A New Subtractive Clustering Based ANFIS System for Residential Load Forecasting

The load forecasting plays a pivotal role for buildings' energy management system to schedule and control loads effectively while maintaining the reliability, stability and quality of the utilized power within the buildings. This work proposes a new subtractive clustering based adaptive neuro fuzzy inference system (ANFIS) for residential load forecasting. The proposed ANFIS prediction system considers the temperature and a new variable calculated based on occupancy and week/special days for predicting the building's energy consumption. The performance of the proposed ANFIS system has been compared with that of the artificial neural network (ANN) based prediction method. Both ANFIS and ANN predicted data are simulated in MATLAB and results are tabulated in terms of performance indices. Simulations results demonstrate the efficacy of the proposed ANFIS system for residential load forecasting. Also, the performance of the proposed ANFIS method outperforms the ANN approach.