Flight time prediction for fuel loading decisions with a deep learning approach

Under increasing economic and environmental pressure, airlines are constantly seeking new technologies optimizing flight operations to reduce fuel consumption. However, the current practice on loading, which has a significant impact aircraft weight consumption, yet be thoroughly addressed by existing studies. Excess is loaded dispatchers (or) pilots handle consumption uncertainties, primarily caused time cannot predicted Flight Planning Systems (FPS). In this paper, we develop novel spatial weighted recurrent neural network model provide better predictions capturing air traffic information at national scale based multiple data sources, including Automatic Dependent Surveillance - Broadcast (ADS-B), Meteorological Aerodrome Reports (METAR), airline records. model, layer designed extract dependences among delay states (i.e. average each airport of Origin-Destination (OD) pair for specific interval). Then, training procedure associated with introduced OD-specific weights then integrate into one nationwide network. Long short-term memory (LSTM) networks used after temporal behavior patterns states. Finally, features from delays, weather, schedules fed fully connected predict particular flight. The proposed was evaluated using year historical an airline’s real operations. Results show that our can more accurate than baseline methods, especially flights extreme delays. We also that, improved prediction, loading optimized resulting reduced 0.016%–1.915% without depletion risk.