Optimization of the phase generated carrier demodulation algorithm based on support vector regression.

Abstract

The phase generated carrier (PGC) algorithm is often used in the demodulation of interference signals in optical fiber sensors for its high sensitivity, good linearity, and large dynamic range. However, the PGC demodulation method is often distorted by the amplitude of the interference signal and the depth of modulation. In this study, the support vector regression (SVR) method is used to compensate the distortion of the PGC demodulation schemes. Simulation results showed that the SVR algorithm can effectively reduce the nonlinear error of the PGC demodulation system. The fitting accuracy of the SVR algorithm is 97.5% and greater than 90% in noiseless and noise systems, which is better than the back propagation (BP) neural network algorithm. Also, the SVR-based algorithm can better restore the amplitude with smaller mean square error and good correlation. A vibration monitoring system has been built, and experiment results confirm that the performance of the SVR-based algorithm is better than direct PGC demodulation and BP algorithm with the mean square error of 0.0005 and relevance of 0.94.