基于时频相似度的油纸绝缘多局部放电源脉冲群分离与识别策略

发布时间：2018-01-11 22:12

本文关键词：基于时频相似度的油纸绝缘多局部放电源脉冲群分离与识别策略　出处：《电工技术学报》2014年12期 　论文类型：期刊论文

【摘要】：由于变压器内部液-固绝缘结构和运行环境的复杂性,导致其内部往往是多个局部放电源同时存在,引起局部放电PRPD图谱交叉和重叠而无法识别。为解决多局部放电源的识别问题,本文提出了基于脉冲时频分析(TFA)结合近邻传播聚类(APC)的变压器油纸绝缘多局部放电源脉冲群分离与识别策略。首先,将基于S变换(ST)时频分布的脉冲相似度(时频相似度)矩阵输入近邻传播聚类(APC)分离多局部放电源脉冲群。然后,采用局部放电相位分布(PRPD)统计特征与粒子群优化支持向量机(PSO-SVM)识别ST+APC分离的子脉冲群,并验证分离的有效性。对实验室人工缺陷模型的局部放电数据进行分析,结果表明ST+APC算法可以有效去除脉冲型干扰(PSN)和分离油纸绝缘多局部放电源脉冲群。
[Abstract]:Because of the complexity of the internal liquid-solid insulation structure and the operating environment of the transformer, it often exists multiple local discharge power sources at the same time. In order to solve the problem of identification of multi-PD power supply, the partial discharge (PD) PRPD spectra are crossed and overlapped and can not be identified. In this paper, a strategy of pulse group separation and identification for transformer oil-paper insulation multi-local discharge power supply based on pulse time-frequency analysis (TFAA) and nearest neighbor propagation clustering (APC) is proposed. The pulse similarity (time-frequency similarity) matrix based on S-transform ST-time-frequency distribution is input into the nearest neighbor propagation cluster (APCs) to separate the multi-local discharge pulse group. Partial discharge phase distribution (PD) and particle swarm optimization support vector machine (PSO-SVM) are used to identify the sub-pulse groups separated by St APC. The validity of the separation is verified. The partial discharge data of the lab artificial defect model are analyzed. The results show that the St APC algorithm can effectively remove the pulse-type interference (PSN) and separate the pulse group of oil-paper insulation multi-local discharge power supply.
【作者单位】：中国电力科学研究院;重庆大学输配电装备及系统安全与新技术国家重点实验室;国网技术学院;
【基金】：国家自然科学基金资助项目(51277187)
【分类号】：TM855
【正文快照】： 1引言变压器是电网中的核心设备,其安全运行直接决定了电能供应的可靠性[1]。在设备制造、运输以及长期运行过程中会不可避免地产生绝缘缺陷并引发的局部放电(Partial Discharge,PD),严重影响变压器的可靠运行。因此,局部放电源的分析与诊断一直是变压器状态评估的重要内容,能

【参考文献】