激光网络故障数据的弱关联挖掘技术  

Weak correlation mining technique for laser network failure data

作　　者：谢光[1] 潘玉霞[1] XIE Guang;PAN Yuxia(Sanya College,Sanya Hainan 572000,Chin)

机构地区：[1]三亚学院

出　　处：《激光杂志》2018年第7期134-138,共5页Laser Journal

基　　金：海南省科技厅自然科学基金项目资助(No.617182)

摘　　要：为了提高对激光网络的故障诊断能力,需要对故障数据的弱关联特征进行有效挖掘,提出一种基于自相关功率谱密度估计的激光网络故障数据的弱关联挖掘技术。首先采用奈奎斯特采样方法进行激光网络传输数据的时间序列采样,对采集的样本数据进行短时傅里叶变换处理,实现采样数据的时频域展开。然后在时频域内对采样数据进行自相关功率谱密度估计,实现故障数据的弱关联特征挖掘。最后进行仿真测试,结果表明,采用本文方法进行故障数据挖掘,能有效实现激光网络故障检测和诊断,对目标故障数据的准确挖掘性能较好。In order to improve the laser network fault diagnosis ability,effective mining for the weak correlation characteristics of fault data is required. Therefore,put forward a weak correlation mining technology of laser network fault data based on autocorrelation power spectral density estimation. First,Nyquist sampling method is used for time sorting of laser network transmission data. Carry out short-time Fourier transform for sampling processing to realize time frequency expansion. Then,the autocorrelation power spectral density data of the sampled data is estimated in the time-frequency domain,and the weak correlation characteristics of the failure data are excavated. Finally,carry out the simulation test,the results show that the method in this paper can effectively achieve the network fault detection and diagnosis and with good performance accurately.

关 键 词：激光网络 故障数据 弱关联 挖掘 诊断 

分 类 号：TN911[电子电信—通信与信息系统]