Traditional parameter identification methods are difficult to accurately identify the physical parameters (stiffness, damping) of time-varying (TV) systems. Aiming at this problem, a novel time-frequency algorithm based on adaptive polynomial chirplet transform (PCT) is proposed. First, the corresponding first and second derivative basis functions are derived based on the wavelet basis function of PCT. Second, based on the obtained derivative basis functions, PCT is performed on the acceleration signal, and the velocity and displacement signals can be reconstructed. In this step, the conversion rules of acceleration, velocity and displacement signals are obtained. Finally, the vibration differential equation of the TV system is transformed into a TV linear equation, and the physical parameters can be identified. Compared with the traditional identification methods, PCT can better track the TV parameters due to the introduction of the frequency modulation slope parameter. In order to verify the performance of the proposed algorithm, a numerical simulation under various TV conditions of a three degree of freedom TV structure is conducted. The results indicate the excellent identification accuracy of the proposed algorithm.