Research On Short-term Trajectory Prediction Method Based On LSTM-ARIMA And Visualization System Development

Accurate trajectory prediction technology plays an increasingly important role in solving the problem of increasing shortage of airspace resources.It is also one of the key technologies of Air Traffic Management concept based on trajectory operation in the future.It has great significance in cooperative control,trajectory optimization,conflict detection and release and other fields.With the development of China's air transportation industry,air traffic flow is more and more,and the flight data of Air Traffic Control System is also increasing dramatically.It is an inevitable trend for Air Traffic Control big data to extract useful information from massive and highly complex data,deeply mine the organization mode and evolution rule of air traffic flow,and effectively apply it to the field of trajectory analysis.In this paper,a short-term trajectory prediction algorithm based on LSTM-ARIMA is proposed for a large number of flight data.The basic idea of the algorithm is to use the combined model based on LSTM(Long Short-Term Memory,LSTM)neural network and ARIMA(Autoregressive Integrated Moving Average,ARIMA)model to model the historical flight data,to mine the hidden dependency of the historical flight data,and to apply it to new trajectory prediction.Firstly,by analyzing the existing flight data of historical flights,two new feature data of the distance to the target airport and the turning state,are added through feature extension to build a trajectory prediction model based on LSTM neural network.Then,in view of the shortcomings of the algorithm based on LSTM,combined with the characteristics of LSTM neural network and ARIMA model,we use LSTM to mine the non-linear relationship of longitude,latitude and height in the flight data of historical flights,and use ARIMA model to model the linear relationship of height,and the height prediction values of the two models are fused with the method of CRITIC(Criteria Importance Though Intercrieria Correlation,CRITIC).The three-dimensional position of the predicted trajectory is constituted by the height values after fusion and the longitude,altitude predicted by the LSTM model.Finally,the actual flight data is used for verification and analysis.The analysis results show the feasibility and accuracy of the trajectory prediction model based on LSTM-ARIMA.In addition,in order to display the trajectory more intuitively and vividly,further mine the information in the trajectory data,and maximize the use value of the historical flight data,a set of trajectory data visualization system based on Web is developed in this paper.The system not only integrates the trajectory prediction algorithm is proposed in this paper,realizes the dynamic display of the trajectory prediction results,but also makes further mining and analysis of flight data,and displays the analysis results in the form of visual charts,which improves the use value of the flight data.