A novel active islanding fault detection based on even harmonics injection and set-membership filtering

This paper proposes a novel active islanding fault detection method based on even harmonics injection and set-membership filtering for the distributed generation systems. Traditional active islanding detections are mainly adding disturbances to frequency or phase of the voltage reference signal. The proposed active method detects islanding fault by injecting even-order harmonics into the current reference signal and monitoring the changes of corresponding voltage harmonics using set-membership filtering. This paper gives the principle of this proposed detection method and analyzes the range of injected even harmonics. We compare the differences of the even-order harmonic voltage between grid-connected and grid-off conditions. The proposed method has a number of advantages such as fast detection speed, almost no non-detection zone, and little impact on power quality. The effectiveness of the proposed method is demonstrated by MATLAB/Simulink simulations. Kalman filtering contrast experiments are presented to confirm that the set-membership filtering is more effective in islanding detection.