Abstract: Fully exploiting the schedulable potential of demand-side resources can provide reliable data support for power system scheduling. However, the schedulable potential of demand-side resources is closely related to user’s internal psychological factors, external environmental factors and power grid operating conditions. So it needs to be modeled and analyzed in detail. The air conditioning load and electric vehicle, which account for the largest part in the residential electricity load, are taken as the research objects. Firstly, a mathematical model was established from the user comfort and physical level to fully tap the scheduling potential of demand-side resources. Secondly, on the basis of schedulable potential, we adopted the direct load control mode in which load aggregators and residents sign an agreement, and took the maximization of the benefits of load aggregators as the objective function. The optimal bidding decision among various aggregators was solved by using the non-cooperative game theory. Simulation examples show that the model can not only stimulate flexible resources on the demand side to actively participate in the regulation of the power market to maximize the economic benefits of all parties, but also effectively reduce the peak-valley difference of load demand to relieve the pressure of balance between supply and demand of the power grid.