In order to solve the problem of pseudo-fault points in fault location in a distribution network with radial networks based on the single terminal impedance method, a fault section location method based on residual comparison of state estimation is proposed. This is based on the monitoring information provided by distributed intelligent measurement equipment in the network. This method first traverses all nodes and establishes the state equations of different node fault networks by adding faulty branches. Secondly, the total residual detection value sequence containing all nodes of the network is obtained by solving the state estimation equation. Then the faulty node is identified by detecting the node with the maximum residual detection value. Finally, the fault section is determined according to the phase difference of impedance angle between the head and the end of adjacent sections. A large number of simulations show that the proposed method produces good fault location with different fault types, transition resistance and other working conditions. The method takes into account the random variable components generated by the measurement, and uses the complementarity and multi-sources of the measurement data of different attributes to improve to some extent the anti-interference ability of different measurement data errors. This greatly reduces the fault finding range of the distribution network under limited measurement conditions. This work is supported by the National Natural Science Foundation of China (No. 51907134) and the Science and Technology Project of China Southern Power Grid (No. SGSW0000GHJS1900107).