基于谐波分量的配电电缆绝缘劣化状态带电检测技术

A Live Detection Technology of Distribution Network Cable Insulation Deterioration State Based on Harmonic Components

由于受城市电网停电检修的限制,配电电缆的谐波电流带电检测技术有望成为传统离线诊断方法的有效补充,以提升配电电缆绝缘状态的诊断实时性。该文搭建了10 kV配电电缆真型实验平台,并制备了受潮与长时热老化的典型缺陷电缆。通过COMSOL有限元电磁仿真获取了典型缺陷下电缆绝缘体的磁通演变规律,通过实验测试得到电缆典型缺陷时谐波电流特性及统计特征规律。在此基础上,利用谐波特征数据,构建了基于LASSO回归分析的配电电缆劣化程度分析方法,进一步提出基于聚类分析的缺陷类型辨识方法。结果表明,配电电缆谐波电流的3、4、5、11次谐波与电缆劣化状态密切相关。融合主成分分析(PCA)数据降维和期望最大化聚类分析的模型用于受潮与正常电缆状态辨识时,识别准确度最高可达75.64%。该文提出的带电检测手段及评估方法能有效排查具有潜伏性缺陷的高危电缆。

Due to restrictions imposed by power outages for maintenance on the urban power grid,the online detection technology of harmonic currents in distribution network cables is expected to become an effective supplement to traditional offline diagnostic methods,aiming to enhance the real-time diagnosis of the insulation status of distribution network cables.In this study,COMSOL finite element software was used to simulate the pores and water tree defects in the XLPE insulation layer of cables.The distribution of magnetic field strength in XLPE cables under different defect conditions was compared and analyzed.A real experimental platform for 10 kV distribution network cables was established,and typical defective cables with moisture and long-term thermal aging were prepared.The induced currents of aging defects and water tree defects in the XLPE cable insulation layer were collected in experiments,and the 2nd to 11th harmonic currents under different insulation defects were extracted.The influence patterns of pore depth caused by aging and external moisture intrusion on induced currents were obtained.To effectively assess the degree of cable degradation,a cable degradation assessment method based on the harmonic features of induced currents was constructed using LASSO regression.In the prediction analysis of normal thermally aged cables,the root mean square error of LASSO regression was 17.1 days,accounting for 14%of the actual aging time range,indicating high accuracy.The prediction of aging time for moisture-affected cables was longer,consistent with the actual state of accelerated insulation aging due to moisture.To accurately identify the type of insulation layer defects in cables,a defect identification method combining principal component analysis and clustering algorithm was developed based on the data sample set.When using harmonic data of a 300 A test current for clustering identification of moisture-affected and normal aging cables,the accuracy reached 75.64%,effectively distinguishing between moisture-affected cables and normal cables.The cable insulation live diagnosis technology proposed in this study,based on the harmonic current characteristics,integrates cable insulation degradation status,induced harmonic current features,and clustering analysis algorithms,achieving intelligent identification of cable insulation defects.