Study of Power Spectrum Replication Algorithm of Random Vibration
This paper discusses both the FRF (frequency response function) estimation and the generation of drive signal in time domain in classical PSD (power spectral density) replication algorithm of random vibration. First, the FRF estimator based on the EV (Errors-in-Variables) model is presented to reduce the bias error of FRF H2 estimator. The H2 estimated value is corrected by the noise PSD in the output of the system and the control signal PSD. Second, the Parks-McClellan method is used to design the FIR (finite impulse response) filter. And then the drive signal in time domain is generated by filtering a series of independent white noise with the designed filter. The method avoids the time domain randomization process in the frequency-to-time transformation in the classical PSD replication algorithm, so that the real-time property of the control system is improved. The improved algorithm is applied to the PSD replication test on the hydraulic vibration table. The result shows that it is favorable to improve the control precision of the PSD replication.
PSD replication the EV model FRF estimation FIR filter
Guan Guanfeng Cong Dacheng Han Junwei Li Hongren
School of Mechanical Electronic Engineering, Harbin Institute of Technology, Harbin, 150001, China
国际会议
北戴河
英文
494-497
2007-06-06（万方平台首次上网日期，不代表论文的发表时间）