Air traffic controller fatigue has become a significant concern for flight safety. With the sharp rise in global air traffic, it is imperative to assess controller fatigue, as it directly impacts the safety and efficiency of air traffic control operations. Our study introduces a non-intrusive method to detect fatigue by analyzing the facial and vocal characteristics of air traffic controllers. Initially, we developed fast and accurate schemes for facial feature extraction, which allowed us to measure the “percentage of eyelid closures” and yawn frequency from video recordings. Subsequently, we extracted several vocal features from audio recordings, including average fundamental frequency, short-time average magnitude, short-time zero-crossing rate, harmonic-to-noise ratio, jitter, shimmer, loudness, and Mel-frequency cepstrum coefficient. We then created temporal sequences of these facial and vocal features to feed into a bi-directional long short-term memory gated recurrent unit network. This data, combined with the Stanford Sleepiness Scale, facilitated the identification and precise prediction of controller fatigue levels. Our experimental findings validate the effectiveness of the proposed detection method, which demonstrated a recognition accuracy rate of 95.12% on the test audio and video datasets.

The construction of a smart aviation transportation system can improve the informatization level and decision support ability of aviation traffic management and services, and is an effective way to improve aviation traffic management, reduce traffic accidents, and improve people's quality of life. In air transportation, due to factors such as severe weather, there are repeated routes in the air transportation route, resulting in problems such as long transportation distance and reduced transportation efficiency. In intelligent aviation path planning, ant colony optimization (ACO) is a popular path solving strategy and has been widely applied. This paper proposes an air transportation path planning method based on deep learning (DL) and ACO for intelligent aviation path planning. By combining Convolutional Neural Networks (CNN) and Long Short Term Memory Networks (LSTM), accurate prediction of the shortest path in complex environments has been achieved. Under various constraints, real-time optimization of ACO is used to obtain the optimal solution of the model and achieve air transportation path planning. The simulation results show that the algorithm designed in this article has good feasibility in complex environments and can save costs.

Confronted with the imperatives of sustainable development within the civil aviation sector, the precision in prognostication of aircraft fuel expenditure emerges as a critical imperative. To overcome the shortcomings in the current study of aircraft range fuel prediction, we propose a novel fuel consumption prediction model that integrates Wavelet Packet Decomposition (WPD) with an Improved Arctic Puffin Optimization (IAPO) optimized Bidirectional Long-Short-Term Memory network–Kolmogorov-Arnold network (BiLSTM–KAN). Initially, Pearson’s correlation coefficient is employed to select the most significant features. Subsequently, WPD decomposes the raw fuel consumption data into subsequences across various frequency bands. The BiLSTM network effectively captures long-term dependency features within the sequence data, which are then input into KAN to further elucidate the complex nonlinear relationships present in the data. Additionally, the SPM chaotic mapping strategy is utilized for population initialization, while the introduction of the golden sine operator variation strategy enhances the local search capabilities of the algorithm. The adaptive swoop switching strategy adjusts the search intensity, thereby improving the global search performance and convergence speed of the Arctic Puffin Optimization (APO). Ultimately, the multi-strategy improved APO is employed to optimize the hyperparameters of the BiLSTM–KAN model, allowing for the superposition of each subsequence to yield the final prediction results. Experimental results indicate that in the B737 aircraft model, the Mean Squared Error, Normalized Root Mean Squared Error (NRMSE), Mean Absolute Percentage Error (MAPE), and Coefficient of Determination (R²) of the proposed WPD–IAPO–BiLSTM–KAN model are 34.57, 0.0061, 3.81, and 0.9974, respectively. In the A320 aircraft model, the MSE, NRMSE, MAPE, and R² of this model are 40.71, 0.0078, 2.61, and 0.9934, respectively. In the B747 aircraft model, the MSE, NRMSE, MAPE, and R² of the model are 242.17, 0.0110, 3.88, and 0.9828, respectively. The WPD–IAPO–BiLSTM–KAN model surpasses other comparative models in prediction accuracy, exhibiting a low prediction error. This model represents a novel and effective approach to predicting airline fuel consumption and provides valuable insights for reducing aircraft fuel consumption.

We utilized the International Civil Aviation Organization (ICAO) standard emission model, refined with adjustments for fuel flow, LTO cycle work mode time, and emission indices, to investigate the environmental footprint of airports at different altitudes. Airports categorized as high (above 5000 $ \mathrm{f}\mathrm{t} $), medium (1500–5000 $ \mathrm{f}\mathrm{t} $), and low (below 1500 $ \mathrm{f}\mathrm{t} $) altitudes were selected to provide a comprehensive representation of the altitude spectrum. The analysis was anchored over the period spanning 2016 to 2017. Emission inventories for air pollutants and noise were computed for these airports, focusing on the LTO (Landing and Take-Off) cycle. Our findings indicated that high altitude airports exhibit the highest $ {\mathrm{N}\mathrm{O}}_{\mathrm{x}} $ emissions, reaching 406.4 $ \mathrm{t} $, whereas low altitude airports record the highest noise levels at 73.1 $ \mathrm{d}\mathrm{B} $. Significant disparities in emission profiles were observed across different phases of the LTO cycle at airports of varying altitudes. Notably, during the climb phase, the types and proportion of $ {\mathrm{N}\mathrm{O}}_{\mathrm{x}} $ emissions at high altitude airports were as high as 71.8%, contrasting with the 45.6% at low altitude airports. Additionally, emissions of gaseous pollutants from major aircrafts, exemplified by the A320 model, escalated with altitude. Specifically, $ {\mathrm{N}\mathrm{O}}_{\mathrm{x}} $ emissions increased from 10.55 $ \mathrm{k}\mathrm{g}/\mathrm{c}\mathrm{y}\mathrm{c}\mathrm{l}\mathrm{e} $ at low altitude to 20.48 $ \mathrm{k}\mathrm{g}/\mathrm{c}\mathrm{y}\mathrm{c}\mathrm{l}\mathrm{e} $ at high altitude, and $ \mathrm{C}\mathrm{O} $ emissions from 10.88 $ \mathrm{k}\mathrm{g}/\mathrm{c}\mathrm{y}\mathrm{c}\mathrm{l}\mathrm{e} $ to 22.89 $ \mathrm{k}\mathrm{g}/\mathrm{c}\mathrm{y}\mathrm{c}\mathrm{l}\mathrm{e} $. A robust correlation between $ {\mathrm{N}\mathrm{O}}_{\mathrm{x}} $ emissions and $ {L}_{den} $ was identified among airports at different altitudes, with correlation coefficients of 0.96 for low altitude, 0.97 for medium altitude, and 0.93 for high altitude airports. This study delineates the distinct characteristics of air pollutant and noise emissions from airports across altitudes, offering novel insights for the environmental assessment of airport operations.

Currently, improving runway utilization under operational safety constraints has become a critical concern for small and medium airports. Existing research focuses primarily on landing-phase runway occupation time, while predictive studies on the takeoff phase remain limited. Analysis of 1749 Quick Access Recorder (QAR) records from ten airports reveals that departure runway occupation time is strongly correlated with ground speed at liftoff (0.72) and airport elevation (0.67) but weakly correlated with aircraft weight and meteorological conditions, providing guidance for feature engineering. To address the prediction of departure runway occupation time, this study proposes a TCN-FEP hybrid model. The model employs an enhanced Temporal Convolutional Network (TCN) module with multi-scale convolutions (kernel sizes 3, 5, 7) and dilated convolutions (rates 2, 4, 8) to capture multi-scale feature interactions, alongside a Feature Enhancement Projection (FEP) module that maps local features into a high-dimensional latent space for implicit relationship mining and global information integration. Experimental results demonstrate that the proposed TCN-FEP model achieves an MSE of 90.20, RMSE of 9.49, MAE of 5.84 s, MAPE of 3.80%, and R² of 0.97, outperforming Informer (MSE 117.95), Longformer (MSE 132.11), XGBoost (MSE 92.30), and LightGBM (MSE 91.45). Under 5% outlier injection, MSE increases by 7.9%, compared to 24.3% for LSTM and 18.4% for Informer. With 94% of prediction errors within ±5 s, the model’s accuracy may offer a useful reference for runway resource optimization at small and medium airports.