A Hybrid Model of Clustering and Neural Network Using Weather Conditions for Energy Management in Buildings

Abstract

For the conservation of energy in buildings, it is essential to understand the energy consumption pattern and make efforts based on the analyzed result for energy load reduction. In this research, we proposed a method for forecasting the electricity load of university buildings using a hybrid model of clustering technique and neural network using weather conditions. The novel approach discussed in this paper includes clustering one whole year data including the forecasting day using K-means clustering and using the result as an input parameter in a neural network for forecasting the electricity peak load of university buildings. The hybrid model has proved to increase the performance of forecasting rather than neural network alone. We also developed a graphical visualization platform for the analyzed result using an interactive web application called Shiny. Using Shiny application and forecasting electricity peak load with appreciable accuracy several hours before peak hours can aware the management authorities about the energy situation and provides sufficient time for making a strategy for peak load reduction. This method can also be implemented in the demand response for reducing the electricity bills by avoiding electricity usage during the high electricity rate hours.