Incident Research Of Civil Aviation Passenger Transport Based On Data Mining

"Prioritizing safety,emphasizing prevention,and implementing comprehensive management" is the fundamental principle of safety production in our country.In order to minimize the occurrence of aviation accidents and incidents,and to obtain as detailed information as possible in the first instance of such incidents,according to the recommendations in Annex 13 of the ICAO,countries and organizations are required to establish a mandatory accident reporting system,as well as a non-punitive voluntary accident reporting system.As early as the 1970s,NASA establishing a voluntary reporting system,which later developed into the earliest implemented aviation safety voluntary reporting system,ASRS.This article presents a comprehensive study on the voluntary reporting system for civil aviation safety.Based on the aviation safety reporting ASRS system,a structured 6W expression model is established to conduct complete and systematic research on information from different dimensions and different data structures.According to the data types in the 6W structured expression model,two data mining methods are selected:multidimensional analysis and text classification.Firstly,from the perspective of multi-dimensional analysis,extract association information to the maximum extent.In the study,the correspondence analysis method was used to conduct two-dimensional and multi-dimensional variable interaction analysis,respectively,to determine dangerous and unsafe situations that may not have caused accidents during flight operations.Then,a supervised classification method is used to mine the association between"Narrative" text and causal factors in the report.The classification experiment design is based on the characteristics of civil aviation safety report text and tags.The final multi-classification model achieves automatic classification of "Contributing Factors/Situations"and"Primary Problem".