Hourly electric load forecasting for buildings using hybrid intelligent modelling

Abstract

Because of the rapidly increasing total electric load of buildings, effective electric load management should be achieved quickly. This can be realized via electric load forecasting. In this study, a novel clustering-based hybrid prediction model is proposed to predict the 24-daily electric load of buildings. In this study, fuzzy c-means (FCM) clustering, ensemble empirical mode decomposition (EEMD), and some intelligent prediction algorithms are combined. FCM is used to extract the daily data exhibiting similar features, whereas EEMD is used for breaking down the optimal prediction algorithm is selected for each component, and the prediction results are integrated. When compared with the remaining conventional prediction models based on real data, the proposed hybrid model exhibits higher prediction accuracy.