With the developmental trend of active distribution network, this paper presents a multi-objective reconfiguration model of active distribution network based on the improved genetic algorithm. First of all, the power loss, voltage offset and system stability margin are combined with the objective function, in which the weight of each index can be acquired through the judgment matrix method and the multiple problems will be transformed to single problem by using the linearity weighted method. Secondly, aiming at reducing the drawbacks of traditional genetic algorithm, it puts forward several methods for swapping chromosomes based on the principle of "properly matched marriage". In this case, the several methods can enrich the evolutionary ways of population, and the computational efficiency and the searching ability of the algorithm are greatly enhanced. Finally, the active distribution network will gain the ability of active reconfiguration and self-perception to improve the security and stability of the distribution network by setting the threshold. The results of simulation show that the model, algorithm and the concepts of active reconfiguration proposed in this paper are in accord with the new ideas of active distribution network, which have wide application prospects. This work is supported by National Natural Science Foundation of China (No. 51407186).