Image-Based Processing Mechanism for Peak Load Forecasting in Smart Grids

Smart homes dispatchable loads provide an opportunity for utility companies to develop demand response mechanisms to balance the demand and supply of energy during peak times. However, in order to determine the right time to dispatch/curtail energy from smart homes, efficient energy forecasting algorithms are needed to precisely determine the peak times. This paper proposes a novel forecasting system of daily electricity consumption prediction based on image processing of load curve structure. While there are several algorithms that deal with energy forecasting, one major challenge is that most of these algorithms are prone to high number of errors when predicting abnormal energy consumption days (e.g. shoulder season days, holidays, etc.). Our focus in this study is on reducing the peak prediction error in such days. Our system proposes a classification model based on the convolution image process and k-mean clustering to select better training sets and optimize the forecasting mechanism. The description and results of our model are captured in this paper.