For the island partition of distribution network considering the output fluctuation of distributed generators (DG) and load uncertainty

a new optimal island partition model and its solving method are presented based on tree knapsack problems (TKP). The basic idea of the paper is first

a new stochastic island partition model considering the uncertainty of DGs and loads is developed. Then the Latin hypercube sample method

the deterministic tree knapsack algorithm and mathematical statistics method are adopted to determine the composition of the initial island

including DGs and loads. Finally

for each initial island

the stochastic optimal power flow is used to optimize and adjust the initial island on the basis of island reconfiguration. Through the above optimization and adjustment

the security

optimization and economy for each DG island operation are ensured. The method has achieved optimal islanding under the uncertainty of DGs and loads

the islands most likely to appear according to the real-time situation

which has a stronger credibility compared with the condition of certainty. The results of an PG&E 69-bus distribution system with several DGs show the validity of the proposed model and method. This work is supported by National Natural Science Foundation of China (No. 51177107).