Abstract: To enhance the accuracy of instantaneous frequency （IF） identification for non-stationary response signals of time-varying structures， a locally optimized multi-synchrosqueezing-short time fractional Fourier transform （LOMS-STFRFT） algorithm is proposed in this paper. Firstly， the local rotation parameters of the short time fractional Fourier transform （STFRFT） are optimally selected in this method. Subsequently， the time-frequency coefficient matrix projected to the fractional domain is obtained through STFRFT. After that， IF estimation and multiple iterations are performed on the time-frequency coefficient matrix. The time-frequency coefficient matrix is reassigned by the multi-synchrosqueezing operator， and IF curves are then extracted via the local mode maxima method. The accuracy of the proposed method is validated through a numerical example of a multi-component signal and a linearly time-varying cable test. The results demonstrate that the proposed LOMS-STFRFT algorithm behaves better than traditional multi-synchrosqueezing transform on IF identification of non-stationary signals from time-varying structures.