Forecasting the Numbers of Pole-Mounted Transformer Failures via Tripple Exponential Smoothing

Availability of adequate numbers of spares of pole-mounted transformers can help to reduce the impact of failures of these transformers on customers. Forecasting can assist to determine the optimum number of these spares to be kept so that funds are not locked in unproductive assets and not less than adequate number of spares are available so that responses to failures can be as quick as possible. This research presents the work on using the exponential smoothing methods of forecasting a time series to forecast the quarterly numbers of pole-mounted transformer failures. To derive the smoothing parameters, the Excel Solver is used as the optimisation tool. The ATES outperforms the MTES and the multiplicative decomposition models but is itself outperformed by the additive decomposition model. The MTES is the least performing of all the four models considered in this work. With a view to improving the performance of TES models, it is proposed that the potential of metaheuristic methods in determining the optimal parameters of the TES models be investigated.