Short-Term Load Forecasting Considering the Separation and Identification of Generalized Load

Abstract

Affected by the distributed generation (DG), traditional load forecasting models have been difficult to forecast generalized load (GL). In this paper, a short-term load forecasting method considering the separation and identification of GL is proposed. First, the key influencing factors of GL are obtained by grey relational analysis, which are mainly temperature and DG. Secondly, the proposed improved back propagation (BP) neural network is used to realize the separation and identification of temperature-sensitive load (TSL) and DG in GL. Finally, long short-term memory (LSTM) network is used for TSL and DG output forecasting, Autoregressive Integrated Moving Average (ARIMA) model is used for normal load forecasting. The short-term GL forecasting results are obtained by summing. The practical example shows the effectiveness of the proposed method.