Recurrent Neural Network Model for Forecasting Electricity Demand in Nigeria
No Thumbnail Available
Date
2016-12
Authors
Abdulsalam, K.A.
Adegbenro, O
Akinbulire, T.O.
Journal Title
Journal ISSN
Volume Title
Publisher
Journal of Engineering Research
Abstract
This work uses modular recurrent neural network to estimate the electricity demand in Nigeria from 2015 to 2050. The network is a 2-layer multi-input, single-output model with twelve neurons trained using Levenberg-Marquardt algorithm. The data structure used for training is cell array of sequential concurrent data. The Recurrent Neural Network model was simulated as Non-linear Auto Regressive with eXogenous (NARX) model in Matlab environment and the predicted load for 2015 is about 550GWh; and an expected demand increase of 7.5 % every five year.
Description
Keywords
energy, Levenberg-Marquardt, modular network, neurons, power