Abstract: In order to consider the variability of transformer voltage and improve the accuracy of topology estimation and practicality of fault detection, a distribution network topology recognition and fault detection method based on constrained recursive grouping algorithm is proposed. A new variable was introduced to measure time-varying transformer voltage for estimating voltage sensitivity coefficient, while considering changes in transformer voltage. A distribution network topology recognition method based on constrained recursive grouping algorithm was proposed to ensure generation of appropriate metering node pairs. We introduced an improved graph similarity metric that can accurately capture global changes, and designed an online data-driven fault detection algorithm. The effectiveness of the proposed method was verified through experiments.