Date of Award
Dissertation - Open Access
Doctor of Philosophy in Aviation
College of Aviation
Dahai Liu, Ph.D.
First Committee Member
Dothang Truong, Ph.D.
Second Committee Member
Houbing Song, Ph.D.
Third Committee Member
Craig H. Neuman, Ph.D.
The purpose of this study was to apply support vector machine (SVM) models to predict the severity of aircraft damage and the severity of personal injury during an aircraft approach and landing accident and to evaluate and rank the importance of 14 accident factors to the severity. Three new factors were introduced using the theory of inattentional blindness: The presence of visual area surface penetrations for a runway, the Federal Aviation Administration’s (FAA) visual area surface penetration policy timeframe, and the type of runway approach lighting.
The study comprised 1,297 aircraft approach and landing accidents at airports within the United States with at least one instrument approach procedure. The dataset was gathered from a combination of the National Transportation Safety Board (NTSB) accident database, the NTSB accident reports, and the FAA’s Instrument Flight Procedure Gateway website. Four SVM models were developed in using the linear, polynomial, radial basis function (RBF), and sigmoid kernels for the severity of aircraft damage and another four SVM models were developed for the severity of personal injury. Five-fold cross-validation was used for testing the model accuracy and measures including evaluation of confusion matrices, misclassification rates, accuracy, precision, sensitivity/recall, and F1-scores for model comparison. The best kernel models were selected and its model hyperparameters were optimized for the best model performance.
The SVM models using the RBF kernel produced the best machine learning models, with a 96% accuracy for predicting the severity of aircraft damage (0.94 precision, 0.95 recall, and 0.95 F1-score) and a 98% accuracy for predicting the severity of personal injury (0.99 precision, 0.98 recall, and 0.99 F1-score). The top predictors across both models were the pilot’s total flight hours, time of the accident, pilot’s age, crosswind component, landing runway number, single-engine land certificate, and any obstacle penetration. Specifically, the visual area surface obstacle penetration status ranked ninth across both SVM models. However, as a sub-category, an obstacle penetration on final approach was the seventh overall predictor and the second highest of the categorical predictors. The FAA visual area surface policy was ranked eighth as the overall factor, and the FAA policy from 2018 to 2019 was the third highest categorical predictor. Finally, the type of runway lighting was the sixth ranked prediction factor.
This study demonstrates the benefit of SVM modeling using the RBF kernel for accident prediction and for datasets with categorical factors. It is recommended for the NTSB to add the collection of all three new factors into the NTSB database for future aviation accident research. Lastly, flight training should include information on a pilot’s susceptibility to inattentional blindness and the risks of potential obstacles in their flight path.
Scholarly Commons Citation
Silagyi, Dezsö V., "Prediction of Severity of Aviation Landing Accidents Using Support Vector Machine Models" (2022). Doctoral Dissertations and Master's Theses. 679.