Predication Model optimized by an Intelligent Algorithm for Energy Consumption
The steadily rising costs of energy is naught but wrought by man’s own doing; our over-reliance on precious,
unrenewable fuel to power everything we use has led to a slow bleeding of a very finite resource pool. This paper proposes
one such measure that utilizes the nascent Internet of Things framework in a lightweight system that is capable of
manipulating and monitoring energy usage at a localized level, thus providing vital information that would aid in an optimal
configuration of a building-wide energy dissemination system. In this work, an improved hybrid model based on the
Autoregressive Integrated Moving Average (ARIMA) and Gaussian Sum Particle Filtering (GSPT) is proposed to predict
energy usage. Observations with similar features can be grouped in the same cluster using clustering-based algorithm which
is inspired by Imperialist Competitive Algorithm (ICA) and k-nearest neighbor (kNN) classification, which is referred to as
ICA-kNN, and it is commonly used to reach optimum clustering N objects into K clusters. ICA-kNN algorithm is used to
develop the hybrid model and optimize the ARIMA model parameters. The ICA process is not only able to immensely cut
the computation load, but also to enhance the model performance. By combining the ARIMA and an optimized ICA-kNN
algorithm, the GSPF accomplishes the best performance. Moreover, proposed model constructs an accurate tool to predict
the global energy consumption issues which has not been encountered effectively so far.
Keywords: Energy consumption, Prediction, ARIMA, Intelligent algorithm.