Short‐term multivariate airworthiness forecasting based on decomposition and deep prediction models

Abstract

This study introduces a model for predicting airworthiness in terms of meteorology information within the viewpoint of not only formal regulations but also informal rules based on acquired indicators from flight training organization experience (AIs‐FTOE). The case study is carried out in the Hasan Polatkan Airport which is used by the Department of Flight Training of Eskişehir Technical University (ESTU‐P), which is also recognized as a flight training organization. Within the study, the constraints (derived from regulations and AIs‐FTOE) and the data set used in models are explained. Also, the models are introduced based on the gated recurrent unit (GRU) and long short‐term memory (LSTM) with the use of empirical mode decomposition (EMD) and variational mode decomposition (VMD). Finally, a model‐selective mechanism (MSM) is proposed to use the models in common. The findings show that the models presented in the study produce successful results that can be used in flight training organization's (FTO) planning studies. The MSM uses GRU and LSTM together with decomposition techniques to provide more advanced prediction capabilities. When the literature is examined, it is observed that although meteorological conditions are of vital importance in the efficiency of FTOs, there are not enough studies on airworthiness based on meteorology. So, a model that will assist in scheduling plans is presented for FTOs. Airworthiness analysis of forecasting can provide a comprehensive reference to support planning efficiency in FTOs. To the authors' knowledge, this study will be the first in the literature on airworthiness that presents the MSM using a hybrid deep learning algorithm and decomposition of time series models in concurrent.