In order to solve the problems caused by analyzing users' load characteristics only and to achieve orderly power utility of demand side management, the load characteristic analysis method based on clustering method is proposed. In the view of the background of industrial enterprises DSM scheme, the fuzzy C means clustering method is applied, and the power load of the important enterprise users are clustered. The clustering validity function index based on probability distribution is chosen as the classification basis, and the load characteristics of a certain automobile manufacturing company are analyzed in detail, so does the clustering results. Analysis results show that, during the peak load period, a preliminary estimate of the demand response to events can provide weather sensitive load interruption of production capacity and load interruption capacity, and get the corresponding demand response types which suites to the needs. This work is supported by Science and Technology Program of Hunan Province (No. 2012WK3036).