Load Forecasting Based on Improved Long Short-Term Memory Artificial Neural Network

To tackle the problems caused by randomness and uncertainty in load prediction in smart grid, this paper proposes an innovative model by combing grey correlation theory and long short-term memory (LSTM) algorithm. The proposed method firstly studies the correlation between input variables and the load dataset so as to eliminate redundant data as well as less relevant variables. Afterwards, the study uses a LSTM artificial neural network to predict the load by setting up a load forecasting model. In this paper, the actual data from a local distribution power grid is used for evaluating the effectiveness of the proposed method. The results show that the prediction accuracy is improved by comparing with other three traditional methods.