在机械设备盲信号处理中，机械振动的源数估计是一个重要的课题。针对机械设备多通道观测信号数目小于源信号数目的欠定问题，试将小波理论引入盲信号处理，并结合奇异值分解，提出了基于小波和小波包分析的机械振动源数估计算法；首先将机械设备观测信号进行小波分解和重构，得到小波系数重构信号，其次重组新的多维观测信号并进行奇异值分解，最后根据奇异值估计机械设备的源信号数目。仿真试验表明：在欠定情况下，机械振动源数估计的小波方法能够正确地估计机械设备的源信号数目。%Source number estimation of mechanical vibration is a key research in blind signal processing of mechan- ical equipments. Usually, multi-channel observation signals are less than source signals in fault diagnosis of mechanical equipments. Under this underdetermined condition, wavelet decomposition algorithm for source number estimation was proposed. The present algorithm integrated blind signal processing with wavelet analysis and singular value decomposition, in which the analysis steps included that the first step 1 was to decompose and reconstruct observation signals of mechanical equipments with wavelet analysis and obtain wavelet reconstruction signals, the second step. To recombine new multi-channel observation signals step 3 was to estimate source signals number of mechanical mental resultss showed that wavelet decomposition algorithm number under underdetermined conditions. and obtain singular value decomposition and the final equipments according to singular values. The experi- could correctly estimate independent mechanical source
展开▼
