Abstract: The large scale deployment of renewable distributed generation (DG) is generally seen as the most promising option for displacing fossil fuel generators, but the uncertainty and fluctuation of its output cause many problems to the distribution network. Soft Normally Open Point (SNOP) is a power electronic device which can solve those problems by adjusting reactive power compensation and active power transmission. In order to balance the safety and economy in power flow optimization better, a self-adaptive operation optimization method for multi-terminal SNOP in active distribution network based on real-time operation condition is put forward. Both weighted voltage deviation and the network loss are taken into objective function and the method mainly has two improvements: one is taking initial voltage deviation and penetration of the DG into consideration to build the self-adaptive weight model of the sum of voltage deviation; the other is putting forward the weighting strategy for node voltage deviation which decided by the electrical distance, DG position and correlation between source output and load. After that, three improved IEEE33 node feeders are used to verify the validity of the strategy.