利用窄带噪声辅助多元经验模态分解算法检测换流变压器用有载分接开关机械状态

发布时间：2018-03-06 16:05

本文选题：换流变压器　切入点：有载分接开关　出处：《电工技术学报》2017年10期 　论文类型：期刊论文

【摘要】：为有效检测换流变压器用有载分接开关的机械状态,提出一种窄带噪声辅助多元经验模态分解的方法对有载分接开关切换过程中采集到的多通道振动信号进行分析。具体应用时,在原始多通道振动信号中增加若干通道窄带噪声信号,并在统一高维超球面坐标系下进行分解计算,从而在有效抑制经验模态分解过程中模态混叠现象的同时,极大提高了多维振动信号分解的准确性并降低了运算复杂性。进而根据区间最大功率特征计算固有模态函数的功率矩阵,对有载分接开关的机械特征进行描述。对有载分接开关样机正常与典型机械故障时振动信号的计算结果表明,其不同工况下的功率特征有较大区别,所定义的矩阵相似度指标可较好地衡量有载分接开关典型故障时的振动差异程度。此外,触头松动时与正常工况的相似度指标低于绝缘板松动故障,说明触头松动故障对有载分接开关切换过程的影响更为明显。
[Abstract]:For effective detection of converter transformer with load mechanical state switch, a narrowband noise assisted method of multivariate empirical mode decomposition of multi channel vibration signal on load tap in the switching process collected were analyzed. The specific application, the original multi-channel vibration to increase the number of channel narrowband noise signal in the calculation and decomposition in a unified high Weichao spherical coordinates, and aliasing in effectively inhibit the process of empirical mode decomposition mode at the same time, greatly improves the accuracy of the multi-dimensional vibration signal decomposition and reduces computation complexity. Then calculate the power matrix of intrinsic mode function according to the maximum power characteristic interval, on load mechanical characteristics of switch are described. The OLTC prototype of normal and typical mechanical fault vibration signals are calculated. The results show that the different conditions The power characteristics of different matrix similarity index defined can be used to measure the OLTC vibration level difference of typical faults. In addition, contact loose under the normal condition the similarity index is lower than the insulation board looseness, contact that looseness fault effect on load tap switching process is more obvious.

【作者单位】：电力传输与功率变换控制教育部重点实验室(上海交通大学);
【基金】：国家自然科学基金资助项目(51207090)
【分类号】：TM403.4

【相似文献】