The invention provides a commercial load characteristic clustering analysis method based on quadratic clustering, comprising the following steps: (1) determining the upper and lower limits Kmax and Kmin of a clustering number K through a system clustering method, setting parameters, and letting K=Kmin; (2) initializing a K-means algorithm; (3) carrying out clustering to get the new-generation population individual fitness; (4) determining whether a fitness function converges, outputting a final clustering center and a final clustering result if the fitness function converges, or going to step (3); and (5) judging whether K is greater than Kmax, determining a most suitable K value if K is greater than Kmax, or, adding 1 to K and going to step (2). Initial K value setting and local convergence of the K-means algorithm are improved, and the clustering result is made more accurate and objective.