Externally bonded (EB) carbon fiber reinforced polymer (CFRP) is widely used in structural strengthening and retrofitting. Premature debonding of the FRP severely limits the efficiency of CFRP utilization. The application of CRRP anchorage system offers a solution to the debonding problem. However, the understanding of damage mode identification of this combined system still remains elusive. Acoustic emission (AE) technique is employed to identify the damage mode of this CFRP anchorage system, due to its high sensitivity and the ability to detect damage in real-time. The objective of the current study is to identify the failure mechanisms of CFRP-strengthened beam by applying advanced pattern recognition techniques to the collected AE data. Firstly, four-point test of CFRP-strengthened beam was carried out until failure with simultaneous recording of AE signals. Then, correlation analysis was adopted to select the AE characteristic parameters, and principal component analysis (PCA) was used for dimensionality reduction. Lastly, the AE signals of the CFRP-strengthened beam was clustered to track the evolutionary behavior of the different damage modes by Gaussian mixture model (GMM) algorithm. Three main damage modes of CFRP-strengthened beam were identified by GMM clustering: concrete cracking, debonding of CFRP sheet and fracture of CFRP sheet. This study explores the damage evolution mechanism of combined system and provides a basis for achieving health monitoring of CFRP-strengthened structures.@en