Abstract: At present, the scale of distributed power connected to the distribution network is increasing rapidly. To reduce the influence of reactive power disturbance on power quality during normal grid-connection and improve the accuracy of islanding detection under disturbance,this paper presents an islanding detection method based on wavelet packet fuzzy entropy and reactive power disturbance. The method uses the wavelet packet transform fuzzy entropy to extract and amplify the high-frequency component feature information of the islanding state in the voltage signal of the Point of common coupling（PCC）,and completes the calibration of the suspected islanding state. After that,the reactive power disturbance is injected to make the frequency shift rapidly after the islanding occurs until the limit is exceeded. The simulation and data analysis are carried out by using PSCAD and Matlab to verify the accuracy and anti-interference of the proposed algorithm, and the simulation results show that the proposed method can accurately detect the islanding phenomenon in the multi photovoltaic access distribution network system and effectively prevent misjudgment under non islanding disturbance.