基于机器学习的光纤多参量探测

Optical Fiber Multi-Parameter Measurement Based on Machine Learning

提出一种借助机器学习算法从信号非完整信息提取待测参量的方法,该方法以只包含信号部分信息的功率谱幅度数据取代包含脉冲幅度和相位全部信息的数据来完成参量提取,克服了复杂光信号相位信息测量困难的问题。通过模拟仿真,验证了使用机器学习算法实现从脉冲演化提取传输介质参量信息的能力以及利用缺失相位信息的脉冲功率谱实现光纤多参量探测的可行性。仿真结果表明,采用适当的机器学习算法,所提方法的均方误差可控制在0.3%以内。

This paper proposes a method to extract the parameters to be measured from the incomplete information of the signal by machine learning.Instead of the data containing all the pulse amplitude and phase information,the method employs the power spectrum amplitude data containing only part of the signal information for parameter extraction.It overcomes the difficulty in measuring the phase information of complex optical signals.Simulations verify the ability to utilize machine learning algorithms to extract the parameter information of transmission medium from pulse evolution and the feasibility of using the power spectrum of pulse without phase information to realize optical fiber multi-parameter measurement.The simulation results show that the mean square error of this method can be controlled below 0.3%with proper machine learning algorithms.