Abstract: Considering the deficiency of single algorithm in selecting typical daily load curve,this paper proposes a new integrated clustering method by combining the improved possible fuzzy C-means（PFCM）algorithm with fuzzy linear discriminant analysis（FLDA）. Firstly,the original PFCM is improved to get the improved PFCM,and it is applied to the selection of the optimal number of clusters. Then the improved PFCM is combined with FLDA,and the integrated clustering algorithm is applied to the clustering of the load curves. Finally,the validity of this method in selecting the typical days is verified by the annual load data of a power grid.