LSTM-based track prediction method
The invention discloses a track prediction method based on LSTM (Long Short Term Memory). The track prediction method comprises the following steps: step 1, obtaining and preprocessing a track sequence data set; wherein the acquisition and preprocessing of the track sequence data set aims to process...
Saved in:
Main Authors | , , , , , |
---|---|
Format | Patent |
Language | Chinese English |
Published |
17.10.2023
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | The invention discloses a track prediction method based on LSTM (Long Short Term Memory). The track prediction method comprises the following steps: step 1, obtaining and preprocessing a track sequence data set; wherein the acquisition and preprocessing of the track sequence data set aims to process acquired ADS-B original data into a required stable multivariable time sequence data set with uniform length; step 2, constructing an LSTM track prediction model, wherein the construction of the LSTM track prediction model aims at performing training learning on the acquired ADS-B data through an LSTM model to realize track prediction; 3, inputting a track to be detected into the model to obtain a predicted track; inputting a to-be-detected track into the model to obtain a predicted track so as to realize a track prediction function by using the model. According to the LSTM track prediction model combined with the ADS-B data, the difference between the prediction result and the actual value is small, the fitting d |
---|---|
Bibliography: | Application Number: CN202310800941 |