Abstract: Aiming at the problems of large network loss, high voltage offset and unbalanced load distribution in urban cable double loop network, a dynamic recombination method of double loop network considering the characteristics of load period is proposed. Firstly, a double layer topology model of trunk loop network layer and branch network layer is constructed according to the cable double loop network structure. Second, the restructuring of moment is presented based on the power, and a method for determining the number according to the imbalance of power moment degree determine daily load restructuring changes, and puts forward the improved double truncated gaussian mixture distribution model of daily load time, the periodic distance instead of traditional model of Euclidean distance, improved the model of time division accuracy. Thirdly, the load recombination optimization model was constructed with the minimum network loss, the lowest voltage offset and the highest load balance as the indexes, and the bus load recombination scheme and branch load recombination scheme were solved respectively by using binary particle swarm optimization algorithm at the trunk ring network layer and branch network layer. Finally, the effectiveness and practicability of the model are verified by an example in a city cable double loop network in east China.