Publications

Abstract : The use of time series forecasting of load has enhanced the operational reliability of power systems in recent years. Load forecasting technique is able to predict how the demand varied at the load side for a specific duration of time. This study compared the use of Auto Regressive Integrated Moving Average (ARIMA), Long Short-Term Memory (LSTM), and Recurrent Neural Network (RNN) models to forecast the load in a smart microgrid. Each model is trained and tested using five-year historical load data. To evaluate the results of the load forecasting models, the regression score (R2-Score) and Root Mean Square Error (RMSE) were considered. As a result, it is possible to identify which model is optimal for load forecasting in smart microgrid environment. LSTM model has shown superior performance than RNN and ARIMA models, since the predicted and actual plots have comparable properties and the R2-score value is approaching unity, demonstrating that it is suitable for load forecasting.