Abstract: A PMV (predicted mean vote)-based dynamic thermal comfort (cool/hot) complaint event model and an energy consumption model are proposed for the control method in which PMV values change alternatively between comfortable and energy-saving zones. An improved multi-objective algorithm based on discrete PSO (particle swarm optimization) is applied to calculating optimal values of parameters in dynamic comfort control system according to the balance (specified by users) between comfort and energy conservation. This method only needs to measure data of thermal environment and occupant's thermal sensation, without building the physical analytic model. Experiment results demonstrate the effectiveness of the proposed control method. In addition, the realizability of the optimal control to dynamic comfort is also verified.