Based on the Improved LSTM Model Aircraft Long Track Prediction Research

Abstract

The plane long track forecast for half an hour or more forecast for a long period of time, is a military action to intentions and cracking down on the key technology of auxiliary decision, also it can be used in the field of civil aviation traffic control, strengthen the management of the aircraft. This article adopts data mining methods to improve the LSTM network structure. It proposes an LSTM network that combines convolutional neural networks and attention mechanisms and uses radar-measured aircraft trajectory data for long-term prediction. The predicted results show that the improved LSTM network structure has certain performance improvements in prediction. Based on this, the article analyzes the impact of prediction steps on the algorithm and presents the analysis results.