基于模糊诊断的变电站开关柜局部放电检测方法

Partial Discharge Detection Method of Substation Switchgear Based on Fuzzy Diagnosis

在开展变电站开关柜局部放电检测的过程中,由于原始的局部放电信号中存在干扰噪声,导致检测结果较大,为此提出基于模糊诊断的变电站开关柜局部放电检测方法研究。为了有效抑制局部放电的背景噪声,以经验模态分解(Empirical Mode Decomposition,EMD)为基础,结合开关柜局部放电信号的特征,对其进行适应性改进,在原始的信号中加入了固定的白噪声信号后,进行自适应集成经验模态分解(Adaptive Ensemble Empirical Mode Decomposition,AEEMD)处理,再将分解后的白噪声信号加入到开关柜局部放电信号AEEMD分解的一阶内涵模态分量(Intrinsic Mode Functions,IMF)中,循环至重构开关柜局部放电信号IMF中,噪声信号的能量密度不再发生变化。在局部放电(Partial Discharge,PD)检测时,设置了开关柜局部放电信号指标隶属函数模糊边界值,并具体计算了超出模糊边界范围的局部放电强度。由测试结果表明,对于不同类型、不同干扰条件下的局部放电强度检测结果始终保持较高的稳定性,误差水平也相对较低,其中最大误差为0.25 V,最小误差为0.07 V。

In the process of partial discharge detection of substation switchgear,due to the interference noise in the original partial discharge signal,the detection result is larger.Therefore,a detection method of partial discharge of substation switchgear based on fuzzy diagnosis is proposed.In order to effectively suppress the background noise of partial discharge,based on Empirical Mode Decomposition(EMD)and combined with the characteristics of the partial discharge signal of the switch cabinet,the adaptive improvement is made.After adding a fixed white noise signal to the original signal,Adaptive Ensemble Empirical Mode Decomposition(AEEMD)decomposition processing is carried out.Then the decomposed white noise signal is added to the first-order Intrinsic Mode Functions(IMF)component of the partial discharge signal AEEMD decomposition of the switch cabinet,and the energy density of the noise signal in the IMF component of the reconstructed partial discharge signal of the switch cabinet does not change.In Partial Discharge(PD)detection,fuzzy boundary value of PD signal index membership function is set,and the partial discharge intensity beyond the fuzzy boundary range is calculated.In the test results,the test results of the design method always maintain high stability for the detection results of partial discharge intensity under different types and different interference conditions,and the error level is relatively low,where the maximum error is only 0.25 V and the minimum error is only 0.07 V.