摘要
针对信号中的噪声影响基于振动信号的变压器早期故障诊断结果,提出一种基于集合经验模式分解自相关系数峰度阈值的降噪方法。该方法将含噪信号分解为多个固有模式分量,其中包括噪声分量和有用信号分量,然后根据两者自相关函数特性的不同,利用峰度值剔除噪声,拾取有用信息。仿真信号和变压器振动信号的降噪结果表明,该方法的降噪性能优于其他方法。
The fault detection performance of power is degraded by the noise. The novel de-noising method based on the ensemble empirical mode decomposition method（ EEMD） is presented. EEMD decomposes a noisy signal into several intrinsic mode functions（ IMF）,which include noise IMF and useful IMF. They are different in kurtosis ratio of autocorrelation function. According to this,an adaptive de-noising method is designed,and an improved kurtosis threshold method is used to pickup useful signal from noise IMF. Applying the method to de-noise simulation signal and the power vibration signal,the results show its efficiency and feasibility.
出处
《黑龙江电力》
CAS
2014年第3期189-193,共5页
Heilongjiang Electric Power
