FBG反射谱中心波长检测算法仿真与实验分析  

Simulation and experiment analysis of central wavelength detection algorithm for FBG reflection spectrum

作　　者：尹成群[1] 王梓蒴[1] 何玉钧[1] 李永倩[1] 

机构地区：[1]华北电力大学电子与通信工程系,河北保定071003

出　　处：《红外与激光工程》2011年第2期322-327,共6页Infrared and Laser Engineering

基　　金：国家863计划资助项目(2006AA092141);河北省高等学校科学技术研究指导项目(Z2010292)

摘　　要：为了进一步提高光纤布拉格光栅(FBG)中心波长的检测精度,结合其反射谱波长检测方法,给出了一种反射谱数据处理流程并对流程中的数据拟合方法进行了重点分析。通过仿真和实验,对高斯曲线拟合、一般多项式拟合、三次样条拟合和径向基函数(RBF)神经网络拟合的检测精度进行了比较,并利用高斯拟合分析了不同数据采样间隔和不同的数据截取宽度对检测精度的影响。仿真及实验结果表明,相同条件下,RBF神经网络拟合的检测精度最高,但数据量大时不适合实时运算;高斯曲线拟合可兼顾实时性与高精度的双重要求;数据采样间隔超过5 pm时误差明显增大;截取的数据宽度在3dB带宽内时可明显降低检测误差。To increase the central wavelength detection accuracy of Fiber Bragg Grating （FBG）, combining the reflection spectrum wavelength detection method, a data process flow for FBG reflection spectrum was presented and the data fitting method in the flow was analyzed. By the simulation and experiment, the detection accuracies of fitting methods, which included the Gaussian curve fitting, the General polynomial fitting, the Cubic spline fitting and the RBF neural network fitting, were compared, and the effects of different sampling intervals and data interception width on detection accuracies were analyzed with Gaussian curve fitting. According to the simulation and experiment results, RBF neural network fitting possesses the highest detection accuracy, but it is not suitable for realHartmann-Shacktime data processing with large quantity of data, Gaussian curve fitting can take account of both high accuracy and real Hartmann-Shacktime. The detection error increases clearly in the case of the data sampling interval is more than 5 pm and the data interception bandwidth is more than 3 dB.

关 键 词：波长检测 FBG 曲线拟合 反射谱 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]