Table 2 Performance comparison in different flight phases

From: Flight trajectory prediction enabled by time-frequency wavelet transform

Phases

Models

MAE↓

MRE(%)↓

RMSE↓

MDE↓

  

Lon

Lat

Alt

Lon

Lat

Alt

Lon

Lat

Alt

 

Cruise

A1

0.0043

0.0050

0.58

0.0041

0.0180

0.07

0.0166

0.0142

5.90

0.7789

 

A2

0.0043

0.0044

0.59

0.0040

0.0159

0.07

0.0155

0.0136

5.47

0.7173

 

A3

0.0041

0.0042

0.62

0.0038

0.0152

0.07

0.0169

0.0144

6.38

0.6837

 

A4

0.0045

0.0050

0.56

0.0042

0.0173

0.06

0.0155

0.0140

5.58

0.7954

 

A5

0.0042

0.0033

0.09

0.0041

0.0133

0.01

0.0612

0.0458

1.90

0.5479

 

WTFTP

0.0022

0.0019

0.36

0.0020

0.0070

0.04

0.0164

0.0136

6.35

0.3405

Climb

A1

0.0084

0.0079

3.16

0.0079

0.0277

0.57

0.0118

0.0111

4.42

1.3473

 

A2

0.0079

0.0087

3.12

0.0075

0.0306

0.57

0.0116

0.0118

4.46

1.3701

 

A3

0.0070

0.0067

3.12

0.0066

0.0234

0.57

0.0104

0.0092

4.30

1.1225

 

A4

0.0060

0.0070

3.00

0.0057

0.0245

0.56

0.0084

0.0094

4.18

1.0802

 

A5

0.0046

0.0040

7.46

0.0044

0.0144

1.45

0.0521

0.0512

28.08

0.9032

 

WTFTP

0.0030

0.0028

3.01

0.0028

0.0098

0.55

0.0043

0.0038

4.16

0.4846

Descent

A1

0.0066

0.0060

1.94

0.0062

0.0211

0.55

0.0097

0.0085

2.89

1.0363

 

A2

0.0061

0.0062

1.91

0.0057

0.0222

0.54

0.0086

0.0086

2.82

1.0128

 

A3

0.0060

0.0052

2.00

0.0056

0.0183

0.56

0.0088

0.0073

2.95

0.9196

 

A4

0.0050

0.0060

2.09

0.0047

0.0211

0.63

0.0075

0.0080

2.95

0.9175

 

A5

0.0041

0.0047

3.94

0.0040

0.0127

0.92

0.0504

0.0726

18.85

0.7996

 

WTFTP

0.0024

0.0021

1.81

0.0023

0.0074

0.50

0.0040

0.0030

2.73

0.3753

  1. A1–A5 represent the baselines: Vanilla LSTM, TCN, CNN LSTM, Transformer and FlightBERT, respectively.
  2. The Lon, Lat and Alt stand for longitude, latitude and altitude, respectively.
  3. In the MAE and RMSE metrics, the Lon and Lat are measured in degrees, and the Alt is measured in 10 m. The MDE is measured in kilometers.
  4. The bold ones denote the best performance on the corresponding metric.
  5. ↓ represents minimization indicators.