Research On The Characteristics And Recognition Technology Of DC Series Fault Arc In Photovoltaic System

The rapid development of photovoltaic power generation,not only brought clean green energy,and brought a lot of security issues,caused widespread concern at home and abroad.According to media reports,most of photovoltaic fire accidents are caused by the DC side of the fault arc.Based on this phenomenon,the 2011 version of the National Electrical Code(NEC)proposed requirement of installing DC fault arc protection devices in photovoltaic system.As most of the photovoltaic system power supply is made up of a large number of photovoltaic panels in series,the probability of series fault arc is higher,for this reason the DC series fault arc of photovoltaic system is taken as the research focus in this paper.At present,the detection technology of photovoltaic DC fault arc is in the exploratory stage both at home and abroad,and arc burning is affected by various factors with randomness and complexity.Therefore the mechanism of arc generation is described in this paper,the Cassie arc model is selected to simulate in order to guide the experiment.And then an arc generator was made by ourselves,a simulated arc experiment platform for simulating PV system was established,the characteristics of the fault arc were further studied and analyzed.In order to avoid the misjudgment caused by single criterion of the characteristic analysis of the fault arc,in this paper,the problem of series arc fault in photovoltaic system is studied by combining time domain analysis and frequency domain analysis.The differences between the peak-to-peak current and the current average in normal operation and series fault arc are acquired,the feasibility of the current time domain feature in the series fault arc identification is analyzed.Get the conclusions that the standard deviation of the low-frequency coefficients of the three-layer wavelet decomposition and the standard deviation of the high-frequency coefficients are all higher than the normal state.Finally,the neural network is used to verify the feasibility and accuracy of the arc characteristic.