| 引用本文: | 杨贤,周丹,王朋,林春耀,王丰华,马佳琪,盛戈皞.基于优化TQWT和LE的变压器绕组状态检测[J].电力自动化设备,2023,43(8):188-194 |
| YANG Xian,ZHOU Dan,WANG Peng,LIN Chunyao,WANG Fenghua,MA Jiaqi,SHENG Gehao.Detection of transformer winding condition based on optimized TQWT and LE[J].Electric Power Automation Equipment,2023,43(8):188-194 |
|
| 摘要: |
| 为实现短路冲击下变压器绕组状态的准确检测，提出了优化可调品质因子小波变换(TQWT)和拉普拉斯特征映射(LE)相结合的方法对变压器短路冲击下的振动信号进行分析。提出归一化奇异值熵作为TQWT分解过程中关键特征参数的选取准则，然后对TQWT分解后的振动信号子带能量序列进行LE，用以获取表征绕组状态的振动信号敏感特征。对某110 kV变压器短路冲击试验下振动信号的计算结果表明：优化TQWT算法可有效提高短路暂态振动信号分解的准确性，经LE获取的振动信号敏感特征可更加清晰地反映变压器绕组机械状态的劣化过程。当特征向量距离的变化超过3倍时，需要重点关注变压器绕组状态，从而为变压器绕组状态检修策略的制定提供依据。 |
| 关键词: 电力变压器 拉普拉斯特性映射 绕组状态 可调品质因子小波 振动信号 |
| DOI：10.16081/j.epae.202212006 |
| 分类号:TM407 |
| 基金项目:国家重点研发计划项目(2020YFB1709701) |
|
| Detection of transformer winding condition based on optimized TQWT and LE |
|
YANG Xian1, ZHOU Dan1, WANG Peng2, LIN Chunyao1, WANG Fenghua3, MA Jiaqi3, SHENG Gehao3
|
|
1.Electric Power Research Institute of Guangdong Electric Power Grid Co.,Ltd.,Guangzhou 510080, China;2.CSG Digital Grid Research Institute Co.,Ltd.,Guangzhou 510623, China;3.Key Laboratory of Control of Power Transmission and Conversion, Ministry of Education, Shanghai Jiao Tong University, Shanghai 200240, China
|
| Abstract: |
| To accurately detect the winding condition of power transformer under sudden short circuit impact, the optimized tunable Q-factor wavelet transform(TQWT) and Laplacian eigenmaps(LE) are proposed to analyze the vibration signals of transformer under sudden short circuit impacts. The normalized singular value entropy is applied to select the key parameters of quality factor and redundancy in the decomposition process of TQWT. With the sub-band energy sequence of vibration signals decomposed by TQWT, the most sensitive features of vibration signals are obtained by LE to represent the winding condition. The calculated results of vibration signals for an 110 kV transformer under sudden short circuit tests show that the optimized TQWT algorithm can effectively improve the decomposition accuracy of the transient vibration signals under short circuit. The sensitive features of vibration signals obtained by LE method is capable of clearly reflecting deformation process of mechanical condition of transformer winding. When the variation of feature vector distance(FVD) exceeds three times, it is necessary to concern the mechanical condition of transformer winding, so as to provide the reference for the proposal of transformer winding state maintenance strategy. |
| Key words: power transformer Laplacian eigenmaps winding condition tunable Q-factor wavelet transform vibration signals |
