There are problems of voltage violations and branch overload caused by the high proportion of renewable distributed energy in a distribution network. Thus a three-layer planning model of soft open point (SOP) is proposed based on network reconstruction and conditional value-at-risk (CVaR) theory. First, in order to simulate the output uncertainty of distributed generation, an optimal scenario is constructed based on Wasserstein distance. Secondly, the upper layer model determines the location and capacity of SOPs considering the two objectives of minimizing the annual comprehensive cost and security risk simultaneously; the middle level model carries out network reconstruction with the goal of minimizing the running cost of each scene; in the optimization model of the lower layer, active various regulation measures such as on-load tap changers, switching capacitor banks, demand response and power transmission of SOPs are considered. In order to reduce the calculation complexity of the model, the hybrid method of the LDBAS algorithm based on grey target decision-making technology and second-order cone optimization is used to analyze the three-layer planning model. Finally, the proposed planning model is verified and analyzed by taking the modified IEEE 33 bus distribution system as an example. This work is supported by the National Natural Science Foundation of China (No. 52067020 and No. 51667020).