Electricity load forecasting using artificial neural networks
Abstract
Electricity load forecasting has become increasingly important for the power industry. To
generate "what is reasonably required" one needs forecast the future electricity demands.
However, the accurate load prediction remains a challenging task due to several issues
such as the nonlinear character of the time series or the seasonal patterns it exhibits. The
objective of the study is to use Artificial Neural Network to solve the problem for
Kenya's Electrical Power sector. The tool used in this study attempts to address the
challenge of nonlinearity of the load function.
In this project "day ahead short term load forecast" is studied using sample data. Raw
data is collected cleaned and loaded into the proposed model to forecast the electricity
needs for the next day. The test results showed that the hour-by-hour approach is more
suitable and efficient for a day ahead load forecasting. The work suggests that
incremental training approach of a neural network model should be implemented for online
testing application to acquire a universal final view on its applicability.
Conclusion is then drawn based on the data collected to ascertain whether the proposed
model addresses the issue of nonlinearity in the load function. Also discussed under this
project are other popular techniques that have been used to address the short term load
forecasting problem but are only analyzed by means of explanatory methods. The study
then suggests future research directions in the area of electricity load forecasting.
Citation
Master of Science in Information SystemsSponsorhip
University of NairobiPublisher
University of Nairobi School of Computing and Informatics