基于改进ECOC分类器的直流电缆终端接头局放模式识别

发布时间：2018-04-30 00:47

本文选题：直流电缆 + 局放信号图　；参考：《中国电机工程学报》2017年04期

【摘要】：绝缘缺陷问题直接影响直流XLPE电缆的运行安全,而准确的绝缘状态诊断对保证直流输电系统正常运行具有重要意义,由于直流电缆的故障诊断目前研究处于起步阶段,且局部放电特征与交流XLPE电缆具有明显区别。针对直流XLPE电缆出现的绝缘缺陷以及产生的局部放电特点,该文设计了4种直流XLPE电缆终端接头典型的绝缘缺陷物理模型,根据q-(35)t-n(即放电幅值,放电间隔,放电次数)局放信号图,提出了基于改进ECOC分类器的直流电缆终端局放模式识别法。首先,对局放信号图进行轮廓波(Contourlet)变换,并计算各子带系数的Tsallis熵,将其作为特征向量,带入自适应布谷鸟优化稀疏编码阵的ECOC分类器(ACS-SR-ECOC)实现缺陷模式识别。通过对大量试验数据测试,验证了所提出的识别方法在直流XLPE电缆终端接头绝缘缺陷的诊断效果,相比与4种传统的ECOC分类器,所提出的ACS-SR-EOCO分类器的识别准确率更高,至少提高10.3%。
[Abstract]:The problem of insulation defect directly affects the safety of DC XLPE cable operation, and accurate insulation condition diagnosis is of great significance to ensure the normal operation of HVDC transmission system, because the fault diagnosis of DC cable is still in its infancy. The characteristic of partial discharge is obviously different from that of AC XLPE cable. Aiming at the insulation defects and partial discharge characteristics of DC XLPE cables, four typical physical models of insulation defects in terminal joints of DC XLPE cables are designed in this paper. Based on the improved ECOC classifier, a DC cable terminal PD pattern recognition method is proposed. Firstly, the contour wave Contourlet (Contourlet) transform is applied to the partial discharge signal map, and the Tsallis entropy of the coefficients of each sub-band is calculated, which is used as the eigenvector and brought into the ECOC classifier ACS-SR-ECOC, which is an adaptive cuckoo sparse optimized coding matrix, to realize the defect pattern recognition. By testing a large number of experimental data, the results show that the proposed method can diagnose insulation defects of terminal joints of DC XLPE cables. Compared with four traditional ECOC classifiers, the proposed ACS-SR-EOCO classifier has higher recognition accuracy. Increase at least 10. 3.
【作者单位】：上海交通大学电气工程系;
【基金】：国家重点基础研究发展计划项目(973项目)(2014CB239506) 国家电网公司科技项目资助(52110115007J)~~
【分类号】：TM855

【相似文献】