Abstract:
Load forecasting plays a paramount role in the operation and management of power
systems. Accurate estimation of future power demands for various lead times facilitates
the task of generating power reliably and economically. In this paper, econometric top
down modelling is used for the medium/long-term load and energy forecasting and an
Artificial Neural Network (ANN) and Box Jenkins model are used for short term load
forecasting.
The econometric model incorporates different sectors (residential, commercial, industrial,
own consumption etc) are related with the influencing factors such as customer
number/population growth, per capita consumption, GDP, economic growth, GNP, load
growth, demographic changes, electricity price, etc. After the energy demand is predicted
for the next 25years, the peak demand has also been forecasted along with system load
factor (57%).
The Artificial Neural Network Approach for Short Term Load Forecasting for Ethiopia is
studied with inputs parameters such as past 24 hours load, temperature, humidity, wind
speed, season (month) and day of the week to forecast 24 hours ahead load demands ( output).And, the result shows that it is reliable forecast as it includes load predictors even
if the correlation value is lesser than Box Jenkins model. In contrast, the Box Jenkins
model (AR, ARIMA, MA) has been selected based on the behavior of SAC and SP AC
after checking of stationarity of the time series.The result shows that the AR Box Jenkins
Model has lesser value of MAPE (10.35 %) comparing with ARIMA model. Moreover,
the AR Box Jenkins model has less value of MAPE and MEFV than the artificial neural
network model.
The contribution of karadobi hydropower on overall I.C.S of Ethiopia has been studied
