|
| 摘要: |
| 现有的基于信号奇异性分析的故障检测方法不同程度地受到噪声的影响。提出了将电力信号滤除工频周期分量后,通过提取适当尺度上的小波模极大值点检测故障的方法。通过预处理滤除工频周期分量,消除了小波变换在信号峰值处的模极大值,从而避免了对故障的误判。小波变换将预处理信号中的故障分量和噪声分解在不同的尺度空间中,保证了故障特征的提取和算法的抗噪性能,简化了的Mallat信号奇异性检测方法,在降低算法计算量的同时,可保持故障的定位精度。仿真研究表明:故障定位准确,且对噪声不敏感,可推广应用到其他周期信号的分析中。 |
| 关键词: 小波分解,故障检测,模极大值,周期信号 |
| DOI： |
| 分类号:TM711 |
| 基金项目: |
|
| Wavelet fault detection based on filtering pretreatment |
|
LIU Yi-hua ZHAO Guang-zhou
|
| Abstract: |
| Noise degrades most existing fault detection methods based on singularity analysis.A method based on wavelet module maxima of the pretreated electric signal is proposed.As wavelet module maxima occur at peaks of signal,the pretreatment removes power frequency periodic components from electric signal and thus avoids misjudgments of fault.Wavelet transform decomposes fault component and noise into different scale spaces,which guarantees the abstraction of fault signature and the de-noising performance.The simplified Mallat singularity detection method not only reduces the computational complexity,but also keeps the localization accuracy.Simulations show that the fault detection method is precise for fault locating and insensible to noise.The method can also be applied to other periodic signal analysis. |
| Key words: wavelet decomposition,fault detection,module maxima,periodic signal |
