The increasing penetration of distributed photovoltaics(PV) brings volatility and uncertain power outputs to micro-grids. Larger local regulation capacity is needed for maintaining the system balance between power supply-side and demand-side. It is promising to utilize widely distributed demand-side resources to provide regulation services, such as battery energy storage system (BESS), heating, ventilation, and air conditioning (HVAC), et al. However, most heterogeneous demand-side resources are regulated without coordination, resulting in the insufficient utilization of the regulation potential. To address this issue, this paper establishes a multi-time scale optimization model for micro-grids considering large-scale heterogeneous BESS and HVAC. Firstly, elements inside the urban micro-grids are modelled, where the HVAC systems and buildings are modelled as building based energy storage systems(BBESS), providing short-term energy storage. Then, a day-ahead optimization is carried out with the participation of day-ahead electricity market and ancillary market. Next, an intra-day rolling optimization is carried out in the real-time market. Finally, the case study shows that lower operation cost of the urban micro-grid and higher self-consumption rate of PV can be achieved by applying the proposed method, and BBESS can replace the demand for energy storage construction to a large extent.