Aircraft surface trajectory prediction method based on long-short term memory (LSTM) neural network

Abstract

The invention provides an aircraft surface trajectory prediction method based on a long-short term memory (LSTM) neural network. The LSTM neural network and a polynomial fitting method are combined toimplement the trajectory prediction technology. By setting an incremental sampling period, the position at any moment with in a long period of 60 seconds is theoretically predictable. However, when the position within an excessively long period is predicted, the training data generated by preprocessing has low quality so as to result in low prediction precision and has no practical effect on thedetection of surface taxiing conflicts. Thus, it is relatively appropriate to predict the position at any moment with in a middle-long period of 30 seconds. The method, by means of the historical memory of the LSTM neural network, can implicitly simulate the surface motion state of an aircraft according to the context of a trajectory sequence, can be used for predicting the position of the aircraft on the taxiway and the runway of an airport in the future, avoids the surface taxiing conflicts of the aircrafts, paves the way for real-time path planning, ensures the safe and efficient operationfor the airport.

Description

technical field [0001] The invention relates to an airport aircraft scene track prediction technology. Background technique [0002] With the rapid and continuous development of the global air transport industry, the traffic on the airport surface is becoming more and more busy, especially in the process of traffic control on the airport surface, resources such as space, time, and manpower are not fully utilized. There are still potential safety hazards of taxiing conflicts, which directly affect the operational safety and efficiency of the airport scene. The trajectory prediction technology can be used to judge the geographic location of the aircraft in the future time period, and can also estimate the arrival time and departure time of the aircraft at key intersections, reduce traffic congestion on the scene, avoid taxiing conflicts, shorten the total taxiing time of aircraft, ensure high-speed operation of the airport, and improve service quality. [0003] At present, t...

AI Technical Summary

Problems solved by technology

However, using LSTM can only achieve position prediction at some discrete time in the future. This time is determined by the sampling period of the training data, and it is impossible to achieve position prediction at any time in the medium and long term in the future.

Images