Research On DC Series Arc Fault Detection Technology Of Photovoltaic System

Many photovoltaic power stations have been put into operation in Hubei,Jiangsu,Anhui and other places in our country.With the increase in the service life of the equipment in the power station,many failures appear.Among them,DC series arc faults have become one of the most difficult problems due to some factors such as manual inspections that are difficult to find and no zero-crossing points.If it is not found in time,the continuous arc burning will reach the ignition point of the weeds around the power station and cause a fire.Therefore,photovoltaic power plants urgently need a set of DC series arc detection algorithms with high judgment accuracy.This paper studies the arc detection algorithm based on BP neural network,using the average value,maximum value,maximum rate of change of the signal after current filtering,wavelet coefficient variance and wavelet coefficient modulus maximum value to characterize the signal state.Experimental results show that the actual output of this method has a large deviation from the ideal output,which may not be suitable for high-noise detection environments.In order to maintain high detection accuracy in a high-noise environment,this paper proposes a novel detection method based on convolutional neural network.First,the time-frequency spectrum of the signal is constructed,and then the value corresponding to each time-frequency point in the time-frequency spectrum is used as the input of the convolutional neural network,and the convolutional neural network algorithm is designed to realize the arc fault detection.Both Short-time Fourier Transform(STFT)and Discrete Wavelet Transform(DWT)can generate time-spectrograms.STFT expresses the value of each time-frequency point in terms of energy density,and DWT expresses the value of wavelet coefficient modulus.Experimental verification shows that both methods can clearly distinguish the characteristics of arc fault current and normal operating current.In the laboratory test,the DC series arc fault of the photovoltaic system can be accurately detected.Taking into account the number of iterations of the training group and the convergence speed of the output results,the DWT+CNN method is better.The last part of this research focuses on system-level detection and protection.When the signal is mixed with switching noise,the time-frequency spectrum can still reflect the characteristics of the arc fault.By analyzing the impact of DC series arc faults on the DC side of the photovoltaic system,the coordination method between arc protection of each line and the distinguishing method from other types of photovoltaic array faults are proposed.