Research On Islanding Detection Strategy Of Grid Connected Photovoltaic Power System

Islanding of photovoltaic system may lead to the deterioration of power quality,damage of power system equipment and other problems,the relevant standards at home and abroad require that photovoltaic system should be equipped with islanding protection device.Traditional islanding detection methods are mainly focused on single inverter system,few research includes the islanding protection of multi-inverter based photovoltaic power plant.There will exist detection failure problem when applying traditional islanding detection strategy in photovoltaic power plant,new islanding strategy suited for photovoltaic needs to be researched.The thesis builds an equivalent simplified simulation model of photovoltaic power plant based on the actual photovoltaic power plant system,the model is the base of simulation verification.The influence factor of the change of voltage amplitude and frequency before and after islanding is analyzed,the failure mechanism of traditional passive islanding detection methods is concluded.Take frequency-reactive power feedback method as an example,when applying traditional active methods in the islanding detection of photovoltaic power plant,the chosen of feedback factor,the mutual interference of disturbance signal and other factors will lead to detection failure.The thesis proposes an islanding detection suited for photovoltaic power plant based on the regulation of the outputted reactive power of SVG.When the outputted active power of photovoltaic power plant is below the threshold,taking cyclical fluctuation in the certain interval of power factor at PCC as constrain factor,regulate the reactive power outputted by the SVG to lead to the active mismatch of the reactive power of the photovoltaic power plant and local loads.The reactive power outputted by the SVG cause little effect to the normal operation of the power system,when the photovoltaic power plant starts islanding operation,the mismatch of reactive power will lead to change of frequency exceeding the threshold in short time and islanding will be detected quickly and correctly.In order to reach the purpose of avoiding setting the thresholds of reference quantities manually and narrowing the islanding non-detection zone,the thesis propose an islanding detection suited for photovoltaic power plant based on Adaptive boosting intelligent algorithm.Adaptive boosting(Adaboost)intelligent algorithm combine some low accuracy classification models into a high accuracy classification model,the model is formed by learning the study data set which is composed by the reference quantities of different operation situation of photovoltaic,the model can distinguish the system operating at which of two operation situations,islanding situation or grid connected situation,the model is verified by the test data set which come from the same source with the study data set.Input the reference quantities of the real-time operation situation of the system,the model will output the system's operation situation,this will detect the islanding of the photovoltaic power plant correctly.The method avoids to set the threshold of reference quantities manually,and it will have a high detection accuracy by detecting the islanding with several reference quantities...