The optimization study of FBG Gaussian fitting peak-detection based on Levenberg-Marquardt algorithm

In order to find a more accurate and stable peak-detection algorithm of FBG, a method of solving the peak position by Gaussian Fitting Algorithm based on Levenberg-Marquardt Algorithm (LM) was studied. In this paper, the principles of Gaussian fitting and Levenberg-Marquardt algorithm are described, and their calculation steps in peak searching algorithm are given. The MOI's SM125 sensor demodulator is used to get full-spectrum data under different temperatures. We analyzed the influence of optical power threshold on peak finding accuracy, and got the optimal threshold. The full spectrum data intercepted by the optimal threshold is used to search peaks, and we compared result with the wavelength got by SM125. The fitting error line graph at different temperatures is drawn, and the algorithm's average fitting error is 1.33pm. The results showed that Gaussian-LM algorithm has the advantages of high accuracy, good stability, which is important to improve the demodulation accuracy.