Multi-peak FBG reflection spectrum segmentation based on continuous wavelet transformation

Abstract A method to segment multi-peak FBG reflection spectrum based on continuous wavelet transformation (CWT) is proposed and experimentally demonstrated. First order Gaussian or Mexican hat wavelet basis with the suitable scale is selected in CWT processing. The pairs of extremum points in the processed signal are used to segment original signal. Multi-peak FBG spectrum is simulated for the two cases, with no-overlap peaks and noise, with partial overlapped peaks and noise. An experiment is conducted on four FBGs connected together in series. The simulating and experimental results show it has advantages to segment the original signal by CWT in adjusting suitable segmented length and obtaining more accurate central wavelength, compared to traditional methods.