Research On Flight Safety Risk Assessment Of Airline

With the continuous development of aviation industry in China, the scale of the airlineskept on expanding in recent years, transport capacity put on routes and airline numbers hadincreased year by year, while the pressure of flight safety was increasing. The flight safetyincidents and accidents occurred in recent year poses a severe challenge to the flight safetymanagers of Chinese airlines. Meanwhile, the flight safety is related to the development ofairlines and is an important influential factor on airlines’international image, and also it is atopic that passengers concerned deeply. Evaluation index system was identified thatinfluenced the flight safety, and BP neural network evaluation method was used to evaluatethe risk management level of flight safety, and some recommends were given that couldimprove flight safety management.First, comprehensive importance analysis of flight safety for the air transportation systemand airlines was studied, on the base of which, using a system theory, evaluation index systemof airlines’flight safety was identified with scientific, comprehensive, comparable andfeasible principles. The evaluation index system was composed of four first-level indicatorsand twenty-seven second-level indicators.Second, on the bases of analyzing the application characteristics of Analytic HierarchyProcess, Triangular Fuzzy Analysis, Fuzzy Comprehensive Evaluation and Evidence Theory,the BP Neural Network was elaborated to evaluate flight safety risk. Then algorithm of BPNeural Network was described, and the improved algorithm of BP Neural Network was alsodescribed because of the slow convergence speed, low learning efficiency and instability inthe training process of standard BP Neural Network algorithm.Then, in order to overcome the influence of subjective factors from experts, thequalitative indexes were quantified by Entropy Method using MATLAB software, and to beused as input numbers of the network mode. Fifty samples were selected to build a networkmodel and four samples were selected to test the feasibility of the network model. The meansquare deviation of the network model is 0.0036 and the standard deviation is 0.06 using theMATLAB software programming, which met the evaluation accuracy and method’s feasibilitywas demonstrated.Finally, the sensitivity analysis of the flight safety indexes was studied, and the keyindexes was identified, which included safety and technical training management, aircraftmaintenance standards, implemention of emergency management training, quality of aircraftmaintenance personnel and illegal operation of airplane of aircrews, and corresponding measures of this key indexes was put forward, Which was of certain directive significance forstrengthening flight safety management, could improve flight safety risk level and ensureflight safety.