Abstract: Regarding the difficulties in detecting the partial blockage in underground drainage pipeline and the degree of blocking,a clustering recognition method based on acoustic features was proposed in this paper.Firstly,the sound pressure level was calculated from the acoustical pressure signals which were collected from the pipeline in order to increase the discrimination between blockage conditions.Then the comprehensive ensemble empirical mode decomposition (EEMD) was applied to the sound pressure level data ,the first 4 IMF components were selected using the Pearson's correlation coefficient and the energy proportion was extracted as the clustering feature.Finally,the principle component analysis (PCA) was adopted to proceed the dimensionality reduction onto the feature vectors,the FCM( Fuzzy C-means),GK(Gustafson-Kessel ),GG (Gath-Geva) algorithms were used to cluster feature vectors into classes and to further identify the blockage conditions.The experiment results have suggested that the proposed method is capable of identifying partial blockage conditions of drainage pipeline in different degrees,and presents a certain value for the engineering applications.