ABSTRACT
The paper aims to develop the risk prediction model for aircraft hard landing with flight data analysis. The statistic data shows nearly half of the incidents occurred in the aircraft landing stage and hard landing is one of the main contributions. The paper firstly analyzes the possible factors of the hard landing and the data feature of flight data. Secondly, flight data are preprocessed by height slice and Principal Component Analysis (PCA). It is to solve the prediction accuracy and data redundancy problems. Then the study builds the mathematical model with the objective function of maximize the probability of hard landing accidents through historical samples. An algorithm based on golden section was provided, and threshold values of each index were found. Finally, the proposed method is validated by empirical research. The result suggests that the proposed method is feasible in hard landing risk prediction problem.
