Amid the context of a sustainable development strategy, there is a growing interest in renewable energy as an alternative to traditional energy sources. However, as the penetration rate of clean energy gradually increases, its inherent features, such as randomness and uncertainty, have led to a surging demand for flexibility and regulation in power systems, highlighting the need to enhance the flexibility of power systems in multiple dimensions. This paper proposes a method for evaluating the adjustable power capacity of a virtual power plant (VPP), which considers the high-energy-consuming industrial load in the day-ahead to real-time stages and establishes an optimization scheduling model for auxiliary service markets based on this method. Firstly, within the day-ahead phase, the VPP is categorized and modeled based on its level of load flexibility regulation. The assessable capacity is then evaluated to establish the adjustable power range of the VPP, and the capacity of the VPP is subsequently reported. Secondly, the adjustable loads inside the VPP are ranked using the performance indicator evaluation method to obtain the adjustment order of internal resources. Finally, on the real-time scale, an optimization scheduling model to minimize the net operating cost of the VPP is established based on real-time peak-shaving and frequency regulation instructions from the auxiliary service market and solved using the CPLEX solver. The case study results show that the proposed method effectively reduces the net operating cost of the VPP and improves the stability of its participation in the auxiliary service market, which verifies the effectiveness of the proposed method.