光纤光栅传感信号寻峰误差研究

Research on peak searching algorithms errors of fiber bragg grating sensing signal

光纤光栅传感器是通过检测被测物理量引起的光纤光栅中心波长的漂移量来得到该物理量的变化,波长检测精度常受到寻峰算法的影响。本文研究了5种寻峰算法,分析了加窗截断长度、信噪比及波长分辨率对寻峰算法误差的影响。截断长度太长或太短都会造成寻峰误差骤增,每种寻峰算法对应的加窗截断长度稍有不同,略大于3dB带宽的2倍为最佳;信噪比越高,寻峰误差越小;波长分辨率越高,寻峰误差越小。通过与实验对比,相同条件下,高斯公式非线性拟合在5种算法中寻峰误差最小,质心算法寻峰误差最大。实验中高斯公式非线性拟合算法的寻峰误差可以达到1.00pm。

Fiber Bragg grating sensor makes acquisition to physical change by detecting the FBG central wavelength drift that caused by physical change,while the wavelength detection precision is limited by peak searching algorithm.Five peak searching algorithms were studied.The influence of truncating length,signal to noise ratio（SNR）and wavelength resolution on peak searching algorithms were analyzed.It is found that the peak searching error is increasing rapidly if the truncating length is either too long or too short.The truncating length of each peak searching algorithm is generally different,while slightly larger than 2times of the bandwidth of 3dB is a better choice.We clarify a fact that the higher the SNR and wavelength resolution,the smaller the peak searching error.Combined with the experiment,the peak searching error of Gaussian nonlinear fitting is the smallest,while the centroid algorithm perform worst.The peak searching error of Gaussian nonlinear fitting can reach 0.01 pm in experiment.