An Efficient Load Disaggregation Methodology Based on Two-Phase Electrical Service Data for Residential Communities

Residential load estimation is one of the most important factors for power systems. With the development of smart grids and controllable loads, load disaggregation is crucial for load shifting, short-term residential load estimation, and demand response. The necessity for a methodology that can be used for accurate load disaggregation for large residential communities is urgent. We propose a novel methodology to disaggregate the loads in houses by using only a small fraction of the houses to train the model and by applying the trained model to the rest of the houses. The proposed methodology will cluster the houses into different groups, and the optimal number of clusters is decided during the process by knee point. Within each cluster, the methodology requires two-phase electrical service data of one house as input to train the load disaggregation model. Our proposed methedology is validated by numerical experiments with real-world data.