Machine Learning and Deep Learning Techniques for Residential Load Forecasting: A Comparative Analysis

Load forecasting has become a very important parameter in modem power systems. These smart power systems require flexibility, smooth operation, scalability, and better demand-side management. Thus, making load forecasting is an essential thing. However, accurate load forecasting is a very challenging task as it involves variables such as the number of devices in the residential household and their many types, time, season, area, and occupants' behavior. In this study, a comparative analysis has been performed between different machine learning and deep learning-based residential load forecasting models. These models are trained based on the dataset of an Estonian household and they are tested, and forecasting has been made for a day-ahead load. Based on the simulation results, it was observed that Recurrent Neural Network (RNN) based algorithms give more accurate forecasting as it showed the lowest lower Root Mean Square Error (RMSE) value compared to other algorithms.