Nonlinear least squares algorithm with embedded Kalman filter for Bragg wavelength detection in fiber Bragg grating sensors

In a fiber Bragg grating (FBG) sensing system, the accuracy of Bragg wavelength detection is much affected by the presence of unwanted interferometric signal in the system. In this paper, a nonlinear least-square (NLS) algorithm with an embedded Kalman filter is used to remove this unwanted signal to enhance the measurement accuracy. This hybrid approach avoids the disadvantage of a pure NLS estimation, which is rather model sensitive, and the disadvantage of an extended Kalman filter, which might fail to converge. Computer simulations and experimental results are provided to demonstrate the effectiveness of this proposed method. Improvements of the accuracy in Bragg wavelength detection are observed.