变压器绕组变形故障诊断及脉冲频率响应研究

Transformer Winding Deformation Fault Diagnosis and Pulse Frequency Response Study

当前通常采用频率响应分析方法对变压器绕组变形故障实施检测,但是,单纯的频率信号抗干扰性能较差,存在诊断精度低及响应曲线特征失真的弊端。研究一种新的变压器绕组变形故障诊断及脉冲频率响应方法,采用短时Fourier变换算法处理变压器绕组变形故障的暂态信号,得到绕组脉冲频率响应图谱,以曲线图谱特征代替单纯信号特征,分析不同脉冲频率响应曲线的波形特征和差异。在利用支持向量机原理设计多个分类器,去除干扰,实现变压器绕组变形故障的准确诊断。实验结果说明,该方法可对变压器绕组变形脉冲频率响应进行准确、全面的分析,噪声较低,对绕组变形故障的诊断精确度高。

At present,frequency response analysis method is usually used to detect the transformer winding deformation fault. However,the simple frequency signal has poor anti-interference performance,low diagnostic accuracy and distortion of response curve features. This paper studies a new method of transformer winding deformation fault diagnosis and pulse frequency response. Using short-time Fourier transform algorithm to process the transient signal of transformer winding deformation fault,the frequency spectrum of the winding impulse response was obtained. The waveform features and differences of different pulse frequency response curves were analyzed by replacing the simple signal features with curve map features. Then,several classifiers were designed based on support vector machine（ SVM） principle to remove the interference and achieve accurate diagnosis of transformer winding deformation fault. The experimental results show that this method can analyze the transformer winding deformation pulse frequency response accurately and comprehensively,with low noise and high diagnostic accuracy for winding deformation fault.