CN111144290A - Multi-view-based flight situation accurate sensing method - Google Patents
Multi-view-based flight situation accurate sensing method Download PDFInfo
- Publication number
- CN111144290A CN111144290A CN201911363768.4A CN201911363768A CN111144290A CN 111144290 A CN111144290 A CN 111144290A CN 201911363768 A CN201911363768 A CN 201911363768A CN 111144290 A CN111144290 A CN 111144290A
- Authority
- CN
- China
- Prior art keywords
- flight situation
- flight
- thermodynamic diagram
- time
- thermodynamic
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Remote Sensing (AREA)
- Astronomy & Astrophysics (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Radar Systems Or Details Thereof (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention relates to the technical field of air traffic management, and provides a flight situation accurate sensing method based on multiple views, which is characterized in that all track points in a certain airspace are clustered according to a set unit time interval and are mapped into a flight situation thermodynamic diagram sequence; acquiring a meteorological radar map corresponding to the flight situation thermodynamic diagram time; considering time dimension influence factors of flight situation, constructing a plurality of independent convolutional neural networks corresponding to the time dimension influence factors; training a convolution neural network by taking the flight situation thermodynamic diagram and the meteorological radar chart as input; fusing the training results to obtain a preliminary model; sensing a flight state thermodynamic diagram by using a primary model, comparing the flight state thermodynamic diagram with a corresponding actual flight state thermodynamic diagram to obtain a mean square error, and performing back propagation training by taking the mean square error as an input to obtain a final model; the final model is used for perception. The method has high sensing precision and provides a basis for flight flow allocation and airspace dynamic management; the method is simple and feasible, and is suitable for popularization and application.
Description
Technical Field
The invention relates to the technical field of air traffic management, in particular to a flight situation accurate perception method based on multiple views.
Background
At present, airspace aviation management is mainly sensed based on flight plan data, the flight plan comprises the time when an aircraft takes off, passes through key points and lands, and the flight situation is calculated through the flight plan.
The biggest problem of the traditional method is that situation perception based on a flight plan is a static perception method, and in the process of aviation operation, a plurality of peripheral interference factors are applied, so that the dynamic change factors of actual air traffic control operation cannot be reflected in the result, and finally, the perception result has great deviation.
At present, the prior art which fully considers the influence of dynamic factors on the flight situation is not seen.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, provides a flight situation accurate sensing method based on multiple views, can accurately sense the flight situation thermodynamic diagram at a specific future moment, and provides a basis for the flight situation optimization and the response plan of an airspace.
The invention adopts the following technical scheme:
a method for accurately sensing flight situation based on multiple views comprises the following steps:
s1, clustering all track points in a certain airspace within a period of time according to a set unit time interval, and mapping the track points into a flight situation thermodynamic diagram sequence according to the space density of the points;
s2, acquiring a meteorological radar map corresponding to the flight situation thermodynamic diagram time;
s3, considering time dimension influence factors of flight situation, and constructing a plurality of independent convolutional neural networks corresponding to the time dimension influence factors; training a plurality of convolution neural networks by taking the flight situation thermodynamic diagram and the meteorological radar diagram as input; fusing the training results to obtain a preliminary model;
and S4, obtaining a perceived flight state thermodynamic diagram by using the preliminary model, comparing the perceived flight state thermodynamic diagram with a corresponding actual flight state thermodynamic diagram to obtain a mean square error, taking the mean square error as an input for back propagation training, and continuously iterating the process to obtain parameters of a final model.
Further, the method further comprises:
and S5, taking the current actual flight situation thermodynamic diagram, the current meteorological radar diagram and the meteorological radar diagram at the future set time point as input, sensing the flight situation thermodynamic diagram at the future set time point by using the final model, making a corresponding plan according to the calculated flight situation thermodynamic diagram at the future set time point, and carrying out flight flow allocation or airspace dynamic allocation.
Further, in step S1, the certain airspace is national airspace or certain region of interest airspace.
Further, in step S1, clustering is performed based on the track density in the unit time, and different track densities are mapped into the flight situation thermodynamic diagram by using different colors.
Further, in step S1, all tracks are clustered from 0 point per day, national aircraft tracks are clustered in units of 30min, and mapped into a flight situation thermodynamic diagram, 48 national flight situation thermodynamic diagram time sequences are generated online each day, and the flight situation thermodynamic diagram time sequences are stored in a time line, wherein T is a time starting point for future sensing (for example, the flight situation thermodynamic diagram 2 hours after the current time is sensed by taking the current time as a sensing time starting point, and the flight situation thermodynamic diagram 10:00 earlier than the last wednesday is sensed by taking the last wednesday earlier 8:00 as a sensing time starting point).
Further, in step S2, the national weather radar view sequences [.
Further, in step S3, the time dimension influencing factors of the flight situation are adjacency, periodicity and trend, and 3 independent convolutional neural networks are constructed accordingly.
Further, each Convolutional neural network selects three Convolutional neural Networks of ResNet (Deep residual error network), DenseNet (dense Connected Convolutional network) and VGG for ensemble learning.
Further, for the convolutional neural network corresponding to the adjacency, p thermodynamic diagrams which are forward adjacent to the T moment and p meteorological radar view mixtures corresponding to the thermodynamic diagrams are selected, namely [ Fh _ T-1, Fh _ T-2,.. Fh _ T-p and Fh _ T-p ] are used as input, and the output is F _ T;
for a convolution neural network corresponding to the periodicity, selecting n thermodynamic diagrams of each week corresponding to T time and n meteorological radar view mixtures corresponding to the thermodynamic diagrams, namely [ Fh _ T-W, Fr _ T-W, Fh _ T-2W, Fr _ T-2W.. Fh _ T-n W, Fr _ T-n W ] as input and F _ p as output;
for the convolutional neural network corresponding to the trend, M thermodynamic diagrams per month corresponding to the T time and M meteorological radar view mixtures corresponding to the thermodynamic diagrams are selected, namely [ Fh _ T-1M, Fr _ T-1M, Fh _ T-2M, Fr _ T-2M.
Further, fusing results F _ c, F _ p and F _ T after training of the 3 convolutional neural networks to obtain a perception flight situation thermodynamic diagram at the moment T + 1:
F’_T+1＝W_c*F_c+W_p*F_p+W_t*F_t；
wherein, W _ c, F _ c and W _ T are all weight coefficients, and the weight coefficients are optimized through back propagation training in step S5 according to the squared difference with the flight situation thermodynamic diagram F _ T +1 at the actual time T + 1.
The invention has the beneficial effects that: according to the method, the flight situation of a certain airspace at a set time point (for example, 0.5 hour) after the current time can be accurately sensed according to the historical flight situation and the meteorological information, and a technical basis is provided for the optimization and early warning of the flight situation of the airspace; the method is simple and easy to implement, high in sensing precision and suitable for large-scale popularization and use.
Drawings
Fig. 1 is a schematic flow chart of a method for accurately sensing a flight situation based on multiple views according to an embodiment of the present invention.
Detailed Description
Specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be noted that technical features or combinations of technical features described in the following embodiments should not be considered as being isolated, and they may be combined with each other to achieve better technical effects. In the drawings of the embodiments described below, the same reference numerals appearing in the respective drawings denote the same features or components, and may be applied to different embodiments.
As shown in fig. 1, an embodiment of the present invention provides a method for accurately sensing a flight situation based on multiple views, including:
s1, clustering all track points of a certain airspace according to a set unit time interval, and mapping the track points into a flight situation thermodynamic diagram sequence;
s2, acquiring a meteorological radar map corresponding to the flight situation thermodynamic diagram time;
s3, considering time dimension influence factors of flight situation, and constructing a plurality of independent convolutional neural networks corresponding to the time dimension influence factors; training a plurality of convolution neural networks by taking the flight situation thermodynamic diagram and the meteorological radar diagram as input; fusing the training results to obtain a preliminary model;
and S4, calculating a flight state thermodynamic diagram by using the preliminary model, comparing the flight state thermodynamic diagram with a corresponding actual flight state thermodynamic diagram to obtain a mean square error, taking the mean square error as an input for back propagation training, and repeatedly iterating the process to obtain a final model and final model parameters.
And S5, taking the current actual flight situation thermodynamic diagram, the current meteorological radar diagram and the meteorological radar diagram at the future set time point as input, sensing the flight situation thermodynamic diagram at the future set time point by using the final model, making a corresponding plan according to the calculated flight situation thermodynamic diagram at the future set time point, and carrying out flight flow allocation or airspace dynamic allocation.
It should be noted that a certain airspace may be a national airspace, or may be a certain airspace of interest, such as a certain province airspace, or an airspace of several provinces; without loss of generality, in the following embodiments, national airspace is employed.
The method of the invention can accurately sense the flight situation of the set time point after the current time point, and the set time is not specially limited, for example, the flight situation sensing 2 hours after the current time point is performed in the following embodiment.
The time dimension influence factors in the method provided by the embodiment of the invention are selected from adjacency, periodicity and trend, and are the most significant influence factors obtained by regression analysis in a large number of practices.
The method of the embodiment comprises the following steps:
1. clustering all tracks from 0 point every day, clustering the national aircraft tracks by taking 30min as a unit, taking track density as a basis, representing different track densities by different colors, mapping the different track densities into a flight situation thermodynamic diagram, generating 48 national flight situation thermodynamic diagram time sequences on line every day, and storing all the thermodynamic diagram time sequences by a time line [. 9.,. Fh _ T-2, Fh _ T-1 and Fh _ T ];
2. and acquiring the view sequences of the national weather radar in the corresponding time period by matching with each flight situation thermodynamic diagram [. multidot.Fr _ T-2, Fr _ T-1 and Fr _ T ].
3. The current time is T, the flight situation of T +2h is sensed, influence factors in the aspects of adjacency, periodicity and trend 3 of the flight situation time are considered in the time dimension, 3 mutually independent integrated learning channels based on convolutional neural networks are constructed for finding complex spatial correlation inside the flight situation thermodynamic diagram, each integrated learning channel selects three convolutional neural networks of ResNet, DenseNet and VGG for integrated learning, and the integrated learning channels are different in input layer, specifically:
1) for the trend, selecting M thermodynamic diagrams per month corresponding to the T time and M meteorological radar view mixtures corresponding to the thermodynamic diagrams, namely [ Fh _ T-1M, Fr _ T-1M, Fh _ T-2M, Fr _ T-2M., [ Fh _ T-M M, Fr _ T-M ] as input and outputting as F _ c;
2) for the periodicity, selecting n thermodynamic diagrams of each week corresponding to the T time and n meteorological radar view mixtures corresponding to the thermodynamic diagrams, namely [ Fh _ T-W, Fr _ T-W, Fh _ T-2W, Fr _ T-2W.. Fh _ T-n W, Fr _ T-n W ] as input and F _ p as output;
3) for adjacency, selecting p thermodynamic diagrams which are forward adjacent to the time T and p meteorological radar view mixtures corresponding to the thermodynamic diagrams, namely [ Fh _ T-1, Fh _ T-2,.. Fh _ T-p and Fh _ T-p ] as input, and outputting F _ T;
4. fusing results F _ c, F _ p and F _ t of the three time channels after convolution ensemble learning:
F’_T+1＝W_c*F_c+W_p*F_p+W_t*F_t；
wherein, W _ c, F _ c and W _ t are all weight coefficients, and the weight coefficients are optimized through back propagation training in the step 5.
5. And calculating the mean square error of the perception thermodynamic diagram and the actual thermodynamic diagram, then carrying out back propagation training, and carrying out multiple iterations to obtain a final model and parameters.
6. And taking the current actual flight situation thermodynamic diagram, the current meteorological radar diagram and the future 2-hour meteorological radar diagram as input, sensing the future 2-hour flight situation thermodynamic diagram by using the final model, making a corresponding plan according to the future 2-hour flight situation thermodynamic diagram obtained by calculation, and making a flight flow allocation or airspace dynamic allocation.
While several embodiments of the present invention have been presented herein, it will be appreciated by those skilled in the art that changes may be made to the embodiments herein without departing from the spirit of the invention. The above examples are merely illustrative and should not be taken as limiting the scope of the invention.
Claims (10)
1. A method for accurately sensing flight situation based on multiple views is characterized by comprising the following steps:
s1, clustering all track points in a certain airspace within a period of time according to a set unit time interval, and mapping the track points into a flight situation thermodynamic diagram sequence according to the space density of the track points;
s2, acquiring a meteorological radar map corresponding to the flight situation thermodynamic diagram time;
s3, considering time dimension influence factors of flight situation, and constructing a plurality of independent convolutional neural networks corresponding to the time dimension influence factors; training a plurality of convolution neural networks by taking the flight situation thermodynamic diagram and the meteorological radar diagram as input; fusing the training results to obtain a preliminary model;
and S4, sensing the flight state thermodynamic diagram by using the preliminary model, comparing the flight state thermodynamic diagram with the corresponding actual flight state thermodynamic diagram to obtain a mean square error, performing back propagation training by taking the mean square error as an input, and performing multiple iterations to obtain parameters of the final model.
2. The method for multi-view based accurate perception of flight situation as claimed in claim 1, further comprising:
and S5, taking the current actual flight situation thermodynamic diagram, the current meteorological radar diagram and the meteorological radar diagram at the future set time point as input, sensing the flight situation thermodynamic diagram at the future set time point by using the final model, making a corresponding plan according to the calculated flight situation thermodynamic diagram at the future set time point, and carrying out flight flow allocation or airspace dynamic allocation.
3. The method for accurately sensing flight situation based on multiple views according to claim 1, wherein in step S1, the certain airspace is national airspace or regional airspace of interest.
4. The method for accurately sensing the flight situation based on multiple views as claimed in claim 1, wherein in step S1, clustering is performed based on the track density in unit time, and different track densities are mapped into the flight situation thermodynamic diagram by using different colors.
5. The method for accurately sensing the flight situation based on multiple views as claimed in any one of claims 1 to 4, wherein in step S1, all tracks are clustered by 30min from 0 point every day, and mapped into flight situation thermodynamic diagrams, 48 national flight situation thermodynamic diagrams are generated online every day, and the flight situation thermodynamic diagrams time series [. multidot.,. Fh _ T-2, Fh _ T-1, Fh _ T ], where T is a time starting point for future sensing, are stored in a time line.
6. The method for accurately sensing flight situation based on multiple views as claimed in claim 5, wherein in step S2, the sequences of views of the national weather radar [.
7. The method for accurately sensing flight situation based on multiple views as claimed in claim 1, wherein in step S3, the time dimension influence factors of the flight situation are adjacency, periodicity and trend, and 3 independent convolutional neural networks are constructed accordingly.
8. The method for accurately sensing flight situation based on multiple views according to claim 7, wherein each convolutional neural network selects three convolutional neural networks of ResNet, DenseNet and VGG for ensemble learning.
9. A method for the accurate perception of a multi-view based flight situation according to claim 7 or 8,
for the convolutional neural network corresponding to the adjacency, p thermodynamic diagrams which are forward adjacent to the T moment and p meteorological radar view mixtures corresponding to the thermodynamic diagrams are selected, namely [ Fh _ T-1, Fh _ T-2,. Fh _ T-p and Fh _ T-p ] are used as input, and the output is F _ T;
for a convolution neural network corresponding to the periodicity, selecting n thermodynamic diagrams of each week corresponding to T time and n meteorological radar view mixtures corresponding to the thermodynamic diagrams, namely [ Fh _ T-W, Fr _ T-W, Fh _ T-2W, Fr _ T-2W.. Fh _ T-n W, Fr _ T-n W ] as input and F _ p as output;
for the convolutional neural network corresponding to the trend, M thermodynamic diagrams per month corresponding to the T time and M meteorological radar view mixtures corresponding to the thermodynamic diagrams are selected, namely [ Fh _ T-1M, Fr _ T-1M, Fh _ T-2M, Fr _ T-2M.
10. The method for accurately sensing the flight situation based on multiple views according to claim 7 or 8, wherein the results F _ c, F _ p and F _ T after training of 3 convolutional neural networks are fused to obtain the sensed flight situation thermodynamic diagram at the time T + 1:
F’_T+1＝W_c*F_c+W_p*F_p+W_t*F_t；
wherein, W _ c, F _ c and W _ T are all weight coefficients, and the weight coefficients are optimized through back propagation training in step S5 according to the square difference with the flight situation thermodynamic diagram F _ T +1 at the actual time T + 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911363768.4A CN111144290B (en) | 2019-12-26 | 2019-12-26 | Multi-view-based flight situation accurate sensing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911363768.4A CN111144290B (en) | 2019-12-26 | 2019-12-26 | Multi-view-based flight situation accurate sensing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111144290A true CN111144290A (en) | 2020-05-12 |
| CN111144290B CN111144290B (en) | 2023-07-25 |
Family
ID=70520266
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911363768.4A Active CN111144290B (en) | 2019-12-26 | 2019-12-26 | Multi-view-based flight situation accurate sensing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111144290B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112466103A (en) * | 2020-11-12 | 2021-03-09 | 北京航空航天大学 | Aircraft flight threat evolution early warning method, device, equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101201870A (en) * | 2007-12-20 | 2008-06-18 | 四川川大智胜软件股份有限公司 | Method for dynamic simulation of air traffic flight posture |
| JP2009229131A (en) * | 2008-03-19 | 2009-10-08 | Toshiba Corp | Radar information display device |
| CN105487409A (en) * | 2016-01-29 | 2016-04-13 | 中国航空无线电电子研究所 | Unmanned plane spatial domain comprehensive flight safety control demonstration and verification platform |
-
2019
- 2019-12-26 CN CN201911363768.4A patent/CN111144290B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101201870A (en) * | 2007-12-20 | 2008-06-18 | 四川川大智胜软件股份有限公司 | Method for dynamic simulation of air traffic flight posture |
| JP2009229131A (en) * | 2008-03-19 | 2009-10-08 | Toshiba Corp | Radar information display device |
| CN105487409A (en) * | 2016-01-29 | 2016-04-13 | 中国航空无线电电子研究所 | Unmanned plane spatial domain comprehensive flight safety control demonstration and verification platform |
Non-Patent Citations (2)
| Title |
|---|
| 李作君;薛雯;祁青青;: "基于时空联合可视化的快速态势感知方法" * |
| 王;贺筱媛;吴琳;张大永;: "面向联合作战的赛博态势关键技术研究" * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112466103A (en) * | 2020-11-12 | 2021-03-09 | 北京航空航天大学 | Aircraft flight threat evolution early warning method, device, equipment and storage medium |
| CN112466103B (en) * | 2020-11-12 | 2021-10-01 | 北京航空航天大学 | Aircraft flight threat evolution early warning method, device, equipment and storage medium |
| US11790767B2 (en) | 2020-11-12 | 2023-10-17 | Beihang University | Method, apparatus, device and storage medium for pre-warning of aircraft flight threat evolution |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111144290B (en) | 2023-07-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109191925B (en) | Multi-airspace track planning and negotiation method for four-dimensional track operation | |
| CN103678671B (en) | A kind of dynamic community detection method in social networks | |
| CN106326585B (en) | Prediction analysis method and device based on Bayesian Network Inference | |
| CN105677648B (en) | A kind of Combo discovering method and system based on label propagation algorithm | |
| CN107247938A (en) | A kind of method of high-resolution remote sensing image City Building function classification | |
| CN112327907B (en) | Unmanned aerial vehicle path planning method | |
| CN108763376A (en) | Syncretic relation path, type, the representation of knowledge learning method of entity description information | |
| CN113222031B (en) | Photolithographic hot zone detection method based on federal personalized learning | |
| CN113837191B (en) | Cross-star remote sensing image semantic segmentation method based on bidirectional unsupervised domain adaptive fusion | |
| CN110378744A (en) | Civil aviaton's frequent flight passenger value category method and system towards incomplete data system | |
| CN104732545A (en) | Texture image segmentation method combined with sparse neighbor propagation and rapid spectral clustering | |
| CN113109816B (en) | Echo block tracking method, device and storage medium of radar echo image | |
| CN102170516B (en) | Color space transition method based on fuzzy logic and neural network | |
| CN109658347A (en) | Data enhancement methods that are a kind of while generating plurality of picture style | |
| CN108241889A (en) | Learning system and learning method | |
| CN111814528B (en) | Connectivity analysis noctilucent image city grade classification method | |
| CN111144290A (en) | Multi-view-based flight situation accurate sensing method | |
| CN115659807A (en) | Method for predicting talent performance based on Bayesian optimization model fusion algorithm | |
| CN114626886A (en) | Questionnaire data analysis method and system | |
| CN113240265A (en) | Urban space division method based on multi-mode traffic data | |
| CN104537254A (en) | Fine drawing method based on social statistical data | |
| CN115797732A (en) | Image retrieval model training method and system used in open category scene | |
| CN114529096A (en) | Social network link prediction method and system based on ternary closure graph embedding | |
| CN113033653A (en) | Edge-cloud collaborative deep neural network model training method | |
| CN111798275A (en) | Domestic flight price prediction method |
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 |