摘要
针对相关滤波等经典频域分析方法提取动不平衡信号时,近频干扰抑制能力及参数估计精度严重依赖数据长度的问题,提出了一种基于残差MUSIC(multiple signal classification)谱分析的正弦参数估计方法,以残差MISIC谱中给定频率点的幅度值为观测变量判定参数拟合效果,提取该频率成分的幅值和相位。实验表明此方法与相关滤波法相比具有更高的频率分辨率,对抑制近频干扰的能力更出色,较好地解决了提高动不平衡信号提取精度与提高动平衡试验效率难于两全的问题。
The frequency resolution of traditional frequency domain analysis methods depends heavily on the length of data samples. So it is difficult for them to eliminate the influence of nearby frequency. Aiming at removing or suppressing the influence of nearby frequency interference to unbalance signal extraction, this paper presents a parameter estimation method based on MUSIC （multiple signal classification）, by which the magnitude and phase angle ofthe unbalance signal is obtained by searching the minimum of the MUSIC power spectrum of the residual signal at the given frequency. The experiment results show that MUSIC based method demonstrates a clear advantage over correlation filtering or LMS fitting methods in unbalance signal extraction especially when the sampled data series is of limited length and mixed with nearby frequency interferences.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2012年第6期1240-1246,共7页
Chinese Journal of Scientific Instrument
关键词
动平衡
MUSIC
正弦拟合
dynamic balancing
multiple signal classification （MUSIC）
sine estimation
