Abstract: The sensitivities of autoregressive coefficients to element stiffness are deduced for damage detection. It is presented that the sensitivities of autoregressive coefficients to element stiffness can be enhanced by assigning the poles of system under detection internationally, which may improve the accuracy of damage detection. A damage detection method using dynamic response data of structures under control and control charts is proposed in the paper. In the study, the principal component analysis is first carried out on all response time histories of a closed-loop system for data compression. A time series model with a fitting order is then constructed using the fist principal component data. Finally, X-bar control charts are built based on the mean values of autoregressive coefficients. The statically significant change of the control charts are employed to judge the damage occurrence. The efficiency and capability of the proposed method is demonstrated by applying the method to a cantilever beam. Numerical simulations show that the control chart is efficient for the identification of early small damage with measurement noise as the sensitivities of autoregressive coefficients are enhanced by feedback control.