Abstract: To balance the demand of users’ electricity and the demand of power grid dispatching and to improve the ability of demand response of temperature-controlled load as well as to further master the operation rules of temperature-controlled load so that to prearrange the dispatching plan, a multi time-scale dynamic correlation analysis model based on improved complete set of adaptive noise and time internal relatedness was proposed. By means of analyzing the dynamic association of load and environmental temperature time series under multi time-scale,its complex variation relation within partial time period was seized, thus both peeling off the fluctuant component of temperature-controlled load off from the load curve and the estimation of the proportion of temperature-controlled load were implemented. The application method of the proposed model was illustrated by the analysis of computing examples, meanwhile the effectiveness of the proposed model was verified.