基于机器学习的光纤多参量传感器准确解调方法研究  

Accurate Demodulation Fiber Optic Multi-parameter Sensor Based on Machine Learning

作　　者：徐欣瑜 赖雁翔 肖悦娱[1,2] 

机构地区：[1]上海大学特种光纤与光接入网省部共建重点实验室,上海200444 [2]上海大学光纤研究所,上海201800

出　　处：《工业控制计算机》2025年第2期28-29,32,共3页Industrial Control Computer

摘　　要：基于干涉光谱的光纤多参量传感器在高精度测量领域有着重要作用,传统的矩阵解调法测量的前提是线性拟合特征波长,但由于交叉灵敏度的干扰以及噪声干扰,其解调准确性较低。为了实现光纤多参量传感器干涉光谱的准确解调,提出了一种基于机器学习的准确解调方法,将整个光谱与传感参量直接映射,可有效提高光纤多参量传感器的解调准确性。实验结果表明可实现解调后的温度决定系数为0.9997、乙醇浓度决定系数为0.9995的准确解调,因此基于机器学习的多参量解调方法可有效提高光纤多参量传感器的解调准确率,为其在多参量测量领域的发展提供了一种新的思路。In order to achieve accurate demodulation of the interference spectrum of fiber optic multiparametric sensors,this paper proposes an accurate demodulation method based on machine learning,which directly maps the whole spectrum to the sensing parameters,and can effectively improve the demodulation accuracy of fiber optic multiparametric sensors.The experimental results show that the accurate demodulation can be achieved with the temperature determination coefficient of 0.9997 after demodulation,the ethanol concentration determination coefficient of 0.9995.Therefore,the machine learning-based multi-parameter demodulation method can effectively improve the demodulation accuracy of fiber optic multi-parameter sensors,and provide a new idea for its development in the field of multi-parameter measurement.

关 键 词：光纤传感器 光谱解调 机器学习 

分 类 号：TN9[电子电信—信息与通信工程]