【摘要】 为了提高10k单相接地故障判断准确率,提出基于混合特征向量-Fisher分类单相接地故障判断算法。首先,将10kV架空线路电流暂态特征和电流、电场稳态特征相结合,构建出3维混合特征向量ξ=(W_value,I,E)。其中暂态特征分量W_value用线路电流db5小波变换第3层小波模系数最大值表示,稳态特征分量I和E分别用线路接地故障发生前和发生后电流有效值和对地电场有效值表示。其次,利用混合特征向量样本建立Fisher分类器,采用Fisher分类器进行单相接地故障报警识别。使用F分布验证了分类器的有效性。最后,针对每条支路参数不同,又进一步研究了分类器参数更新方法。实验表明,该算法可以有效地对新向量进行分类。将仿真结果与其他算法进行了比较,证明了该算法提高了单相接地故障判断准确率,尤其提高了高阻接地识别效果。更多还原

【Abstract】 To improve the accuracy of 10 kV single-phase grounding fault recognition,the algorithm based on Hybrid Eigenvector-Fisher Classifier is proposed.Firstly,a three-dimensional hybrid eigenvector ξ=(W_value,I,E) is constructed by combining the current transient characteristics of 10 kV overhead lines with the current and electric field steady-state characteristics.The transient characteristic component W_value is expressed by the maximum value of the third-level wavelet modulus coefficient of db5 wavelet transform of line current,and the steady-state characteristic component I and E are respectively expressed by the effective value of current before and after the occurrence of line grounding fault and the effective value of earth-electric field.Secondly,a Fisher classifier is established for mixed eigenvector,and the classifier is used for single-phase grounding fault alarm recognition.The F distribution is used to verify the effectiveness of the classifier.Finally,according to the different parameters of each branch,the updating method of classifier parameters is further studied.The experimental results show that the hybrid eigenvector-Fisher classifier can effectively classify the new vectors,especially improve the recognition effect of high resistance grounding.更多还原