As more and more renewable generation resources are integrated into the smart grid, they are likely to cause supply-demand imbalance such as overgeneration due to their uncertain and variable nature, resulting in poor power quality and unreliable power delivery. To balance overgeneration, this paper presents a collaborative control strategy of thermostatically controlled appliances in the timescale of minutes. The proposed strategy includes four parts: state awareness, real-time analysis, scientific decisionmaking, and precise execution. A responsive task-type model is designed to identify the executable task type of an appliance in the state awareness part. A multiobjective 0-1 programming method is used to solve the optimal control combination in the scientific decision-making part. And a thermostat set-point control rule is used to generate the control signals of appliances in the precise execution part. In order to illustrate the advantage of the proposed strategy, four case studies are designed, and the advantage of the strategy is validated by the simulation results of the case studies.