The invention discloses a self-adaptive Robust optimization method for an active distribution network considering a large number of renewable energy connected grids. The method comprises the steps of estimation of a Robust feasible region, construction of a nonlinear self-adaptive function, construction of a self-adaptive Robust active-power-reactive-power optimization model, mapping of kernel function space, and solution with a cutting-plane method and an optimization operation strategy of the active distribution network. According to the method, to solve height uncertainty of the large-scale renewable energy connected grids, an idea of Robust optimization is introduced, and the safe operation of a power system under an optimal decision is guaranteed; a self-adaptive technology is introduced, and self-adaptive changes of the optimal decision with uncertain variables are achieved; a self-adaptive Robust active-power-reactive-power optimization model of the active distribution network is established, and the economy and efficiency of the distribution network under the optimal decision are ensured, a kernel function method and a cutting-plane solution strategy are adopted, the efficient and fast solution of an original problem is achieved, and the reliability, economy and safety during an operation process of the active distribution network are ensured.