New clustering-based forecasting method for disaggregated end-consumer electricity load using smart grid data

This paper presents a new method for forecasting the load of individual electricity consumers using smart grid data and clustering. The data from all consumers are used for clustering to create more suitable training sets to forecasting methods. Before clustering, time series are efficiently preprocessed by normalisation and the computation of representations of time series using a multiple linear regression model. Final centroid-based forecasts are scaled by saved normalisation parameters to create forecast for every consumer. Our method is compared with the approach that creates forecasts for every consumer separately. Evaluation and experiments were conducted on two large smart meter datasets from residences of Ireland and factories of Slovakia. The achieved results proved that our clustering-based method improves forecasting accuracy and decreases high rates of errors (maximum). It is also more scalable since it is not necessary to train the model for every consumer.