DATA PROCESSING ALGORITHMS IN THE PROCESS OF PILOT PREPARATION ON THE COMPLEX AIRCRAFT SIMULATOR

Abstract

The article discusses issues of improving the quality of crew training on the complex simulator of the An-148 aircraft. These studies are aimed at improving safety in special flight cases. The problem of the human factor is currently urgent. On modern simulators, actions are worked out in simple and difficult flight conditions. However, it is impossible to foresee all actions with simultaneous failures. Such failures also lead to increased stress of the human operator. The task is to train pilots not to degrade the quality of piloting technology in such situations. Modern digital simulators provide great opportunities for statistical data processing. Therefore, the introduction of new methods for automating the analysis of flight information and obtaining results from experimental data is up-to-date. Changes in flight parameters are random in nature when the pilot wants to clearly follow the given flight path. As a result of statistical processing of the roll angle data on the glide path, two models for describing the probability distribution laws were determined. In particular, when flying without failures or with single failure, it is proposed to use the normal probability distribution of the roll angle. With simultaneous complex failures, when their number is more than two, it is proposed to use the generalized Weibull distribution of the roll angle. This model for the roll angle indicates a deterioration in the quality of the piloting technique due to an increase in the psychophysiological stress of the aircraft pilot. The synthesis of two algorithms for detecting the fact of increased psychophysiological stress of the pilot in case of complex failures. One algorithm is based on the Neyman-Pearson criterion, and the other corresponds to the optimal Bayesian criterion. During detection algorithms synthesis, the assumption about harmonic component presence in the observed roll angle trend with complex failures was used. This assumption allows to simplify the structure of data processing algorithms.