基于连续小波变换的变压器绕组变形故障类型检测

发布时间：2018-07-25 20:53

【摘要】：利用脉冲频率响应法在线检测电力变压器绕组变形故障时,传统的频率响应曲线并不能刻画暂态信号的时域特性,且快速Fourier变换的局限性可能造成有用信息的缺失。为了克服以上缺陷,提出利用连续小波变换处理脉冲在线注入获得的暂态信号,绘制检测信号小波时频图,以矩阵相似度作为故障评判的量化指标。首先从理论推导入手,构建了单绕组仿真模型,开展了各种故障类型仿真分析;其次,进行了故障绕组试验,分析了信号时频特性,另外,采用矩阵相似度量化评判健康与故障绕组检测信号的关联程度。结果表明:仿真不同故障类型对应的小波时频图与健康绕组的时频图差异明显,且时频图在各频段差异表现出和频率响应相似的规律;与健康绕组相比,试验故障绕组的小波时频图在0.6 MHz以上高频段出现较大偏差,表现出绕组电容性故障应有的特性,而在0~0.6 MHz频段,试验绕组小波时频图的矩阵相似度为0.928 0,大于采用快速Fourier变换获得频响曲线的关联系数0.800 3,证实了小波变换的优越性。仿真分析与实验测试的数据处理结果,均初步证实该方法的可行性。
[Abstract]:The traditional frequency response curve can not describe the time-domain characteristics of transient signals when the power transformer winding deformation fault is detected on-line by the pulse frequency response method, and the limitation of fast Fourier transform may lead to the lack of useful information. In order to overcome the above shortcomings, a continuous wavelet transform (CWT) is proposed to process the transient signals obtained by pulse on-line injection, and to draw the wavelet time-frequency map of the detected signals. The matrix similarity is used as the quantitative index for fault evaluation. The simulation model of single winding is constructed, and the simulation analysis of various fault types is carried out. Secondly, the fault winding test is carried out, and the time-frequency characteristic of the signal is analyzed. The correlation between health and fault winding detection signal is evaluated by matrix similarity quantification. The results show that the time-frequency map of wavelet corresponding to different fault types is obviously different from that of healthy winding, and the difference of time-frequency diagram is similar to that of frequency response in each frequency band, and compared with that of healthy winding, The wavelet time-frequency diagram of the test fault winding shows a large deviation in the high frequency band above 0.6 MHz, which shows the characteristic of the winding capacitive fault, but in the 0.6 MHz frequency band. The matrix similarity of the wavelet time-frequency graph of the test winding is 0.928, which is larger than the correlation coefficient of the frequency response curve obtained by the fast Fourier transform, which proves the superiority of the wavelet transform. The results of simulation analysis and experimental test show that the method is feasible.
【作者单位】：云南电网有限责任公司电力科学研究院;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;
【基金】：国家自然科学基金(51377175) 云南电力公司科技项目(K-YN2014-031)~~
【分类号】：TM407

【参考文献】