It is an important aspect of energy big data to transform the power quality monitoring data into the information which is useful for practical production. The most of voltage sag in the distribution network is due to fault and

therefore

the information of fault position is hidden in the residual voltage amplitude.For this purpose

a kind of fault location method of distribution network by the use of voltage sag monitoring data analysis is proposed in this paper. Firstly

the back propagation neural network（BPNN）for fitting the mapping relation between fault position and residual voltage amplitude is designed with the monitoring data as the input and the fault position as the output. Secondly

the fault position identification method based on the maximum output and node fault probability is proposed. Then

a fault locating method with off-line train and on-line application mode is proposed to perform on-line fault location by way of off-line trained neutral network. Finally

the effectiveness of the method proposed in this paper is simulated by and the influence of incomplete disposition of power quality monitoring device on the location accuracy is discussed. The method proposed in this paper

in case of on line application

is independent of the system parameters and has a definite fault-tolerant ability.