Vibration monitoring based on distributed fiber sensing technology has great application prospects in the current linear engineering field. Asymmetric dual Mach-Zehnder interferometer (ADMZI) based distributed fiber sensing effectively extends the vibration sensing distance. However, the inherent asymmetry of the system leads to the failure of positioning by directly using the cross-correlation algorithm. By extracting the phase variation characteristics of the interference signals, the asymmetry of the sensing system can be eliminated, which in turn enables vibration positioning along the fiber. To further enhance the positioning accuracy, a new positioning method is proposed in this paper.In this study, with the ADMZI distributed vibration sensing system (Fig.1), a high-accuracy positioning method based on the wavelet-based synchrosqueezing transform (WSST) is proposed (Fig.2). First, endpoint detection is performed on the two interference signals, followed by the extraction of large bandwidth sections of the interference signals, whose 20 dB bandwidths are obtained through the power spectrum analysis. Then, the time-frequency analysis of the two interference signals is carried out using the WSST, and the effective phase variation characteristics are extracted adaptively in combination with the 20 dB bandwidths. At the same time, the asymmetry of the sensing system is eliminated. Finally, the time delay between the interference signals is obtained through the cross-correlation algorithm, and the vibration position is demodulated based on the ADMZI sensing principle.To verify the feasibility of the proposed positioning scheme, a series of field tests are conducted on the sensing system. A total of 100 knocking tests are performed at the end of the armored fiber optic cable over 128 km long sensing fiber, and the positioning results obtained using the CWT and proposed WSST positioning method are compared. The experimental results show that the ADMZI vibration sensing system achieves higher positioning accuracy when utilizing the WSST positioning method. The scheme can precisely locate vibrations over 128 km long sensing fiber with a positioning accuracy of approximately 35.43 m. In this paper, a high-accuracy positioning method based on the WSST is proposed to enhance the positioning performance of the long-distance ADMZI vibration sensing system. First, the time-frequency analysis of the two interference signals is performed using the WSST in combination with power spectrum analysis to extract the effective phase variation characteristics. The operation solves the inherent asymmetry of the ADMZI sensing system. WSST concentrates the energy of the time-frequency spectrum while eliminating the effects of substrate noise. Compared with CWT, WSST has higher time-frequency resolution and stronger noise immunity, which contributes to improving positioning accuracy. Adaptive characteristic extraction based on the bandwidth of interference signals simultaneously suppresses the disturbance of low-frequency noise. Then, a cross-correlation algorithm is performed to acquire the time delay between the interference signals using the extracted phase variation characteristics, and the vibration position is demodulated based on the ADMZI sensing principle. To verify the effectiveness of the proposed positioning scheme, a series of field experiments are conducted on the ADMZI sensing system. The experimental results show that the proposed scheme can precisely locate vibrations over 128 km long sensing fiber with a positioning accuracy of approximately 35.43 m. This scheme provides a new research direction for improving the positioning accuracy of long-distance distributed fiber vibration sensing technology. Furthermore, polarization effects and environmental changes introduce phase variations in the interference signals, known as phase noise. By investigating real-time phase noise compensation methods, the positioning accuracy of the ADMZI sensing system is expected to be further enhanced.