Aviation Safety Risk Prediction Model Based on Machine Learning and Data Mining
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
This study addresses the current challenges of delayed accident hazard identification and the lack of dynamic and predictive risk assessment in aviation safety management. This study proposes an aviation safety risk prediction model based on machine learning and data mining, aiming to improve the scientific and real-time nature of aviation risk identification. First, historical flight operation data, meteorological data, and crew reports are cleaned and feature engineered to construct a multi-dimensional indicator system covering flight environment, equipment status, and crew operations. Principal Component Analysis (PCA) is then used for dimensionality reduction, eliminating redundant features while retaining key risk factors. Classification and prediction are then performed using Random Forest (RF) and Support Vector Machine (SVM). RF is used for feature importance ranking and preliminary screening, while SVM is used to accurately predict high-risk flight events. K-Means Clustering is also introduced for unsupervised mining of potential risk patterns, enabling the identification of risk distribution structures. In the experimental part, 200,000 pieces of operational data from the Civil Aviation Administration of China from 2015 to 2022 are selected as samples. After cross-validation, the model prediction accuracy reached 92.8% and the recall rate reached 88.9%. This method can effectively discover potential hidden danger patterns and realize early warning of high-risk flights, and has strong engineering application value and promotion significance.
Fields of Study & Subjects
Evaluation and Optimization Models
Air Traffic Management and Optimization
Data Mining and Machine Learning Applications
Publication details
- Journal / source
- IEEE International Conference on Power, Electronics and Green Energy (ICPEGE)
- Pages
- 615–620
- Publication date
- 28 January 2026
- Language
- EN