Research On The Detection Of Anti-islanding Effect In Photovoltaic Grid-Connected Power Generation System

Solar energy is becoming the most potential energy due to the rapid growth of the powerload, energy shortage and environmental pollution hazards, and grid-connected PV is thetrend for the way solar energy being used. But in the connection system, some questions cannot be ignored and islanding detection is notable. In this way, It is of great importance for theresearch of islanding detection.For the first part, this thesis mainly discusses questions existing in Grid-connected PVsystems and theories about islanding detection, including the standards, models and methodsnow being used. The emphasis has been put on the analysis and research of proactiveislanding detection on frequency drift of positive feedback. In a traditional way, islandingdetection would be failed and then caused an interference to the quality of grid power whenfrequency drift direction of disturbance signal is opposite to that of load resonant frequencyunder AFD.After that, to make up for the shortcomings mentioned in last part, this thesis proposedan improved islanding detection method, which is named as frequency drift fluctuationdetection method of positive feedback. The improved method adjusted the one-way frequencyoffset disturbance signal to make the disturbance semaphore in opposite directions within thetwo consecutive cycles. After selecting the specific direction conformed to local loadcharacteristic, inject disturbance semaphore continuously in that direction and insert a deadzone in the output current waveform when accumulated frequency offset is0.5%above orbelow the standard grid frequency(50Hz), you will find the output voltage frequency responsefluctuates and the voltage frequency is out of the standard grid frequency range soon. In thisway, we can avoid the situation that frequency offset is counteracted each other whenfrequency drift direction of disturbance signal is opposite to that of load resonant frequencycaused by changes in load characteristics, which will cause the detection frequency offsetvalue minish or vanish and then lead failure of islanding detection. By using this new method,we can reduce the impact of disturbance on grid and the adverse effect to the quality of power.In the end, this thesis tried an emulation of the method by the use of Matlab, and theresult therein proves the feasibility and effectiveness.