Intelligently managing energy production and consumption on community basis can enhance the demand side energy efficiency. The fact that the number of energy resources (especially controllable appliances) could be large imposes non-trivial computational burden to community-scale energy management. This paper proposes a community energy management framework that coordinately manages the operations of different kinds of energy resources in a community (i.e., photovoltaic solar power sources, battery energy storage systems (BESSs), and controllable appliances from different households) to minimize the community’s energy cost while sufficiently considering the peak-to-average ratio (PAR) of the community’s load and the occupants’ satisfaction. The proposed framework is with a two-stage design: in the day-ahead stage, multiple “virtual appliances” are formed, where each virtual appliance represents a group of appliances that have similar operational patterns. Based on this, a day-ahead scheduling model is formulated to determine the optimal operation plans of the virtual appliances and other energy resources. The virtual appliances’ schedules are then mapped to actual individual appliances through the developed mapping algorithms. In the real-time operation stage, a receding horizon-based energy management model is developed for correcting the BESS’s operation subjected to the day-ahead plan and real-time solar power output and uncontrollable load. Extensive simulation is conducted to validate the proposed framework based on a real-world dataset.