Correlation Analysis Of Influencing Factors Of Typical Civil Aviation Unsafe Events Based On Improved Apriori Algorithm

The influencing factors of one or more dimensions work together and interact to form different types of civil aviation unsafe events.The study of the influencing factors of civil aviation unsafe events is currently a hot topic of academic research,but the correlation between different influencing factors and the degree of correlation as well as the differences between several typical influencing factors of unsafe events have been rarely discussed.To address this issue,firstly,the indicators were initially selected based on the findings of previous studies,and an impact factor system evaluation model containing 66 indicator variables in 3 levels and 5dimensions was constructed by combining the characteristics of Aviation Safety Reporting System(ASRS)data,based on which the impact factor association rule mining was carried out.The AHP-DEMATEL method was used to assign weights to compensate for the shortcomings of the traditional Apriori association rule algorithm that does not consider the importance differences between different influencing factors and the confusing causal logic of association rules,and the weighted ordered directed constraint Apriori algorithm is proposed.The improved Apriori algorithm was applied to mine association rules for each of the influencing factors of four typical civil aviation unsafe events in the ASRS database,and the mined strong association rules were analyzed and interpreted in detail.In addition,the association rule mining results were compared with the traditional statistical analysis results.The main research findings were as follows:(1)The distribution characteristics of the important influencing factors in the index system were analyzed using mathematical statistical methods,and it was found that the number of unsafe events reported in the last decade showed a fluctuating upward trend;the proportion of severe weather affecting flight safety was in the order of multiple complex weather,turbulence,thunderstorm,rain,hail,windshear,etc.;there were significant differences in the flight phases where the four typical unsafe events often occurred;among the human factors,the error or loss of situational awareness is dominant in the four typical unsafe events,while other factors have different percentages;deviation-procedural Published Material/Policy accounts for the highest percentage of CFIT/CFTT and unstabilized approach events,and aircraft equipment problems was the most common problem causing in-flight loss of control and aircraft fuel events,etc.(2)Through the improved Apriori algorithm to explore the association of factors influencing typical unsafe events,it was found that environmental factors,especially severe weather,such as windshear and turbulence,are significantly associated with in-flight loss of control events;CFIT/CFTT and unstabilized approach events were less influenced by severe weather,but these two types of unsafe events were significantly associated with the distribution of flight phases,and initial approach and final approach were the two main flight phases causing these two types of events;aircraft fuel events mostly occur in winter and were closely associated with aircraft equipment problems,etc.(3)Comparing the association rule mining results with the distribution feature statistics results,it was found that the influencing factors with high frequency occurrence were also mostly found in the strong association rules,but the association rule strength of the high frequency influencing factors was not necessarily high.In general,the dimensionality of the association rule mining results was more comprehensive and more informative.Applying the improved Apriori algorithm to association rule mining of the influence factors of unsafe events can provide theoretical support and technical reference for research in related fields.