Comparison Study of Collaborative Learning Techniques on Residential Short-Term Load Forecasting

The deployment of the advanced metering infrastructure provides the opportunity for performing Short-Term Load Forecasting (STLF) for a single energy user. In the last few decades, Artificial neural networks (ANNs) have been widely implemented in STLF. Conventionally, a user trains an ANN only based on her/his own historical load data. In recent years, collaborative learning techniques have been applied to facilitate multiple users to train the ANNs to enhance the STLF performance that can hardly be achieved by the users individually. This paper presents a comparison study evaluating the performances of three state-of-the-art collaborative learning STLF methods on an Australian “Smart Grid, Smart City” residential power load dataset. The work is expected to reference researchers and engineers the practical implementation of STLF systems in the residential sector.