In view of the uncertainty in the optimization of active distribution network(ADN)operation under high-penetration photovoltaic(PV)access

an adaptive robust optimization method based on approximate value function is proposed. The method considers the stochastic characteristics of PV output and the robustness of the system operation

ensuring the safe operation of the system through adaptive robust optimization and minimizing the operation cost at the same time. First of all

the optimal operation model is established considering various ADN adjustable devices

and the convex optimization technique is used to transform the original problem into a mixed integer second order cone optimization model. Then

the time decoupling adaptive robust optimization model is constructed by using the approximated value function of approximate dynamic programming(ADP)

and the successive projection approximation routine algorithm is used to train the value function parameters being used in the robust model. Finally

column and constraint generation algorithm is adopted to solve adaptive robust model for each period successively by means of approximated value function. Case study on modified IEEE 33-bus and PG&E69-bus distribution system is analyzed and compared with deterministic optimization

ADP and two-stage robust optimization method to verify the feasibility and effectiveness of the proposed method.