Data driven fuel consumption prediction model for green aviation using radial basis function neural network.

Area of Science:

• Aerospace Engineering
• Artificial Intelligence
• Sustainable Aviation

Background:

• Growing demand for sustainable aviation necessitates advanced fuel consumption prediction.
• Traditional methods struggle with high-dimensional, nonlinear flight data.
• Quick Access Recorder (QAR) data offers high-resolution parameters for improved accuracy.

Purpose of the Study:

• To develop a lightweight and computationally efficient fuel consumption prediction model for sustainable aviation.
• To enable accurate, real-time fuel prediction for both ground-based and onboard applications.
• To support airlines in optimizing flight planning and minimizing fuel usage.

Main Methods:

• Utilized Radial Basis Function (RBF) Neural Networks trained on high-resolution Quick Access Recorder (QAR) data.
• Extracted key influencing factors for different flight phases: takeoff/climb, cruise, and descent/approach.
• Validated model robustness using ten-fold cross-validation.

Main Results:

• Achieved prediction errors of 5.73% (takeoff/climb), 3.36% (cruise), and 14.04% (descent/approach).
• Demonstrated significantly superior performance compared to existing models.
• Error variances from cross-validation were low (0.31%, 0.15%, 0.29%), confirming model robustness.

Conclusions:

• The RBF model offers an accurate, efficient solution for predicting aircraft fuel consumption in various flight phases.
• The model is suitable for both pre-flight analysis and real-time onboard deployment in resource-constrained environments.
• This research provides valuable insights for enhancing fuel efficiency and supporting the development of green aviation.