Comparison of Deep Learning-Based Methods for Electrical Load Forecasting

Abstract

The introduction of Artificial Intelligence based methods for forecasting the load in power systems has shown remarkable results in terms of accuracy. A proper forecast of the load ahead can be beneficial in terms of planning, scheduling, and regulating the usage of power to minimize its cost of generation, wastage and to improve the system reliability. Numerous AI-based methods have been used for the purpose of forecasting the load. In this paper, three deep learning algorithms namely Long Short-Term Memory (LSTM) network, Gated Recurrent Unit (GRU) and Vanilla Recurrent Neural Network (RNN) were used for load prediction. The dataset has been taken from PJM (East Region). The accuracy of models was examined and compared on the basis of values obtained for Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE). The experimental results show that the LSTM is reliable and accurate than other two models for the forecasting of electrical load in a power system.