Volume V | Issue 1 | 2026 | Paper ID: IJATEM-V05I01P3

Explainable Swin Transformer Framework for Aircraft Bird Strike Risk Prediction Using SHAP-Based Feature Analysis

Author(s)

, J N Rajesh Kumar

Abstract

Aircraft bird strike risk prediction is important because it prevents catastrophic engine failures, structural damage, and loss of life during low-altitude flights. In this paper, a Swin Transformer (ST) with Shapley Additive explanations (SHAP) based feature analysis is proposed for aircraft bird strike risk prediction. Data preprocessing transforms raw, noisy multi-source aviation and ecological records into clean inputs for predictive models. It handles missing radar tracks, standardizes diverse weather metrics, and encodes cyclical time features, ensuring Deep Learning (DL) algorithms accurately forecast bird strike risks. STs are applied in bird strike risk prediction and airport safety frameworks primarily for high-accuracy small-object detection and real-time visual monitoring of avian hazards. SHAP interprets complex DL models used in aircraft bird strike risk prediction by quantifying the individual impact of environmental, temporal, and operational features on collision probability. The performance analysis is carried out using python software and assesses the model's performance using key metrics such as precision, recall, F1-score, confusion matrix, and ROC-AUC. The proposed ST-SHAP uses two classes, they are no damage and caused damage. The proposed ST-SHAP achieves high accuracy of 96% and better convergence speed. The proposed method outperforms conventional techniques, contributing a robust and reliable tool for aircraft bird strike risk prediction.

Keywords

Swim Transformer (ST), Shapley Additive explanations (SHAP), aircraft bird strike risk prediction, Data Preprocessing.

References

[1] Shlok Misra; Ila Toppo; Flavio Antonio Coimbra Mendonca, Year: 2022, “Assessment of aircraft damage due to bird strikes: a machine learning approach”, International Journal of Sustainable Aviation, Vol: 8, No: 2, pp. 136-151

CrossRef Google Scholar Publisher Link
[2] Luís FFM Santos, Victor Tavares, Flávio Lázaro, Rui Melicio, Duarte Valério, Year: 2026, “Bird Strike Risk in Aviation: A Calibrated Threshold-Optimized Machine Learning Decision Support”, Mathematics, Vol: 14, No: 14.

CrossRef Google Scholar Publisher Link
[3] LIU Zhenjiang; WANG Zheng; CAO Yujie; CAI Shuwei; CHEN Wei; LU Kaili, Year: 2025, “Risk assessment of bird strike at airport based on improved CRITIC-cloud model”, Ecology and Environment, Vol: 34, No: 1.

Google Scholar
[4] Changyang Yu; Fan Zhao; Yu Yin; Yi Wu, Year: 2025, “Entropy-weighted TOPSIS-based bird strike risk assessment for an airport”, International Journal of Computational Intelligence Systems, Vol: 18, No: 1.

CrossRef Google Scholar Publisher Link
[5] Rui Wang; Qiang Zhao; Hui Sun; Xuedong Zhang; Yuyue Wang, Year: 2022, “Risk assessment model based on set pair analysis applied to airport bird strikes”, Sustainability, Vol. 14, No: 19, pp.12877.

CrossRef Google Scholar Publisher Link
[6] Jaya Dipti Lal; Shahnawaz Ayoub; Prashant D Hakim; S. Prabagar; Vijay Kumar Dwivedi; Mohit Tiwari, Year: 2023, “Hybrid Deep Learning based Attack Detection and Classification Model on IoT Environment”, IEEE, pp. 1-5.

CrossRef Google Scholar Publisher Link
[7] Sabziyan Varnousfaderani; Shihab, Year: 2023. Bird movement prediction using long short-term memory networks to prevent bird strikes with low altitude aircraft”, In AIAA AVIATION 2023 Forum, pp. 4531.

CrossRef Google Scholar Publisher Link
[8] Elaheh Sabziyan Varnousfaderani; Syed AM Shihab; Jonathan King, Year: 2026, “Deep Learning Based Multi-Level Classification for Aviation Safety”, Vol: 12, pp. 1-10.

CrossRef Google Scholar Publisher Link
[9] Najiba Said Hamad Alzadjali; Sundara vadivazhagan; Charles Savarimuthu, Year: 2024, “A deep learning framework for real time bird detection and its implications for reducing bird strike incidents”, Sensors, Vol: 24, No: 17.

CrossRef Google Scholar Publisher Link
[10] Jason Powell; Dahai Liu, Year: 2026, “Aircraft Bird Strike Risk Prediction Using LSTM-RNN, XGBoost, and BNN”, SSRN, pp. 1-79.

CrossRef Google Scholar Publisher Link

Full Length Article PDF

Article Details

Explainable Swin Transformer Framework for Aircraft Bird Strike Risk Prediction Using SHAP-Based Feature Analysis

Author(s)

Abstract

Keywords

References

Quick Info