Abstract: The acoustic emission technology (AE) was used to monitor the tensile failure process of T300 carbon fiber composite material plate. Based on the test and progressive damage analysis,K-means clustering was applied to determine failure pattern of tensile damage signals of ［0°/90°］14 T300 carbon fiber composite material plate, and the clustering model was trained and tested under the supervision. The classification model was applied to the other similar data,and signals were well sorted.The above-mentioned method can thus provide a scientific basis for real-time monitoring of failure of carbon fiber composites.