Residential Load Forecasting for Flexibility Prediction Using Machine Learning-Based Regression Model

Flexibility is already an important concept in the context of electricity balancing. Residential customers' consumptions are usually highly volatile and depend on individual behaviour; hence, it complicates forecasts and flexibility extractions. This paper proposes the use of machine learning based regression models to generate load patterns for forecasting the potential flexibility of residential customers and improving both technical and economic smart grid operations. The advantage of proposed method is that it can be used, in on-line and real-time methods, in a wide range of control approaches.