Availa bility Prediction of the Repairable Equ ipment using A rtificial Neural Network, EWM A, AR, MA an ARMA Models

Availability is considered one of the most important criteria in public services quality. In this study, this criterion is evaluated using artificial neural network (ANN). In addition, availability values for future periods are predicted using exponential weighted moving average (E WMA) scheme and several time series models (T SM) including autoregressive (AR), moving average (MA), and autoregressive moving average (ARMA). Results obtained through the comparison of four methods based on ANN, consideringseveral conditions for the ejfective parameters in ANN, show that the generalized regression method is the best method for predicting availability compared to other existing methods. Furthermore, results obtained from E WMA method and the three aforementioned TSMs demonstrate that MA model outperforms other models in predicting the availability values in future periods.

Keywords:

availability , prediction , artificial neural network , exponentially weighted moving average , time series model