An EMD-wavelet algorithm for PDH error signal noise reduction of FBG strain sensor

The paper proposes an empirical mode decomposition (EMD)-wavelet algorithm for the noise suppression of the PDH error signal. The EMD is used to decompose the PDH error signal into a finite number of intrinsic mode functions. The Daubechies 4 wavelet is used to suppress the noise in the intrinsic mode functions. A FBG based Fabry-Perot interferometer is used as the sensing element. A narrow linewidth tunable laser is used as the laser source. The laser is phase modulated by a sinusoidal signal. A lock-in amplifier is used to extract the PDH error signal. A proportional-integral-derivative controller is used to control the laser frequency. The noise in the PDH error signal is effectively suppressed. Moreover, the signal-noise-ratio (SNR) of the PDH error signal is improved from 16 dB to 28 dB. A strain signal (1 nε) is applied on the FBG-FP via a PZT. A high-resolution of 680 fε/√Hz@ 1 kHz is achieved.