Prognostic plays an important role in predictive maintenance and reducing operational cost in avionic industries. Along with it, model complex technical problems using data-driven method. This paper conducted extensive analysis of various state-of-art data-driven prognostic model. From analysis it can be seen the traditional machine learning (ML) based prognostic and predictive model are not efficient for dynamic requirement of aviation industries. Deep learning (DL) and recurrent neural network (RNN) has been applied for building prognostic model. However, these models are not efficient for modeling short contextual sequences as they are trained using historical data. Further, it is slow and time consuming. For reducing time and modeling short context sequence Extreme Learning Machines (ELM) is used. However, these models are not efficient when data is linearly non-separable. In general, the avionic data are imbalanced (i.e., very less fault data and more normal data). Thus, from extensive analysis carried out shows there is need for new model that improves and accuracy and reduce computation time. Thus, for meeting research challenges, the future work would consider building an accurate and efficient hybrid data-driven prognostic and predictive model for industry.
Volume 11 | Issue 7
Pages: 219-225