Improved gray correlation analysis and combined prediction model for aviation accidents

• DOI: 10.1108/ec-06-2022-0384
• Publication Date: 2023-08-08T17:58:10Z
• Authors: Siyu Su; Youchao Sun; Chong Peng; Yuanyuan Guo
• Keywords: Aviation accident; Takeoff; Airplane; Aviation fuel

Purpose The purpose of this paper is to identify the key influencing factors aviation accidents and predict caused by factors. Design/methodology/approach This proposes an improved gray correlation analysis (IGCA) theory make relational find out critical causes accidents. optimal varying weight combination model (OVW-CM) constructed based on gradient boosted regression tree (GBRT), extreme boosting (XGBoost) support vector (SVR) due Findings global accident data from 1919 2020 selected as experimental data. airplane, takeoff/landing unexpected results are leading IGCA. Then GBRT, XGBoost, SVR, equal-weight (EQ-CM), variance-covariance (VCW-CM) OVW-CM used results, respectively. show that has a better prediction effect, accuracy stability higher than other models. Originality/value Unlike traditional (GCA), IGCA weights sample distance more objectively reflect degree influence different built minimizing combined error at points assigns individual models moments, which can full use advantages each accuracy. And parameters XGBoost SVR optimized particle swarm algorithm. study guide provide scientific basis for safety management.