Abstract: The research of loop closing current assessment has great significance for distribution network feeder loop closing operation to supply power. In order to enhance the accuracy of active distribution network feeder loop closing current calculation, an active distribution network feeder loop closing current assessment research method based on double K2 algorithm and probability load flow (DK2-PLF) is proposed in this paper, which also considers the prediction and distribution characteristics of source-load data. Firstly, Bayesian network based on DK2 algorithm is suggested to characterize the correlation of source-load samples. Secondly, the Cholesky decomposition method is proposed to deal with the predictive source-load samples, and the cumulative probability distributions of the loop closing current are calculated by cumulant method. Thirdly, the security assessment of the loop closing current is evaluated in terms of the success rate and the degree of exceeding the limit. Finally, taking a city in Guizhou as an example, the following conclusions are drawn from the study of 10 kV distribution system. Comparing from the aspects of probability density, cumulative distribution, and maximum error, the probability density and cumulative distribution errors of the DK2 algorithm are smaller than those of the K2 algorithm's predicted values, which verifies the superiority of DK2 algorithm. Comparing from the aspects of cumulative distribution, and maximum error, the cumulative distribution error of the cumulant method with Cholesky decomposition is smaller than that of the cumulant method without Cholesky decomposition and Monte Carlo method. The cumulant method with Cholesky decomposition meets the requirements of active distribution network feeder loop closing current assessment. Comparing from the aspects of loop closing success rate and the degree of exceeding the loop closing limit, the safety index of loop closing current computed by the cumulant method deviates from the simulation results is within 5% of the grid empirical error, which indicates that the active distribution network feeder loop closing current assessment method based on DK2-PLF can provide a reference for loop closing assisted decision-making.