A kind of network-reconfiguration scheme based on dimension reduction and search is proposed to solve the oversize dimensions of optimal searching problem for large system that contains too many nodes and lines when using discrete particle swarm optimization. An improved ant colony and discrete particle swarm optimization (DPSO) are combined in the scheme to form skeleton networks. Firstly, an improved ant colony is used to obtain a trunk line, which could reduce the dimensions of the subsequent optimal searching. Secondly, based on the trunk line, discrete particle swarm optimization is employed in forming skeleton networks. The maximum quantity of the comprehensive recovered load, which makes important load into account, is defined as the optimization objective. The optimization must meet the constraint conditions, namely topological connectivity, safe and stable operation constrains. At last, the correctness and validity of the proposed scheme are verified by IEEE 118-bus system and part of Hubei Power Grid. This work is supported by National Natural Science Foundation of China (No. 51677137).