基于机器学习的光纤网络激光器异常功率数据采集系统设计

Design of optical fiber network laser abnormal power data acquisition system based on machine learning

针对传统的激光器异常功率数据采集系统存在采集精度较低、抗干扰性不好等问题,提出并设计了一种基于机器学习的光纤网络激光器异常功率数据采集系统。给出激光器异常功率数据采集系统的总体架构,从硬件和软件2个方向分析,得到激光器异常功率数据采集系统的硬件框图;软件部分,根据激光器异常功率检测,判断功率数据是否异常,结合瞬时谱检测方法,对检测得到的异常功率数据进行时频分布特征分析。同时根据激光器异常功率数据特征的主成分,获取激光器异常功率数据的特征值,在此基础上,采用机器学习方法采集异常功率数据,实现了光纤网络的激光器异常功率数据采集系统设计。仿真结果表明,采用该方法进行激光器异常功率数据采集的精度在86%~95%区间,采集精度较高,输出数据的抗干扰性较好,输出信噪比均高于45 dB。

Aiming at the problems of the traditional laser abnormal power data acquisition system such as low acquisition precision and poor anti-interference,a new optical fiber network laser abnormal power data acquisition system based on machine learning is proposed.The overall structure of the laser abnormal power data acquisition system is given.The hardware block diagram of the laser abnormal power data acquisition system is obtained by analyzing the hardware and software.In the software part,according to the abnormal power detection of laser,the abnormal power data is judged,and the time-frequency distribution characteristics of abnormal power data are analyzed by using instantaneous spectrum detection method.At the same time,according to the characteristics of abnormal power data of laser,the machine learning method is used to collect the feature values of laser abnormal power data.On this basis,the design of laser abnormal power data acquisition system based on fiber network is realized through machine learning.The simulation results show that the accuracy of laser abnormal power data collection by this method is in the range of 86%~95%,which is high,the output data has good anti-interference,and the output signal-to-noise ratio is higher than 45 dB.