Abstract: With the large-scale integration of distributed generator (abbr. DG) into the distribution network, the randomness and volatility of load and DG output have brought big challenges to the operation of the distribution network. By means of the advantage of soft open point (abbr. SOP) in power flow control and reactive power compensation, a two-stage SOP robust optimization operation model, in which the minimized operation cost of active distribution network was taken as the objective function, was established. In the first stage, the operation strategy of SOP and the switch action conditions in each time interval were generated to minimize the operation cost. In the second stage, the worst scenario of the distribution network operation and the minimum operation cost under the corresponding scenario were found in the predefined uncertainty set, and by means of adjusting the robust regulation parameter the conservative degree of the operation scheme could be flexibly changed. Finally, by use of column constraint generation algorithm and duality theory, the primal problem was decomposed into the main problem and the sub-problem, and they were solved by interactive iteration to obtain the optimal solution. Taking the improved IEEE 33-bus system for example, both effectiveness and security of the proposed method are verified.