I. Introduction
With the deep integration of smarter sensing components and more complex communication and control networks in the power system, it has changed prominently toward future power system, also called cyber-physical energy system (CPES), which is combined power networks and cyber-physical systems (CPS), as shown in Fig. 1. The supply and demand balance between power suppliers and customers is the fundamental driving force for the development of CPES. Therefore, accurate load forecasting, especially short-term load forecasting (STLF), will play a key role in CPES by providing power suppliers with data support for the coordination of supply and demand, as well as automatic electricity generation control in modern power systems [1], [2].