Abstract
A neural network modeling approach has been developed to analyze the pilots' landing control at visual approach. In the past study, the effectiveness of this method was proved, but only longitudinal motion was taken into consideration. Lateral motion is coupled to longitudinal motion and more refined network is necessary to be analyzed. In this paper, flight data is obtained in flight simulator, and some types of network with various inputs are modeled. The modeling skill of each network is evaluated using Monte Carlo landing simulation, and what types of network is the best is considered.
