Study On Isolation Detection Method Of Distributed Photovoltaic Grid-connected System

With the development of new energy sources and the widespread application of distributed generation(DG),the safe and reliable operation of distribution networks has been challenged,making island state detection technology a hot topic for scholars at home and abroad.Distributed grid-connected technology has the characteristics of flexible operation and wide application range,especially suitable for photovoltaic power generation.In order to give full play to the advantages of distributed power supply and overcome the problems existing at this stage,this paper studies the isolated island detection method of distributed photovoltaic grid-connected system.The island operating state of distributed power has great security risks for electrical equipment and personal safety.The island detection is to select the correlation between the point of Common Coupling(PCC)voltage and the output current of the DG before and after the island state as the feature quantity to identify the island operation state of the DG.The current island detection methods are roughly divided into two basic passive island detection methods:Over/Under Voltage(OUV)and Over/Under Frequency(OUF).Active Frequency Shifting(Active Frequency)Drift,AFD),Sandia frequency shift(SFS),Sliding mode frequency shift(SMS)three traditional active island detection methods,and remote island detection methods.In this paper,the advantages and disadvantages of various methods of island detection are analyzed in detail,and passive and remote methods are selected for improvement.The improvement method fully exerts the passive method has no influence on the power quality,and the advantages of the remote method Non-Detection Zone(NDZ),avoids the deficiencies of the passive method and the remote method,and greatly improves the detection efficiency and accuracy.The improved method uses wavelet transform technology to collect the amplitude of PCC point voltage before and after the island and the harmonic of DG output current,and process the algorithm to obtain the feature vector,input the trained BP neural network for pattern recognition,and use the remote method to confirm the status..In this paper,the simulation method and simulation verification are carried out to prove the feasibility of the method.