Abstract: In the occasion that there is no measuring device at the zero injection node in low-voltage distribution network (abbr. LVDN), the data of measurable node is single and its accuracy is unsatisfied, so the line impedance is difficult to accurately estimated, for this reason, based on time-sharing perception and partitioned backtracking an impedance estimation method was proposed. Firstly, the simplified typical LVDN was topologically divided into blocks and at the block where the user was located the time series superposition equation set was constructed to obtain the time-scale impedance sequence. Secondly, the fuzzy entropy was used to calculate the disorder of impedance sequence to achieve LVDN user-side topology verification. Thirdly, the silhouette coefficient density-based spatial clustering of applications with noise (abbr. SCD-BSCAN) algorithm was proposed and by means of the clustering of impedance sequence the perception of high quality measured data was realized and on the basis of high quality measurement the partitioned impedance estimation was conducted to determine the impedance boundary of each segment of the line by backtracking. Fourthly, a non-linear impedance estimation function containing impedance boundaries was established to realize the accurate estimation of each segment of the line. Finally, results of calculating example show that compared with existing methods, the identification precision of line impedance of LVDN can be improved by the proposed method.