Abstract
In this paper, load forecasting as applied to a
medium voltage distribution power network servicing a South
African hospital facility is conducted using statistical time series
model. Historical measurements of active power data recorded
over a period of three months are used for this purpose. The
R-Studio package software is sourced to examine the shape of
active power time series pattern. The Box-Jenkins seasonal autoregressive
integral moving average (ARIMA) model is applied
to forecast future active power series data. The accuracy of this
prediction is verified on the basis of the mean absolute percentage
error (MAPE). Results show MAPE deviation of 3.91% from
actual load data measured.