E-Pilots: A System to Predict Hard Landing During the Approach Phase of Commercial Flights

Abstract

Hard landings represent a persistent hazard in commercial aviation, causing structural stress to airframes, increasing maintenance expenditure, and endangering passenger safety. Existing monitoring approaches predominantly perform post-flight analysis, flagging hard landings only after they have occurred. This paper presents E-Pilots, an intelligent software system that analyzes critical flight parameters gathered during the approach phase to predict the probability of a hard landing before the aircraft reaches the runway threshold. The proposed framework ingests parameters including altitude, vertical speed, descent rate, pitch angle, and ambient weather conditions, and processes them through a multi-model prediction pipeline comprising Support Vector Machines (SVM), Logistic Regression, and a Hybrid Long Short-Term Memory (LSTM) network. Feature selection is applied to isolate the most predictive indicators, and a rule-assisted classification layer maps the combined model outputs to three safety states: Safe, Warning, and Hard Landing. Evaluated under realistic flight data scenarios, the system achieves reliable classification with reduced false-positive rates compared to single-indicator threshold methods. The system further provides a graphical user interface for dataset upload, real-time result visualization, and session reporting. E-Pilots demonstrates that data-driven, multi-model prediction can meaningfully augment aviation safety by delivering advance warning during the most safety-critical phase of a commercial flight.

Introduction

Hard landings are a major safety concern in aviation, occurring when an aircraft touches down with excessive force, potentially causing structural damage and safety risks. Although current systems detect such events after landing, predicting them in advance remains limited.

This paper introduces E-Pilots, a Python-based intelligent system that predicts landing risk before touchdown using flight data from the approach phase (e.g., altitude, descent rate, pitch, airspeed, and weather conditions). The system uses a hybrid machine learning pipeline combining Support Vector Machine (SVM), Logistic Regression, and a Hybrid LSTM model to classify landings into three categories: Safe, Warning, or Hard Landing.

The system includes modules for data input, preprocessing (handling missing data and normalization), feature selection (using mutual information), prediction, and result visualization via a graphical interface. It processes multivariate time-series flight data and uses an ensemble voting approach to improve accuracy and reduce false positives.

Compared to traditional threshold-based and post-flight analysis methods, E-Pilots provides proactive prediction and better performance by leveraging multiple features and deep learning. Experimental results show accurate classification, efficient processing time, and improved detection of risky landing conditions.

Despite its effectiveness, limitations include reliance on pre-recorded data and limited representation of extreme weather conditions. Future improvements aim to incorporate real-time data and adaptive learning.

Conclusion

This paper presented E-Pilots, an intelligent system for predicting hard landings during the approach phase of commercial flights. Unlike existing post-flight analysis tools, E-Pilots operates on approach-phase data to deliver advance classification of landing risk before touchdown, supporting timely corrective intervention. The system integrates a multi-model ML pipeline comprising SVM, Logistic Regression, and a Hybrid LSTM ensemble with automated feature selection and a three-tier safety classification output. Validation demonstrates reliable classification performance with reduced false-positive rates compared to single-indicator threshold methods. The graphical interface and session reporting capability make the system practical for aviation engineers and flight operations analysts. Future work will focus on integrating the system with real-time QAR data streams for live monitoring, incorporating additional parameters such as crosswind component, aircraft gross weight, and runway surface condition, and evaluating the system on larger aviation safety datasets. Extension to edge-deployable architectures for onboard advisory systems is also planned.

References

[1] L. M. Bergasa, J. Nuevo, M. A. Sotelo, R. Barea, and M. E. Lopez, \"Real-time system for monitoring driver vigilance,\" IEEE Transactions on Intelligent Transportation Systems, vol. 7, no. 1, pp. 63–77, 2006. [2] C. Cortes and V. Vapnik, \"Support-vector networks,\" Machine Learning, vol. 20, no. 3, pp. 273–297, 1995. [3] S. Hu and G. Zheng, \"Driver fatigue detection from brain oscillations using a single EEG channel,\" Expert Systems with Applications, vol. 38, no. 7, pp. 8764–8771, 2011. [4] S. Hochreiter and J. Schmidhuber, \"Long short-term memory,\" Neural Computation, vol. 9, no. 8, pp. 1735–1780, 1997. [5] Federal Aviation Administration (FAA), Aircraft Hard Landing Inspection Guidelines, FAA Advisory Circular AC 25.571, 2018. [6] International Civil Aviation Organization (ICAO), Manual of Aircraft Accident and Incident Investigation, Doc 9756, 4th ed., 2015. [7] A. Géron, Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow, 2nd ed. O’Reilly Media, 2019. [8] N. V. Chawla, K. W. Bowyer, L. O. Hall, and W. P. Kegelmeyer, \"SMOTE: Synthetic minority over-sampling technique,\" Journal of Artificial Intelligence Research, vol. 16, pp. 321–357, 2002. [9] I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning. MIT Press, 2016. [10] F. Pedregosa et al., \"Scikit-learn: Machine learning in Python,\" Journal of Machine Learning Research, vol. 12, pp. 2825–2830, 2011. [11] M. Abadi et al., \"TensorFlow: Large-scale machine learning on heterogeneous distributed systems,\" arXiv:1603.04467, 2016. [12] Y. Dong, Z. Hu, K. Uchimura, and N. Murayama, \"Driver inattention monitoring system for intelligent vehicles: A review,\" IEEE Transactions on Intelligent Transportation Systems, vol. 12, no. 2, pp. 596–614, 2011.

Copyright

Copyright © 2026 Chollangi Madhuri, Gannavarapu Durga Satish, Jayant Kamdi, Chiratapudi Suresh, Ganji Lakshmi. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.