Short-Term Load Forecasting Using an LSTM Neural Network

Abstract

In this paper, two forecasting models using long short term memory neural network (LSTM NN) are developed to predict short-term electrical load. The first model predicts a single step ahead load, while the other predicts multi-step intraday rolling horizons. The time series of the load is utilized in addition to weather data of the considered geographic area. A rolling time-index series including a time of the day index, a holiday flag and a day of the week index, is also embedded as a categorical feature vector, which is shown to increase the forecasting accuracy significantly. Moreover, to evaluate the performance of the LSTM NN, the performance of other machines, namely a generalized regression neural network (GRNN) and an extreme learning machine (ELM) is also shown. Hourly load data from the electrical reliability council of Texas (ERCOT) is used as benchmark data to evaluate the proposed algorithms.