基于变分模态分解-支持向量机混合算法的TN系统大阻抗接地故障配电线路保护

Protection of Distribution Lines with Large Impedance Earth Fault in TN System Based on Variational Mode Decomposition-Support Vector Machine

低压配电TN系统发生接地故障的配电线路阻抗较大时,采用以剩余电流幅值大小作为动作依据的剩余电流动作保护电器防护措施,不能有效解决接地故障防护问题。从分析剩余电流波形的角度出发,采用一种基于变分模态分解(variational mode decomposition, VMD)及支持向量机(support vector machine, SVM)分类的故障诊断方法。该方法在低压配电系统内测得剩余电流信号波形,对该波形进行VMD分解后得到各剩余电流信号的固有模态函数(intrinsic mode function, IMF),对其进行Hilbert变换并进行积分得到Hilbert边际谱,再求该边际谱的能量熵,将其作为特征向量输入SVM进行分析,最后准确区分正常状态和故障状态。结果表明:发生配电线路大阻抗接地故障时,该方法可以大幅提高用剩余电流动作保护电器(residual current protection device, RCD)作为故障防护的准确率。

When the impedance of the distribution line with earth fault in the low-voltage distribution TN system is large,the residual current action protection measures based on the magnitude of residual current cannot effectively solve the problem of earth fault protection.From the perspective of residual current waveform analysis,a fault diagnosis method based on variational mode decomposition(VMD)and support vector machine(SVM)classification was adopted.The residual current signal waveform was measured in the lowvoltage distribution system.After the VMD decomposition of the waveform,the intrinsic mode function(IMF)of each residual current signal was obtained.Hilbert transformation and integration were performed to obtain Hilbert marginal spectrum.Then,the energy entropy of the marginal spectrum was obtained and inputted to SVM as a feature vector for analysis.Finally,the normal and fault states were accurately distinguished.The results show that this method can greatly improve the accuracy of using residual current protection device(RCD)for fault protection in case of large impedance earth fault of distribution lines.