With higher and higher penetration rate of renewable energy sources in the microgrid

the uncertainty of microgrid operation is increasing remarkably. So it is necessary to evaluate the impact of input variables on the state of system to improve the control level of system operation. This paper considers the uncertainty of distributed generators(DGs)and loads in the islanded microgrid

and proposes a robust Levenberg-Marquardt(LM)method based on P-f/Q-U droop control method to solve the probabilistic power flow(PPF)model of the islanded AC microgrid under integrated control strategy. Based on the uncertainty of system state

a global sensitivity analysis(SA)method based on Sobol' theory is proposed to analyze the sensitivity of islanded microgrid PPF. This method can identify the key random input variables influencing the system state. Finally

the 33-node islanded microgrid with several DGs is simulated. The first order sensitivity and total sensitivity of input factors are calculated to analyze the influence of input variables on the system state

and the effectiveness of the proposed method is verified.