Abstract: In smart grid, the demand response applied to air conditioning load, which is the major load of the commercial and residential electricity system, is of great significance for the stable operation of power network. However, participating in demand response may lead to the reduction of consumers comfort. Thus an efficient optimal control method of air conditioning load is put forward, which allows the decision-making of electricity consumers, and a multi-objective optimization model is built based on the improved immune clonal selection algorithm, which optimizes comfort and electricity expense of air conditioning users. In this paper, the mutation operator of the traditional immune clonal selection algorithm is improved as an adaptive non-uniformity mutation operator, which further enhances the convergence of the algorithm and gets a more optimal Pareto-frontier. Results of simulation and tests indicate that multi-objective optimization can effectively balance the users demand for economy and comfort during the process of demand response based on Time-of-Use (TOU) price. Moreover, the antigen affinity can be optimized by improved algorithm, which verifies the effectiveness and advantage of the improved algorithm.