Affiliations: [a]
Nanjing Tech University, Nanjing, China
| [b]
Nanjing Institute of Measurement and Testing Technology, Nanjing, China
| [c] College of Civil Aviation, Nanjing University of Aeronautics and Astronautics, Nanjing, China
| [d]
Nanjing Special Equipment Inspection Institute, Nanjing, China
Note: [1] Jun-cheng Jiang is currently a full Professor in Nanjing Tech University.
Abstract: To reduce the airline accident risk probability, this paper proposes a new airline safety assessment method based on fuzzy mathematics and Bayesian networks (BN). Herein, we construct a safety assessment system encompassing five aspects— namely maintenance quality, aircraft technical state, environmental effects, emergency rescue, and safety management— and establish a BN model based on this safety assessment system. The fuzzy mathematical and statistical analyses are used to obtain the prior probabilities, and the data training function in GeNIe2.1 is used to obtain the conditional probabilities. Finally, we apply our method to an unspecified airline. The results indicate that the risk probability was 0.826 for the airline to have an excellent safety status in January 2018; this value was 0.886 according to fault tree analysis (FTA). In addition, by using vertical and horizontal analyses, we investigate the factors affecting airline safety. Thus, our BN-based method is more efficient than FTA because compared with FTA, the BN has the advantages, such as polymorphism and accuracy, particularly in detecting the most risky factors in a complex model.
Keywords: Bayesian network, fuzzy mathematics, fault tree analysis, safety assessment
