A novel daily peak load forecasting method using analyzable structured neural network

This paper presents a novel daily peak load forecasting method using an analyzable structured neural network in order to explain forecasting reasons. We propose a new training method for the analyzable structured neural network (ASNN) in order to realize accurate daily peak load forecasting and explain forecasting reasons. ASNN consists of two types of hidden units. One type of hidden units has connecting weights between the hidden units and only one group of input units. Another one has connecting weights between the hidden units and all input units. The former type of hidden units allows to explain forecasting reasons. The latter type of hidden units ensures the forecasting performance. The effectiveness of the proposed training method is shown applying to daily peak load forecasting. ASNN trained by the proposed new training method can explain forecasting reasons more properly than ASNN trained by the conventional method.