The estimate of peak load shifting potential is the foundation of fine management of orderly electric utilization. Based on the peak-valley characteristics of electricity user data, fuzzy C-means clustering is employed to classify consumers in the same industry, then the industry's typical consumers are filtered out from the clustering result according to the peak-valley weight and the annual electricity consumption of the same type of users. This provides a screening method of the typical consumers for the analysis of peak load shifting potential. A case study including 36 industries is carried out; 146 typical consumers are screened out. Due to the limited space, this paper only describes the screening process of three typical consumers and peak load shifting potential in the general and special equipment manufacturing industry. The results prove the feasibility and effectiveness of the proposed screening method.