To implement demand response in residential sector in smart grid, a control algorithm, which controls air conditioning systems in smart grids according to predicted outdoor temperature and electricity price published by power company, is proposed. For this reason, on the basis of power consumption expense index, an index reflecting the comfort degree of air conditioning users quantitatively is led in, and a multi-objective optimization model, which optimizes power consumption expense index and comfort degree index of air conditioning users, is established. During the model, the Monte Carlo simulation and scenarios reduction technique are utilized to cope with the uncertainty due to the error of predicted outdoor temperature. The improved particle swarm optimization algorithm is used to solve the proposed model. Simulation experiments of the proposed algorithm are performed, and simulation results are compared with those by both traditional constant temperature control mode and single-objective control algorithm, which takes the minimized power consumption expense as the objective. Experimental results show that the proposed multi-objective optimization algorithm can well cope with both economical demand and demand on comfort degree of users simultaneously, thus the effectiveness of the proposed algorithm is validated.