Wavelet analysis on vibration modal frequency measurement at a low level of strain of the turbine blade using FBG sensors

Vibration modal frequency measurement of a turbine blade was investigated by using fiber Bragg grating (FBG) sensors for their adaptation for a low level of strain at high frequency. However, the signal-to-noise ratio was so low that it was very difficult to identify dominant modal frequency from a raw signal acquired. An attempt was made to solve this problem. First, a bi-linear transform elliptic filter with pass-band and stop-band was proposed to remove electromagnetic interference noise. Second, discrete stationary wavelet transform with a soft threshold was utilized to de-noise the high-frequency part. Third, wavelet packet transform was exploited to decompose the time signal and to obtain the power spectrum and identification of dominant modal frequency. Experimental and analytic results demonstrated that four stages of dominant modal frequencies of the turbine blade without any constraint condition (free vibration) and three stages of dominant modal frequencies with a constraint condition (A-type vibration) were obtained, respectively. To testify to the validity of analytic results, a professional computer random signal and vibration analysis system (CRAS) was used to measure modal frequency. The results illustrated that modal frequencies obtained by the CRAS platform yielded close agreement with those based on FBG sensors. Obviously, wavelet analysis is successfully employed to decode vibration modal frequency from a raw signal of FBG sensors