摘要：Since 2020, the COVID-19 has spread globally at an extremely rapid rate. The epidemic, vaccination, and quarantine policies have profoundly changed economic development and human activities worldwide. As many countries start to resume economic activities aiming at a “living with COVID” new normal, a short-term load forecasting technique incorporating the epidemic’s effects is of great significance to both power system operation and a smooth transition. In this context, this paper proposes a novel short-term load forecasting method under COVID-19 based on graph representation learning with heterogeneous features. Unlike existing methods that fit power load data to time series, this study encodes heterogeneous features relevant to electricity consumption and epidemic status into a load graph so that not only the features at each time moment but also the inherent correlations between the features can be exploited; Then, a residual graph convolutional network (ResGCN) is constructed to fit the non-linear mappings from load graph to future loads. Besides, a graph concatenation method for parallel training is introduced to improve the learning efficiency. Using practical data in Houston, the annual, monthly, and daily effects of the crisis on power load are analyzed, which uncovers the strong correlation between the pandemic and the changes in regional electricity utilization. Moreover, the forecasting performance of the load graph-based ResGCN is validated by comparing with other representative methods. Its performance on MAPE and RMSE increased by 1.3264 and 15.03%, respectively. Codes related to all the simulations are available on https://github.com/YoungY6/ResGCN-for-Short-term-power-load-forecasting-under-COVID-19.
