Comparison, in the case where not needing to Study first setting, it was demonstrated that the consistency of effect.
The present invention is achieved by the following technical solutions：
The present invention utilizes principal component analysis after being pre-processed by the flight record data of flights a large amount of to airline
Method carry out PCA dimension-reduction treatment, then using density peaks method for fast searching, analysis with detecting aircraft in itself or flight
Abnormal state of flight, the unsupervised approaches for being finally based on density are clustered and analyzed to obtain the boat of abnormal behaviour caused by member
Class.
The pretreatment refers to：
1. it is reinforced with the engine in aircraft QAR (Quick Access Recorder quickly access logger) data, air speed
Increase as taking off beginning, the critical data of the preceding 180s of interception time sequence as takeoff phase, sectional drawing partial parameters are enumerated
In the following table, wherein rolling, this kind of parameter of pitching are provided by multigroup sensor, therefore each parameter has more sets of data, is formed altogether
24 parameters.
The aircraft QAR data include the quantity of parameters flown, location information, air speed, aspect and climb, roll dress letter
Breath, engine working condition, control panel operation logic, machine external environment information, wind field, temperature, cabin ambient etc. in machine.
1 typical flight record data of table
2. each parameter is first normalized, it is 0 to form mean value, and variance is the vector of 1 180*1, then will
The vector of 24 parameters is put into successively in a higher dimensional space vector, so that the parameter of 200 different flights is with comparable
Property.It is X=[X to form final data matrix1,...,Xm]T, m=200, wherein：
XiRepresent all parameters of the i-th frame flight, n=180, the sampling number for representing record is 180s, j=24, and totally 24 fly
Machine recording parameters, the dimension of each vector is 4320, i.e. n*j=180*24 will be every after the data matrix transposition of 200*4320
A line normalizes again, is put into final matrix X.
Data above forms behavior (180*24), and it is sparse to be classified as 200 matrix and the distribution of the data in matrix, is
Solution high dimensional data problem encountered needs to carry out dimension-reduction treatment to it, i.e., high dimensional data is obtained one loyal can reflect
The low-dimensional of initial data inherent characteristic represents.It is main from linear discriminant analysis, canonical correlation analysis, factor analysis and Projection Pursuit
Constituent analysis, independent component analysis, the dimension reduction method based on neural network, including famous Self-organizing Maps SOM, to base
Nonlinear Dimension Reduction, tensor dimensionality reduction, manifold learning and localization dimensionality reduction and semi-supervised dimensionality reduction in core.
The PCA dimension-reduction treatment is specially：By linear projection, the data of higher-dimension are mapped to table in the space of low-dimensional
Show, and it is expected that the variance of the data in the dimension projected is maximum, to use less data dimension, retain more former data point
Characteristic.It can be proved that PCA is to lose a kind of minimum linear dimensionality reduction mode of primary data information (pdi).This is by retaining low order
Principal component ignores what order components were accomplished.The present invention is calculating time complexity, linear reconstruction is missed using PCA dimension reduction methods
Difference, versatility aspect performance are good.
The cluster, i.e. density clustering method, it will cluster the point set for being defined as one group of density and connecting, then
It is clustered by constantly growing region highdensity enough, while can identify noise data.
The abnormal flight and Activity recognition, i.e., after some flight data is marked as noise spot, to its data
Particular concern is carried out, from the visualization figure of key parameter, the curve of abnormal flight is marked, its behavior is analyzed.
The present invention relates to a kind of system for realizing the above method, including：Data preprocessing module, is gathered at dimension-reduction treatment module
Class processing module and Analysis of Policy Making module, wherein：Preprocessing module is connected with dimension-reduction treatment module and transmits original higher-dimension square
Battle array information, dimension-reduction treatment module are connected with clustering processing module and transmit the matrix information after dimensionality reduction, and clustering processing module is with determining
Plan analysis module, which is connected and transmits density and range information, Analysis of Policy Making module after clustering, to be analyzed to obtain abnormal behaviour accordingly
Flight.
Technique effect
Compared with prior art, the present invention can not need to pre-set the initial parameter of algorithm in no a priori criteria
In the case of, fast clustering analysis is carried out to a large amount of data, is airline regulatory pilot guidance, decision branch is done in optimization flight
It holds.It is it is demonstrated experimentally that consistent with the cluster DBSCAN effects based on space density.
Description of the drawings
Fig. 1 is flight path abnormity detecting overall flow schematic diagram；
Fig. 2 is initial data figure display schematic diagram；
Fig. 3 is conventional flight takeoff pattern diagram；
Fig. 4 a are contribution rate schematic diagram of the principal component number to information；
Fig. 4 b are influence schematic diagram of the dimensionality reduction to the cluster time；
Fig. 5 is the decision diagram based on density and distance；
Fig. 6 a, Fig. 6 b are the abnormal flight path dominant marker schematic diagram based on density searching algorithm；
Fig. 7 a, Fig. 7 b are selectedWhen, algorithm excavates 5 frame exception flights, and visualization mark is carried out in initial data
The Fig. 1 and its partial enlarged view shown；
Fig. 8 a, Fig. 8 b are selectedWhen, algorithm excavates 5 frame exception flights, and visualization mark is carried out in initial data
The Fig. 2 and its partial enlarged view shown；
Fig. 9 a, Fig. 9 b are selectedWhen, algorithm excavates 8 frame exception flights, and visualization mark is carried out in initial data
The Fig. 1 and its partial enlarged view shown；
Figure 10 a, Figure 10 b are selectedWhen, algorithm excavates 8 frame exception flights, is visualized in initial data
Indicate obtained Fig. 2 and its partial enlarged view.
Specific embodiment
As shown in Figure 1, the present embodiment includes the following steps：
Step 1) is to be multiplied by 180 sampled points by aircraft parameter number to be formed in pretreatment stage vector, and vector magnitude is
4320*1, the matrix size of 200 flights is 4320*200.If we, which measure 100 parameters, adds up 100s, that obtained list
A vector dimension is 10000 dimensions.Not only data volume is big for the above situation, but also distribution is sparse, is unfavorable for cluster analysis.The present embodiment
The middle expression that initial data is transformed to one group of each dimension linear independence using PCA by linear transformation ensures projecting dimension
The variance of data is maximum on degree, extracts the main feature component of data.
Become 200*88 after the data matrix dimensionality reduction of the original 200*4320 of step 2), the matrix after dimensionality reduction is carried out at core
Reason.Fast search is with finding that the clustering method (DPC) of density peaks can ignore the shape and dimension of cluster, during detecting clusters
The heart excludes abnormal point.In DPC methods, cluster centre typically is provided with two kinds of features：The low point of local density nearby is clustered round,
It is and relatively large with the distance of the point of local density bigger.
DPC methods in the present embodiment introduce two variables, local density measurement ρiWith distance δi, preferably to excavate
Cluster centre：The distance δ of dots of maximum density ii
=maxj(dij), wherein：dijIt is the Euclidean distance of point i, j, dcIt is to block distance.Local density ρiIt is equal to and the distance of point i
Less than dcThe sum of all the points j quantity.Alex Rodriguez demonstrate the d of variation in an experimentcIt can generate almost consistent
As a result.As a rule of thumb, d can be selectedc, the average quantity for making density center neighbour is about the percent of data count
1 to 2.ρ is being calculatedi, δiLater, it draws with ρiFor horizontal axis, δiDecision diagram for the longitudinal axis.Possess high density in data set
Point with big distance is likely to be cluster centre.Abnormal flight path in the present embodiment belongs to noise spot scope, in decision diagram
On, density is small, is possible excavation target apart from big point, display is patterned while data markers.
Step 3) is due to being that aviation aircraft is easiest to occur into nearly landing period (8 minutes) and takeoff phase (3 minutes)
The major accident stage, according to investigation, more than 60% airplane crash is happened at take off or land during.Lead to the original of airplane crash
Many because having, most important factor is the operation error of pilot, has about accounted for the 51% of airplane crash sum.It is in addition, severe
Weather, airplane fault and artificial destruction be also the reason of causing airplane crash.
Into nearly landing period and the analysis all fours of takeoff phase, by detecting the parameter of DME rangefinders, when distance
During 6 nautical miles of land end, start to read parameter, form data matrix, then pre-processed, dimensionality reduction, cluster analysis.The present embodiment
QAR (Quick Access Recorder) data of 200 frame flights are obtained, csv file is pre-processed, is taken off
The time series data of stage 24 parameters, 180 sampled points, vector magnitude 4320*1, the matrix size of 200 flights are
4320*200.The visualization of initial data, which is shown, sees Fig. 2.Routine is taken off flight height, air speed, the angle of climb, the pattern of roll angle
See Fig. 3, it can be seen that aircraft starts climb (1 section=1 nautical mile/hour=1.852 thousand m/h), the angle of climb in 160 section left and right
Kept for 14 degree or so.
Fig. 4 (a) is that for k eigenvalue components to the contribution rate change curve of result, Fig. 4 (b) is dimensionality reduction to cluster before taking
The influence curve of time, the present invention set contribution rate as 91%, k=88, and dimension drops to 88 from 4320, and it is complicated to reduce calculating
Degree.
Fig. 5 is calculates by density peaks fast search algorithm, with local density ρiFor horizontal axis, with distance δiIt is raw for the longitudinal axis
Into Analysis of Policy Making figure.
According to airline's needs, abnormal flight detection ratio is set, then in conjunction with curve shown in Fig. 4, Fig. 5, is calculated
To the setting value of two parameters, the present embodiment mainly has studied the operating condition of takeoff phase flight, in 2% detection ratio,
Two initial parameters that setting DBSCAN algorithm needs are set：Size of Neighborhood epsilon=55, minimum number MinPts=8.Table
1 enumerate be DBSCAN algorithms excavate 5 frame exception flight serial numbers.
The abnormal flight that table 2DBSCAN is excavated
| Abnormal flight serial number | Flag bit |
| 23 | -1 |
| 59 | -1 |
| 92 | -1 |
| 102 | -1 |
| 182 | -1 |
Gained is calculated according to the above method, with ρiFor horizontal axis, δiDecision diagram is formed for the longitudinal axis.The present embodiment combines practical need
Background is sought, excavation is noise point data, therefore improves density peaks fast search algorithm, with flight serial number horizontal axis,For
The longitudinal axis obtains Optimal Decision-making figure, and screening density itself is small, the target data of the big point of Distance Density again relatively far away from.According to right
Extremely the Stringency judged can be selectedExcavate 5 frame exception flights shown in table 2.With reference to the abnormal flight of table 1
Serial number, it can be seen that the 5 frame exception flight serial numbers detected are consistent with DBSCAN algorithms, confirmed two kinds of algorithm effects from side
The consistency of fruit.
The abnormal flight that 3 density peaks searching method of table is excavated
Work as settingWhen, excavate 8 frame exception flights, the font of overstriking and table 1 in careful deck watch 3, it can be found that
8 frame exception flights of detecting enumerate 5 frame flights of DBSCAN detections.Demonstrate the effective of density peaks fast search algorithm
Property.It is detected according to abnormal flight as a result, being marked in initial data, the flight path with special color and figure is possible different
Chang Hangban.The right side of every width figure is the partial enlargement of left figure, is conducive to specific scenario analysis.
The abnormal flight that 4 density peaks searching method of table is excavated
Fig. 8 is the figure domination label of 5 frame exception flights of DBSCAN methods detection, and Fig. 9 is that density peaks are quickly searched
The domination label of 8 frame exception flights of Suo Fangfa detections, different colours are with the abnormal flight that special marking is detection.With reference to
Table and graph data can be seen that the consistency for having confirmed improved density peaks method and DBSCAN methods.Pass through dimensionality reduction
With clustering processing, the handling result of Fig. 7 is obtained, is the partial enlargement of artwork on the right side of every width figure, different colours are with special marking
It is the abnormal flight of detection.The abnormal flight of blue circle mark, takeoff thrust acquisition is smaller, and it is slow compared with regular flight that air speed increases
It is slow, postponement of climbing.Wind field is stronger at that time on airport with reference to where Fig. 7 finds out it, considers to perform the different airport environments of aerial mission
Condition further determines that thrust that high-altitude aerodrome condition brings declines or aircraft person does not drive an airplane by Standard Operating Procedure
Cause.Totally three frame flight departure times are advanced with black for red triangular, accelerate and climb comparatively fast, carried out in 40s or so etc.
Speed turning, it may be possible to which the characteristics of airport course line causes, if it is special need to specific course line to be combined by aeronautics expert there are insecurity
Point is analyzed.
Above-mentioned specific implementation can by those skilled in the art under the premise of without departing substantially from the principle of the invention and objective with difference
Mode carry out local directed complete set to it, protection scope of the present invention is subject to claims and not by above-mentioned specific implementation institute
Limit, each implementation within its scope is by the constraint of the present invention.
Claims (9)
1. a kind of flight path exception detecting method based on density peaks fast search, which is characterized in that by big to airline
Amount flight flight record data pre-processed after using principal component analysis method progress PCA dimension-reduction treatment, then utilize
Density peaks method for fast searching, analysis and detecting aircraft abnormal state of flight in itself or caused by pilot, is finally based on
The unsupervised approaches of density are clustered and analyzed to obtain the flight of abnormal behaviour.
2. according to the method described in claim 1, it is characterized in that, the pretreatment refers to：With the engine in aircraft QAR data
Reinforcing, air speed increase as taking off beginning, the critical data of the preceding 180s of interception time sequence as takeoff phase；To crucial number
Each parameter in is first normalized, and it is 0 to form mean value, and variance is the vector of 1 180*1, then by all ginsengs
Several vectors are put into successively in a higher dimensional space vector, form data matrix.
3. according to the method described in claim 1, it is characterized in that, the PCA dimension-reduction treatment is specially：It will by linear projection
The data of higher-dimension are mapped in the space of low-dimensional and represent.
4. according to the method described in claim 1, it is characterized in that, the cluster, i.e. density clustering method, pass through by
Cluster is defined as the point set of one group of density connection, is then clustered by constantly growing region highdensity enough, simultaneously
It can identify noise data.
5. the method according to claim 1 or 4, it is characterized in that, the cluster, by introducing two variables, part is close
Degree measurement ρiWith distance δi, preferably to excavate cluster centre： The distance δ of dots of maximum density ii=maxj(dij), wherein：dijIt is the Euclidean distance of point i, j, dcIt is to cut
Turn-off from；Local density ρiIt is equal to and the distance of point i is less than dcThe sum of all the points j quantity；Select dc, make density center near
Adjacent average quantity is about 1 to 2 the percent of data count；ρ is being calculatedi, δiLater, it draws with ρiFor horizontal axis, δiIt is vertical
The decision diagram of axis.
6. according to the method described in claim 5, it is characterized in that, to possess density rise in data set from big in decision diagram
It puts as cluster centre, using the big point of density small distance as target is excavated, display is patterned while data markers.
7. according to the method described in claim 6, it is characterized in that, when excavation be noise point data when, then using improve density
Peak value fast search algorithm, i.e., with flight serial number horizontal axis,For the longitudinal axis, Optimal Decision-making figure is obtained, screening density is small, distance is big
And target data relatively far away from.
8. according to the method described in claim 1, it is characterized in that, the analysis, including into nearly landing period and takeoff phase
Analysis.
9. a kind of system for realizing any of the above-described claim the method, which is characterized in that including：Data preprocessing module,
Dimension-reduction treatment module, clustering processing module and Analysis of Policy Making module, wherein：Preprocessing module is connected simultaneously with dimension-reduction treatment module
Original higher dimensional matrix information is transmitted, dimension-reduction treatment module is connected with clustering processing module and transmits the matrix information after dimensionality reduction, gathers
Class processing module is connected with Analysis of Policy Making module and transmits density and range information, Analysis of Policy Making module after clustering to be analyzed accordingly
Obtain the flight of abnormal behaviour.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810019779.XA CN108256285A (en) | 2018-01-09 | 2018-01-09 | Flight path exception detecting method and system based on density peaks fast search |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810019779.XA CN108256285A (en) | 2018-01-09 | 2018-01-09 | Flight path exception detecting method and system based on density peaks fast search |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108256285A true CN108256285A (en) | 2018-07-06 |
Family
ID=62724806
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810019779.XA Pending CN108256285A (en) | 2018-01-09 | 2018-01-09 | Flight path exception detecting method and system based on density peaks fast search |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108256285A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109492683A (en) * | 2018-10-30 | 2019-03-19 | 国网湖南省电力有限公司 | A kind of quick online evaluation method for the wide area measurement electric power big data quality of data |
| CN109977546A (en) * | 2019-03-27 | 2019-07-05 | 北京航空航天大学 | A kind of online method for detecting abnormality of four-dimensional track based on unsupervised learning |
| CN110135451A (en) * | 2019-03-27 | 2019-08-16 | 中电莱斯信息系统有限公司 | A kind of track clustering method arriving line-segment sets distance based on point |
| CN114338348A (en) * | 2021-12-08 | 2022-04-12 | 邵也铮 | Intelligent alarm method, device, equipment and readable storage medium |
| CN115293225A (en) * | 2022-06-17 | 2022-11-04 | 重庆大学 | Pilot flat drift ejector rod cause analysis method and device |
| CN116299318A (en) * | 2023-05-18 | 2023-06-23 | 成都凯天电子股份有限公司 | Method for denoising helicopter atmospheric data based on point cloud density |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104899263A (en) * | 2015-05-22 | 2015-09-09 | 华中师范大学 | Ship trajectory mining, analysis and monitoring method based on specific region |
| CN107480647A (en) * | 2017-08-22 | 2017-12-15 | 中国人民解放军海军航空工程学院 | Based on the abnormal behaviour real-time detection method for concluding formula uniformity abnormality detection |
-
2018
- 2018-01-09 CN CN201810019779.XA patent/CN108256285A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104899263A (en) * | 2015-05-22 | 2015-09-09 | 华中师范大学 | Ship trajectory mining, analysis and monitoring method based on specific region |
| CN107480647A (en) * | 2017-08-22 | 2017-12-15 | 中国人民解放军海军航空工程学院 | Based on the abnormal behaviour real-time detection method for concluding formula uniformity abnormality detection |
Non-Patent Citations (2)
| Title |
|---|
| ALEX RODRIGUEZ AND ALESSANDRO LAIO: "Clustering by fast search and find of density peaks", 《SCIENCE》 * |
| LISHUAI LI ET AL: "Analysis of Flight Data Using Clustering Techniques for Detecting Abnormal Operations", 《JOURNAL OF AEROSPACE INFORMATION SYSTEMS》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109492683A (en) * | 2018-10-30 | 2019-03-19 | 国网湖南省电力有限公司 | A kind of quick online evaluation method for the wide area measurement electric power big data quality of data |
| CN109977546A (en) * | 2019-03-27 | 2019-07-05 | 北京航空航天大学 | A kind of online method for detecting abnormality of four-dimensional track based on unsupervised learning |
| CN110135451A (en) * | 2019-03-27 | 2019-08-16 | 中电莱斯信息系统有限公司 | A kind of track clustering method arriving line-segment sets distance based on point |
| CN110135451B (en) * | 2019-03-27 | 2020-06-26 | 中电莱斯信息系统有限公司 | Flight path clustering method based on distance from point to line segment set |
| CN109977546B (en) * | 2019-03-27 | 2020-08-04 | 北京航空航天大学 | Four-dimensional track online anomaly detection method based on unsupervised learning |
| CN114338348A (en) * | 2021-12-08 | 2022-04-12 | 邵也铮 | Intelligent alarm method, device, equipment and readable storage medium |
| CN115293225A (en) * | 2022-06-17 | 2022-11-04 | 重庆大学 | Pilot flat drift ejector rod cause analysis method and device |
| CN116299318A (en) * | 2023-05-18 | 2023-06-23 | 成都凯天电子股份有限公司 | Method for denoising helicopter atmospheric data based on point cloud density |
| CN116299318B (en) * | 2023-05-18 | 2023-08-11 | 成都凯天电子股份有限公司 | Method for denoising helicopter atmospheric data based on point cloud density |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108256285A (en) | Flight path exception detecting method and system based on density peaks fast search | |
| CN113486938B (en) | Multi-branch time convolution network-based re-landing analysis method and device | |
| Tao et al. | Low-altitude small-sized object detection using lightweight feature-enhanced convolutional neural network | |
| Mangortey et al. | Application of machine learning techniques to parameter selection for flight risk identification | |
| US8204637B1 (en) | Aircraft approach to landing analysis method | |
| CN109598815A (en) | A kind of estimation of Fuel On Board system oil consumption and health monitor method | |
| CN107194876A (en) | A kind of large-scale wild animal population quantity investigation method based on unmanned plane | |
| CN108122434A (en) | A kind of flight monitoring method and device | |
| Nimmagadda et al. | Predicting airline crash due to birds strike using machine learning | |
| Martınez et al. | Forecasting unstable approaches with boosting frameworks and lstm networks | |
| Shmelova et al. | Collective Models of the Aviation Human-Operators in Emergency for IntelligentDecisionSupportSystem. | |
| Puranik | A methodology for quantitative data-driven safety assessment for general aviation | |
| Deshmukh et al. | Temporal logic learning-based anomaly detection in metroplex terminal airspace operations | |
| Çelik et al. | Classification of manifold learning based flight fingerprints of UAVs in air traffic | |
| Fala et al. | Study on machine learning methods for general aviation flight phase identification | |
| Salgueiro et al. | Aircraft Takeoff and Landing Weight Estimation from Surveillance Data | |
| Hook et al. | How digital safety systems could revolutionize aviation safety | |
| CN115293225B (en) | Method and device for analyzing causes of pilot flat-floating ejector rod | |
| Jiang et al. | Research on the flight anomaly detection during take-off phase based on FOQA data | |
| Zixuan et al. | Research on Influencing Factors of Fuel Flow Based on QAR Data | |
| Xing et al. | Discovering latent themes in aviation safety reports using text mining and network analytics | |
| CN104157105B (en) | Runway is boarded a plane the detection warning system of state | |
| CN109131843B (en) | Long-term visual tracking active separation type undercarriage | |
| US20150269861A1 (en) | System and Method for Using Pilot Controllable Discretionary Operational Parameters to Reduce Fuel Consumption in Piloted Aircraft | |
| Clachar | Identifying and analyzing atypical flights by using supervised and unsupervised approaches |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180706 |
|
| RJ01 | Rejection of invention patent application after publication |