The invention discloses a power system load clustering method based on a fuzzy c-means algorithm. The power system load clustering method includes determining a statistics comprehensive molding theory based on a load structure; researching one of the clustering issues in load modeling: primary selection and refined selection of industrial users; researching another of the clustering issues in load modeling: classification and synthesis of comprehensive load static characteristics of a transformer substation; researching and determining the power system load clustering method based on the fuzzy c-means algorithm. Specific steps are subject to the abstract attached drawings. The power system load clustering method solves the problem of load clustering in load modeling of a power system and is an effective method for load clustering.