CN117332207A - Long-time track prediction method, medium and device based on deep learning - Google Patents
Long-time track prediction method, medium and device based on deep learning Download PDFInfo
- Publication number
- CN117332207A CN117332207A CN202311228341.XA CN202311228341A CN117332207A CN 117332207 A CN117332207 A CN 117332207A CN 202311228341 A CN202311228341 A CN 202311228341A CN 117332207 A CN117332207 A CN 117332207A
- Authority
- CN
- China
- Prior art keywords
- track
- long
- deep learning
- longitude
- latitude
- 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
- 238000013135 deep learning Methods 0.000 title claims abstract description 37
- 238000000034 method Methods 0.000 title claims abstract description 31
- 238000007781 pre-processing Methods 0.000 claims abstract description 10
- 238000012549 training Methods 0.000 claims abstract description 9
- 230000006870 function Effects 0.000 claims abstract description 4
- 239000013598 vector Substances 0.000 claims description 45
- 230000007774 longterm Effects 0.000 claims description 10
- 238000001914 filtration Methods 0.000 claims description 9
- 238000009499 grossing Methods 0.000 claims description 9
- 230000015654 memory Effects 0.000 claims description 8
- 238000005070 sampling Methods 0.000 claims description 7
- 230000000694 effects Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 229920001690 polydopamine Polymers 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a long-time track prediction method, medium and device based on deep learning, wherein the method comprises the following steps: step one, acquiring track data; step two, preprocessing track data; thirdly, constructing a track data set by using the preprocessed track data; training a track data set by using a long-time track prediction model based on deep learning, stopping training after the loss function converges, and obtaining a weight file of the long-time track prediction model; step five, preprocessing the track data to be predicted through the step two to obtain a track sample to be predicted; and step six, performing long-time track prediction on the track sample to be predicted by using a long-time track prediction model based on deep learning and a trained weight file to obtain a predicted track. The invention has higher accuracy, wider application range and better universality.
Description
Technical Field
The invention relates to the technical field of track prediction, in particular to a long-time track prediction method, medium and device based on deep learning.
Background
Along with the vigorous development of smart cities, smart traffic and smart navigation, track prediction plays an increasingly important role. Track prediction is classified into single-step track prediction, which predicts the track state of the next sampling time using the track state of the target at the past multiple sampling times, and multi-step track prediction, which predicts the track state of the future multiple sampling times using the track state of the target at the past multiple sampling times.
Although short-time track prediction based on deep learning has been greatly improved, no deep learning technology is directly used in long-time track prediction at present, and the existing long-time track prediction method mainly comprises a method for estimating tracks at a plurality of moments in the future based on a clustering and track matching technology, and the method is not suitable for long-time prediction of complex multi-state tracks; the other is to predict the next time track based on a single time track prediction method and take the next time track as input to predict a plurality of future time tracks through iteration, and as the prediction always has deviation, the prediction error is larger and larger along with the increase of the iteration times, so that the effect of the prediction results of a plurality of future time tracks is poor.
Disclosure of Invention
The invention aims to provide a long-time track prediction method, medium and device based on deep learning, which are used for solving the problems that the current long-time track prediction method is not suitable for long-time prediction of complex multi-state tracks and the effect of the prediction result is poor.
The invention provides a long-time track prediction method based on deep learning, which comprises the following steps:
step one, acquiring track data;
step two, preprocessing track data;
thirdly, constructing a track data set by using the preprocessed track data;
training a track data set by using a long-time track prediction model based on deep learning, stopping training after the loss function converges, and obtaining a weight file of the long-time track prediction model;
step five, preprocessing the track data to be predicted through the step two to obtain a track sample to be predicted;
and step six, performing long-time track prediction on the track sample to be predicted by using a long-time track prediction model based on deep learning and a trained weight file to obtain a predicted track.
Further, in the second step, preprocessing the track data includes:
calculating the navigational speed and the course through longitude and latitude points and time;
smoothing and filtering longitude, latitude, navigational speed and course;
and forming 8-state track data by the longitude, latitude, speed and course before smoothing filtering and the longitude, latitude, speed and course after smoothing filtering.
Furthermore, in the second step, the track data with the duration not meeting the requirement needs to be removed.
Further, in the third step, constructing a track data set using the preprocessed track data includes:
setting a predicted time step N, performing single-step sliding sampling on track data by using window width 2*N, taking longitude, latitude, speed, direction, smoothed longitude, smoothed latitude, smoothed speed and smoothed direction of the 1 st to N th moments in each window as historical tracks, taking longitude and latitude of the n+1 to 2*N th moments as future tracks, and forming 1 track sample in a track data set by the historical tracks obtained by sliding for 1 time and the future tracks.
Further, in the fourth step, the long-time track prediction model based on deep learning is as follows:
the 8 inputs of the long-time track prediction model based on the deep learning are an original longitude with a size of 1 x 24, a smoothly filtered longitude, an original latitude, a smoothly filtered latitude, an original speed, a smoothly filtered speed, an original direction and a smoothly filtered direction respectively;
the 8 inputs respectively pass through the convolution layer, the lstm layer, the convolution layer, the lstm layer and the full connection layer in sequence to obtain corresponding feature vectors with the size of 1 x 64;
the feature vectors of the original longitudes and the smoothed longitudes are added to obtain longitude feature vectors respectively, the feature vectors of the original latitudes and the smoothed latitudes are added to obtain latitude feature vectors, the feature vectors of the original speeds and the smoothed speeds are added to obtain speed feature vectors, and the feature vectors of the original directions and the smoothed directions are added to obtain direction feature vectors;
and splicing the longitude feature vector, the latitude feature vector, the speed feature vector and the direction feature vector to obtain a feature vector of 1 x 96, sequentially using a convolution layer, an lstm layer and a full connection layer to obtain a feature vector with the size of 1 x 64, and predicting by using full connection with the length of 2 x 24 to obtain 24 longitude and latitude points.
The invention also provides a computer terminal storage medium which stores computer terminal executable instructions for executing the long-time track prediction method based on deep learning.
The present invention also provides a computing device comprising:
at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the deep learning based long-term track prediction method described above.
In summary, due to the adoption of the technical scheme, the beneficial effects of the invention are as follows:
1. the invention has higher accuracy. The traditional long-time track prediction recognition method needs a series of complex operations, has more manual intervention and cannot guarantee the prediction accuracy, but the method disclosed by the invention uses a deep learning-based method to fuse and extract the features of the track, and has higher prediction accuracy.
2. The invention has a wider application range. The historical track duration of different targets is different, and the corresponding prediction duration can be set according to different historical tracks, so that different track prediction requirements can be met.
3. The invention has better universality. The traditional long-time track prediction method aims at specific track data, when the difference between a new track to be predicted and an original track data set is large, the prediction error is large, the original prediction method is required to be optimized and changed to adapt to the new track data, the method is not strong in universality, and the system maintenance is difficult, and in the invention, only the encountered track is required to be added into the data set for training, the obtained model is replaced, and the maintenance is convenient.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following description will briefly describe the drawings in the embodiments, it being understood that the following drawings only illustrate some embodiments of the present invention and should not be considered as limiting the scope, and that other related drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a long-time track prediction method based on deep learning in an embodiment of the invention.
FIG. 2 is a schematic diagram of a long-term track prediction model based on deep learning in an embodiment of the invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. The components of the embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
Thus, the following detailed description of the embodiments of the invention, as presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Examples
As shown in fig. 1, the embodiment provides a long-time track prediction method based on deep learning, which includes the following steps:
step one, acquiring J pieces of track data (J is more than or equal to 100), wherein J=600 in the embodiment;
step two, preprocessing flight path data:
(1) The flight path data with the duration not meeting the requirement (less than 8 hours) need to be removed;
(2) Calculating the navigational speed and the course through longitude and latitude points and time; smoothing and filtering longitude, latitude, navigational speed and course; and forming 8-state track data by the longitude, latitude, speed and course before smoothing filtering and the longitude, latitude, speed and course after smoothing filtering.
Thirdly, constructing a track data set by using the preprocessed track data:
setting a predicted time step n=4 hours, sliding and sampling the track data every 10 minutes for 1 time by using a window width of 8 hours, taking the longitude, latitude, speed, direction, smoothed longitude, smoothed latitude, smoothed speed and smoothed direction of the 1 st 10 minutes to 24 th 10 minutes in each window as a historical track, taking the longitude and latitude of the 25 th to 48 th 10 minutes as a future track, and forming 1 track sample in the track data set by the historical track obtained by sliding for 1 time and the future track.
Training a track data set by using a long-time track prediction model based on deep learning, stopping training after the loss function converges, and obtaining a weight file of the long-time track prediction model;
as shown in fig. 2, the long-time track prediction model based on deep learning is as follows:
the 8 inputs of the long-time track prediction model based on the deep learning are an original longitude with a size of 1 x 24, a smoothly filtered longitude, an original latitude, a smoothly filtered latitude, an original speed, a smoothly filtered speed, an original direction and a smoothly filtered direction respectively;
the 8 inputs respectively pass through the convolution layer, the lstm layer, the convolution layer, the lstm layer and the full connection layer in sequence to obtain corresponding feature vectors with the size of 1 x 64;
the feature vectors of the original longitudes and the smoothed longitudes are added to obtain longitude feature vectors respectively, the feature vectors of the original latitudes and the smoothed latitudes are added to obtain latitude feature vectors, the feature vectors of the original speeds and the smoothed speeds are added to obtain speed feature vectors, and the feature vectors of the original directions and the smoothed directions are added to obtain direction feature vectors;
and splicing the longitude feature vector, the latitude feature vector, the speed feature vector and the direction feature vector to obtain a feature vector of 1 x 96, sequentially using a convolution layer, an lstm layer and a full connection layer to obtain a feature vector with the size of 1 x 64, and predicting by using full connection with the length of 2 x 24 to obtain 24 longitude and latitude points.
Step five, preprocessing the track data to be predicted through the step two to obtain a track sample to be predicted;
and step six, performing long-time track prediction on the track sample to be predicted by using a long-time track prediction model based on deep learning and a trained weight file to obtain a predicted track.
Furthermore, in some embodiments, a computer terminal storage medium is provided, storing computer terminal executable instructions for performing the deep learning based long-term track prediction method as described in the previous embodiments. Examples of the computer storage medium include magnetic storage media (e.g., floppy disks, hard disks, etc.), optical recording media (e.g., CD-ROMs, DVDs, etc.), or memories such as memory cards, ROMs, or RAMs, etc. The computer storage media may also be distributed over network-connected computer systems, such as stores for application programs.
Furthermore, in some embodiments, a computing device is presented comprising: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the deep learning based long-term track prediction method as described in the previous embodiments. Examples of computing devices include PCs, tablets, smartphones, PDAs, etc.
The above description is only of the preferred embodiments of the present invention and is not intended to limit the present invention, but various modifications and variations can be made to the present invention by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (7)
1. The long-time track prediction method based on deep learning is characterized by comprising the following steps of:
step one, acquiring track data;
step two, preprocessing track data;
thirdly, constructing a track data set by using the preprocessed track data;
training a track data set by using a long-time track prediction model based on deep learning, stopping training after the loss function converges, and obtaining a weight file of the long-time track prediction model;
step five, preprocessing the track data to be predicted through the step two to obtain a track sample to be predicted;
and step six, performing long-time track prediction on the track sample to be predicted by using a long-time track prediction model based on deep learning and a trained weight file to obtain a predicted track.
2. The deep learning based long-term track prediction method of claim 1, wherein in the second step, preprocessing the track data comprises:
calculating the navigational speed and the course through longitude and latitude points and time;
smoothing and filtering longitude, latitude, navigational speed and course;
and forming 8-state track data by the longitude, latitude, speed and course before smoothing filtering and the longitude, latitude, speed and course after smoothing filtering.
3. The long-time track prediction method based on deep learning according to claim 2, wherein in the second step, track data whose duration does not meet the requirement needs to be removed.
4. The deep learning based long term track prediction method of claim 2, wherein in step three, constructing a track data set using the preprocessed track data comprises:
setting a predicted time step N, performing single-step sliding sampling on track data by using window width 2*N, taking longitude, latitude, speed, direction, smoothed longitude, smoothed latitude, smoothed speed and smoothed direction of the 1 st to N th moments in each window as historical tracks, taking longitude and latitude of the n+1 to 2*N th moments as future tracks, and forming 1 track sample in a track data set by the historical tracks obtained by sliding for 1 time and the future tracks.
5. The long-term track prediction method based on deep learning according to claim 1, wherein in the fourth step, the long-term track prediction model based on deep learning is as follows:
the 8 inputs of the long-time track prediction model based on the deep learning are an original longitude with a size of 1 x 24, a smoothly filtered longitude, an original latitude, a smoothly filtered latitude, an original speed, a smoothly filtered speed, an original direction and a smoothly filtered direction respectively;
the 8 inputs respectively pass through the convolution layer, the lstm layer, the convolution layer, the lstm layer and the full connection layer in sequence to obtain corresponding feature vectors with the size of 1 x 64;
the feature vectors of the original longitudes and the smoothed longitudes are added to obtain longitude feature vectors respectively, the feature vectors of the original latitudes and the smoothed latitudes are added to obtain latitude feature vectors, the feature vectors of the original speeds and the smoothed speeds are added to obtain speed feature vectors, and the feature vectors of the original directions and the smoothed directions are added to obtain direction feature vectors;
and splicing the longitude feature vector, the latitude feature vector, the speed feature vector and the direction feature vector to obtain a feature vector of 1 x 96, sequentially using a convolution layer, an lstm layer and a full connection layer to obtain a feature vector with the size of 1 x 64, and predicting by using full connection with the length of 2 x 24 to obtain 24 longitude and latitude points.
6. A computer terminal storage medium storing computer terminal executable instructions for performing the deep learning based long-term track prediction method of any one of claims 1-5.
7. A computing device, comprising:
at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the deep learning based long-term track prediction method of any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311228341.XA CN117332207A (en) | 2023-09-21 | 2023-09-21 | Long-time track prediction method, medium and device based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311228341.XA CN117332207A (en) | 2023-09-21 | 2023-09-21 | Long-time track prediction method, medium and device based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117332207A true CN117332207A (en) | 2024-01-02 |
Family
ID=89282170
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311228341.XA Pending CN117332207A (en) | 2023-09-21 | 2023-09-21 | Long-time track prediction method, medium and device based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117332207A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117688257A (en) * | 2024-01-29 | 2024-03-12 | 东北大学 | Long-term track prediction method for heterogeneous user behavior mode |
-
2023
- 2023-09-21 CN CN202311228341.XA patent/CN117332207A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117688257A (en) * | 2024-01-29 | 2024-03-12 | 东北大学 | Long-term track prediction method for heterogeneous user behavior mode |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110260870B (en) | Map matching method, device, equipment and medium based on hidden Markov model | |
| CN107038478B (en) | Road condition prediction method and device, computer equipment and readable medium | |
| CN109215372B (en) | Road network information updating method, device and equipment | |
| US20170316328A1 (en) | Pollution prediction | |
| CN114415628A (en) | Automatic driving test method and device, electronic equipment and storage medium | |
| CN117332207A (en) | Long-time track prediction method, medium and device based on deep learning | |
| CN111737377B (en) | Method and device for identifying drift trajectory, computing equipment and storage medium | |
| CN108120991A (en) | A kind of wheelpath optimization method | |
| CN111487624A (en) | Method and equipment for predicting rainfall capacity | |
| CN112668238B (en) | Rainfall processing method, rainfall processing device, rainfall processing equipment and storage medium | |
| CN113342597B (en) | System fault prediction method based on Gaussian mixture hidden Markov model | |
| CN113327264B (en) | Trajectory fitting method, device, equipment and storage medium | |
| CN115393745A (en) | Automatic bridge image progress identification method based on unmanned aerial vehicle and deep learning | |
| CN114061596A (en) | Automatic driving positioning method, system, test method, device and storage medium | |
| CN116861262B (en) | Perception model training method and device, electronic equipment and storage medium | |
| CN116719060B (en) | Information fusion-based method for detecting tight combination navigation faults of deep learning network | |
| CN113496305A (en) | Data processing method and device | |
| CN104121917A (en) | Method and device for automatically discovering new bridge | |
| CN114758493B (en) | Expressway traffic flow monitoring method and system based on data fusion | |
| CN116092048A (en) | Image recognition method and device suitable for vehicle-mounted equipment | |
| CN114578402A (en) | Target position information determining method and device, electronic equipment and storage medium | |
| CN114708470A (en) | Illegal picture identification method, medium and computing device | |
| US11710098B2 (en) | Process flow diagram prediction utilizing a process flow diagram embedding | |
| CN117332312A (en) | Behavior recognition method, medium and device based on deep learning | |
| CN115577511B (en) | Short-term track prediction method, device and system based on unmanned aerial vehicle motion state |
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 |