Aircraft four-dimensional track prediction method based on multi-machine interaction network model

The invention provides an aircraft four-dimensional track prediction method based on a multi-aircraft interaction network model. The method comprises the steps of 1, collecting historical track data of a plurality of aircrafts in the same airspace and processing the historical track data to serve as training data; 2, constructing a multi-machine interaction network model, and training the multi-machine interaction network model by using training data; and 3, inputting historical track data of a plurality of aircrafts in the airspace to be predicted into the trained multi-aircraft interaction network model, and predicting to obtain future track points, including position, speed, height and time information, of the plurality of aircrafts. According to the method, the operation characteristics of the aircrafts and the interaction among different aircrafts in the flight process are considered, the influence factor attention weight system of different types of aircrafts is also established, the constructed multi-a