Abnormality Detection in a Landing Operation Using Hidden Markov Model

The air transport industry is seeking to manage risks in air travels. Its main objective is to detect abnormal behaviors in various flight conditions. The current methods have some limitations and are based on studying the risks and measuring the effective parameters. These parameters do not remove the dependency of a flight process on the time and human decisions. In this paper, we used an HMM-based method which is among the main methods of situation assessment in data fusion. This method includes two clustering levels based on data and model. The experiments were conducted with B_777 flight data and the variables considered in the next generation of ADS_B. According to the results of this study, our method has high speed and sensitivity in detection of abnormal changes which are effective in the flight parameters when landing. With the dynamic modelling, there is no dependency on time and conditions. The adaptation of this method to other air traffic control systems makes its extension possible to cover all flight conditions.