Efficient Neurofuzzy Model to Very Short-Term Load Forecasting

Since adaptive neurofuzzy inference systems are universal approximators that can be used in prediction applications, this study aims to determine the best parameters and their best architectures for the purpose of performing very short-term load demand forecasting in power distribution substations. The system inputs are load demand time series, consisting of data measured at five-minute intervals over seven days. Several input configurations and different architectures were examined to make the forecasting a step forward. The results provided by the adaptive neurofuzzy inference system in relation to the approaches found in the literature are promising.