Gaussian Mixture Model based pattern recognition for understanding the long-term impact of COVID-19 on energy consumption of public buildings

At present, the structural transformation of energy demand of public buildings in the post-pandemic era is not well known, and there is also a lack of fine-grained research on energy consumption pattern identification of public buildings. To fill this gap, this research used the electricity dataset of public buildings in Scotland, and applied Gaussian Mixture Model (GMM) to explore the changes in electricity usage patterns throughout the pandemic, so as to understand the long-term impact of COVID-19 on energy consumption of public buildings. It was found that the basic electricity consumption of selected public buildings in the post-pandemic period not only continued the reduction trend identified in the pandemic period, but also would be likely to further reduce. The peak electricity consumption in the post-pandemic period rebounded to a certain extent, but it still could not reach the peak in the pre-pandemic period. The most significant change of the electricity usage pattern was found for office buildings, and the changed pattern continued into the post-pandemic period. The results provide important implications for policy makers to understand the demand-side changes of building energy consumption in the post-pandemic era, and to formulate supply-side adjustments accordingly.