Abstract: A comprehensive evaluation method based on non-linear mapping and kernel principal component analysis was proposed in this paper. First, non-linear mapping was utilized to reduce the high-dimensional evaluation results of area distribution network to two-dimensions, and then the two-dimensional data was clustered by an improved K-means method. Secondly, kernel principal component analysis was used to reduce the dimension of the high-dimensional distribution network evaluation index and construct the evaluation system of distribution network from four aspects of economy, security, reliability and intelligence. The proposed method was applied in 108 counties distribution networks in a province, and the results can effectively reflected the status of the 108 counties distribution networks, identified their key problems, and verified the effectiveness of the proposed method.