Abstract: Flight safety is the core of air traffic safety management;hence,it is important to develop effective evaluation methods to assess the risk of flight safety.A systematical framework combining two theoretical models including fuzzy cognitive map (FCM) and analytic network process (ANP) is proposed to tackle the interdependence modeling problem among different evaluation criteria.Firstly,criteria interdependence among difference factor groups are discussed and measured utilizing the FCM-ANP hybrid model to furnish the construction of criteria interdependence measurement matrix.Based upon the obtained matrix,ANP is employed to calculate the stable criteria weights and FCM is used to identify the converged state of criteria values.The decision making trial and laboratory method (DEMATEL) is then implemented to analyze the key criteria factors influencing flight safety and based on the identified key factors,different improvement scenarios can be designed and their effects simulated through FCM model to select the best risk mitigation policy.Finally,a case study adapted from a real airline company's operational data is used to demonstrate the proposed method and verify its feasibility.

Key words: flight risk evaluation, fuzzy cognitive maps (FCM), analytic network process (ANP), decision making trial and evaluation laboratory (DEMATEL), criteria influence relationship matrix