Abstract: With the proposed goals of "carbon neutrality and carbon peaking" and "building a new power system dominated by renewable energy", cleanliness and digitization have become urgent needs for the power system. Based on advanced control, metering, and communication technologies, the virtual power plant (VPP) aggregates massive diversified and distributed energy resources (DERs) on the demand side, which can provide necessary flexible support for the power system through diversified adjustment methods, and simultaneously facilitate renewable energy integration. The digital twin (DT) uses digital technologies, such as big data, cloud computing, and artificial intelligence to model the characteristics, behavior, process, and performance of the physical entities of DERs, which is an ideal way to optimize the operation of VPP. In this article, firstly, the technology development of VPP is prospected. Then, based on the concept, architecture, and characteristics of DT, a DT VPP system (DTVPPS) is proposed. The connotation of the DTVPPS is explained in-depth, and a framework based on physical entities, virtual spaces, DT data, connections, and intelligent applications is projected. Finally, the key technologies of DTVPPS are analyzed, and its future application practices are summarized and prospected.