光伏系统直流微弱电弧信号检测研究

Research on DC Weak Arc Signal Detection in Photovoltaic System

光伏系统中电弧时有发生,一旦发生并引起火灾,会带来严重的危害。电弧强烈燃烧时,其时频域特征较为明显,通过常用的方差或小波分析等方法就可以实现检测。但是,当电弧燃烧比较微弱时,电弧电流波动很小,其特征频段与正常信号区分不大,这时无法实现电弧区分,将其称作为微弱电弧信号,其检测具有一定难度。针对这种微弱电弧信号,提出小波–奇异值分解新型信号消噪方法。该方法首先对光伏系统直流母线处采集到的电流信号进行小波变换,对分解后的各层小波系数再进行奇异值分解(singularvalue decomposition,SVD),通过与正常信号各奇异值的分析比较,去除随机噪声与开关频率,提取出微弱电弧信号特征,再对SVD后的小波系数重构得到降噪后的电弧电流,利用阈值法实现弱电弧检测。最后进行实验验证,实验结果表明该方法有良好的抗扰性。

The arc fault occurs occasionally in photovoltaic systems which may cause a fire disaster, bringing serious damage. The frequency-domain characteristics are obvious when the arc burns intensely, so simple methods, such as variance or wavelet can realize detection. However, while the arc burns weakly, the arc current fluctuation is very small, of which characteristic frequency is not distinguishable from normal signals. Therefore arc, called weak electric arc signal, cannot be distinguished, and the arc detection has a certain degree of difficulty. For the weak arc signals, this paper puted forward a new signal denoising method of wavelet-singular value decomposition(SVD), which firstly carried out wavelet transform on the current signal collected at the DC bus of photovoltaic system, and then proceeded SVD on the decomposed wavelet coefficients. By comparing with singular values of the normal signal, the random noise and switching frequency were removed, and the characteristics of weak arc signals were extracted. Then the wavelet coefficients after SVD were reconstructed to obtain the noise-reduced arc currents, and then the threshold method was used to detect weak arcs. Finally, experiments have verified that this method has good noise immunity.