数据中心光纤通信网络传输非平稳数据无损检测技术研究
|
摘要
针对当前方法存在检测准确率低、稳定性差、耗时长等问题,提出一种基于BP神经网络分类器的非平稳数据无损检测方法,构建数据中心光纤通信网络传输数据的噪声模型,采用平均斜率法和最小二乘法相结合的方式,根据数据中心光纤通信网络传输数据中包含的线性趋势项和非线性趋势项进行消除,完成数据中心光纤通信网络传输数据预处理;将得到的数据中心光纤通信网络传输数据预处理结果作为样本训练BP神经网络,采用梯度下降法作为BP神经网络学习规则以及成批训练模式,对BP神经网络各层权值进行初始化设置,并将权值向训练误差减小方向进行调整,同时进行权值更新,完成对BP神经网络分类器训练,将经过预处理得到的数据中心光纤通信网络传输数据样本全部输入到训练好的BP神经网络分类器中,进行非平稳数据无损检测。实验结果证明,相比其他方法,本文方法能够实现光纤通信网络传输非平稳数据高效、准确、稳定的无损检测。
Aiming at the shortcomings of low accuracy,poor stability and long time,this paper proposes a nonstationary data non-destructive detection method based on BP neural network classifier,and builds a noise model for data transmission in data center optical fiber communication network. The method of combining with the least squares method is performed according to the linear trend term and the nonlinear trend term contained in the data transmission data of the data center fiber communication network,and the data center fiber network communication network data preprocessing is completed; the data center fiber communication network will be obtained. The data preprocessing result is used as a sample training BP neural network. The gradient descent method is used as the BP neural network learning rule and the batch training mode. The weights of each layer of the BP neural network are initialized and the weight is reduced to the training error. The adjustment is performed,and the weight update is performed at the same time to realize the BP neural network classifier training. All the data samples of the data center optical fiber communication network obtained through the preprocessing are input into the trained BP neural network classifier for non-stationary data non-destructive detection. The experimental results show that compared with other methods,the proposed method can achieve high-efficiency,accurate and stable non-destructive detection of non-stationary data transmission in optical fiber communication networks.
Aiming at the shortcomings of low accuracy,poor stability and long time,this paper proposes a nonstationary data non-destructive detection method based on BP neural network classifier,and builds a noise model for data transmission in data center optical fiber communication network. The method of combining with the least squares method is performed according to the linear trend term and the nonlinear trend term contained in the data transmission data of the data center fiber communication network,and the data center fiber network communication network data preprocessing is completed; the data center fiber communication network will be obtained. The data preprocessing result is used as a sample training BP neural network. The gradient descent method is used as the BP neural network learning rule and the batch training mode. The weights of each layer of the BP neural network are initialized and the weight is reduced to the training error. The adjustment is performed,and the weight update is performed at the same time to realize the BP neural network classifier training. All the data samples of the data center optical fiber communication network obtained through the preprocessing are input into the trained BP neural network classifier for non-stationary data non-destructive detection. The experimental results show that compared with other methods,the proposed method can achieve high-efficiency,accurate and stable non-destructive detection of non-stationary data transmission in optical fiber communication networks.
引文
[1]丁驰竹,谭佐军.基于光纤探头的洋葱近红外光谱检测的数值模拟[J].华中农业大学学报,2017,36(4):110-116.
[2]李晓敏,高妍,王宇,等.基于光纤传感技术的局部放电超声信号检测方法研究[J].传感技术学报,2017,30(11):1619-1624.
[3]吴巧巧,周志强,陈文,等.油纸绝缘系统局放光纤传感检测方法[J].哈尔滨理工大学学报,2016,21(4):96-100.
[4]刘学,梁红,玄志武.基于小波模极大值模糊熵的遥测振动信号异常检测[J].振动与冲击,2016,35(9):147-151.
[5]刘炜,刘宏昭.应用模糊熵和相似度的管道泄漏检测与定位[J].机械科学与技术,2016,35(3):437-443.
[6]张燕君,刘文哲,付兴虎,等.基于EMD-AWPP和HO-SA-SVM算法的分布式光纤振动入侵信号的特征提取与识别[J].光谱学与光谱分析,2016,36(2):577-582.
[7]孟祥涛,向政,郭景,等.非均匀采样条件下光纤陀螺微小角振动信号检测技术[J].红外与激光工程,2016,45(3):267-271.
[8]何宁,骆湘红,赵中华,等.基于光纤耦合与相干探测的无损检测方法[J].光学学报,2017,37(8):193-200.
[9]王迪,刘佳,王岩峰,等.基于多信号流模型的电子设备故障诊断方法[J].强激光与粒子束,2017,29(7):95-99.
[10]雷洪利,鲁阳,向新,等.同类2FSK混合信号中微弱信号提取方法研究[J].半导体光电,2016,37(2):238-242.
[11]赵林,王纪强,李振.光纤与电子传感器负压波信号对比实验[J].光电工程,2017,44(6):610-615.
[12]刘国涛,陈皓,徐峥,等.用于非线性兰姆波检测的高灵敏度宽带光纤光栅传感器[J].声学技术,2016,35(3):228-230.
[13]万敏,邱细虾,熊刚,等.基于嵌入式系统的电力光纤信号测试仪的设计与实现[J].电子设计工程,2017,25(13):147-150.
[14]聂德鑫,陈连凯,姜国义,等.非本征F-P局部放电光纤传感器设计及灵敏度分析[J].应用光学,2016,37(6):954-960.
[15]邹东伯,刘海,赵亮,等.分布式光纤振动传感信号识别的研究[J].激光技术,2016,40(1):86-89.
[16]蒋立辉,刘杰生,熊兴隆,等.光纤周界入侵信号特征提取与识别方法的研究[J].激光与红外,2017,47(7):906-913.
[17]周永杰,余震虹,张赵良.基于小波包能量分析的红外火焰信号识别[J].红外技术,2017,39(3):232-236.
[18]廖敏敏,陈伟.通信光缆线路中故障点的定位与检测[J].光通信研究,2015,41(1):26-27.
[2]李晓敏,高妍,王宇,等.基于光纤传感技术的局部放电超声信号检测方法研究[J].传感技术学报,2017,30(11):1619-1624.
[3]吴巧巧,周志强,陈文,等.油纸绝缘系统局放光纤传感检测方法[J].哈尔滨理工大学学报,2016,21(4):96-100.
[4]刘学,梁红,玄志武.基于小波模极大值模糊熵的遥测振动信号异常检测[J].振动与冲击,2016,35(9):147-151.
[5]刘炜,刘宏昭.应用模糊熵和相似度的管道泄漏检测与定位[J].机械科学与技术,2016,35(3):437-443.
[6]张燕君,刘文哲,付兴虎,等.基于EMD-AWPP和HO-SA-SVM算法的分布式光纤振动入侵信号的特征提取与识别[J].光谱学与光谱分析,2016,36(2):577-582.
[7]孟祥涛,向政,郭景,等.非均匀采样条件下光纤陀螺微小角振动信号检测技术[J].红外与激光工程,2016,45(3):267-271.
[8]何宁,骆湘红,赵中华,等.基于光纤耦合与相干探测的无损检测方法[J].光学学报,2017,37(8):193-200.
[9]王迪,刘佳,王岩峰,等.基于多信号流模型的电子设备故障诊断方法[J].强激光与粒子束,2017,29(7):95-99.
[10]雷洪利,鲁阳,向新,等.同类2FSK混合信号中微弱信号提取方法研究[J].半导体光电,2016,37(2):238-242.
[11]赵林,王纪强,李振.光纤与电子传感器负压波信号对比实验[J].光电工程,2017,44(6):610-615.
[12]刘国涛,陈皓,徐峥,等.用于非线性兰姆波检测的高灵敏度宽带光纤光栅传感器[J].声学技术,2016,35(3):228-230.
[13]万敏,邱细虾,熊刚,等.基于嵌入式系统的电力光纤信号测试仪的设计与实现[J].电子设计工程,2017,25(13):147-150.
[14]聂德鑫,陈连凯,姜国义,等.非本征F-P局部放电光纤传感器设计及灵敏度分析[J].应用光学,2016,37(6):954-960.
[15]邹东伯,刘海,赵亮,等.分布式光纤振动传感信号识别的研究[J].激光技术,2016,40(1):86-89.
[16]蒋立辉,刘杰生,熊兴隆,等.光纤周界入侵信号特征提取与识别方法的研究[J].激光与红外,2017,47(7):906-913.
[17]周永杰,余震虹,张赵良.基于小波包能量分析的红外火焰信号识别[J].红外技术,2017,39(3):232-236.
[18]廖敏敏,陈伟.通信光缆线路中故障点的定位与检测[J].光通信研究,2015,41(1):26-27.