Predicting the potential risk in the operation of distribution network is great significance to the operation and maintenance of distribution network. Therefore

considers a method for predicting power outage failure based on weather factors

analyzes the correlation between the probability of power outages in the distribution network and weather variables and other possible factor variables

and obtains the fault characteristic variables.Based on the ReliefF algorithm

analyzes the correlation between the collection of fault characteristic variables and the probability of power outages in the distribution network to obtain the fault The importance degree of feature variables is ranked by weight.According to the main fault characteristic variables

forms a prediction model of distribution line fault blackout based on XGBoost algorithm

and evaluates and verifies the prediction results and by calculation examples.The example verification proves that this method can realize the prediction of the probability of power failure of the distribution network line

which is beneficial to the timely release of fault warning information and provides useful operation and maintenance guidance for the operation and maintenance department of the distribution network.