Prediction and Optimization of Civil Aviation Flight Delays Based on Machine Learning Algorithms
Abstract The civil aviation industry continues to face the significant problem of flight delays, which impacts operational efficiency and passenger satisfaction. This research aims to develop an innovative model that accurately predicts civil aviation flight delays and provides insights related to p...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-07-01
|
| Series: | International Journal of Computational Intelligence Systems |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s44196-025-00932-2 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract The civil aviation industry continues to face the significant problem of flight delays, which impacts operational efficiency and passenger satisfaction. This research aims to develop an innovative model that accurately predicts civil aviation flight delays and provides insights related to performance enhancement. The proposed Flight Delay Prediction Network with Spatio-Temporal Learning (FlightNet-ST) is a hybrid deep learning architecture that combines Long Short-Term Memory (LSTM) networks, Graph Convolutional Networks (GCNs), and 3D Convolutional Neural Networks (3D-CNNs) to achieve this goal. The model is trained using datasets of domestic flights that include geographical, operational, and temporal data, such as geographical separation, origin–destination pairs, airline rules, scheduled departure times, and dates. The approach involves running time series data through LSTM to capture temporal dependencies, applying 3D Convolutional Neural Networks (3D-CNNs) to analyze aircraft route grids dynamically, and utilizing Graph Convolutional Networks (GCNs) to discover topological patterns from spatial airport connectivity. Delay prediction is powered by a unified representation that fuses these disparate elements. Based on the experimental data, FlightNet-ST achieves a 14.47% reduction in Mean Absolute Error (MAE). Additionally, an attention mechanism enhances interpretability by highlighting key aspects that influence delays, such as departure time blocks and airport-specific trends. Finally, FlightNet-ST helps with civil aviation flight delay prediction and management with its data-driven, interpretable, and robust solution. This methodology facilitates real-time operational decision-making and provides tactics to mitigate delays. |
|---|---|
| ISSN: | 1875-6883 |