基于模糊贴近算法的海量异常激光数据挖掘方法研究

Abnormal laser data mining technique in a large amount of fiber optic data

光纤网络中存在海量激光数据,对异常激光数据的挖掘存在较大困难。提出基于模糊贴近算法的海量异常激光数据挖掘方法,定义出一个激光数据处理域,在处理域中设置模糊贴近算法的模糊贴近度阈值,根据阈值定义正常隶属度函数与异常隶属度函数,得到正常、异常模糊数据集合。在激光数据库中,利用异常激光数据特征平均值表述数据动态特征,根据隶属度约束条件与网络参数实现模糊数据向精准数据的转化,通过分析数据动态特征完成对海量异常激光数据的定位预测。实验结果表明,所提方法的挖掘效率高,内存占用量小,数据聚合性好。

There is a large amount of laser data in the fiber optic network , which is difficult to excavate the abnormal laser data. The method of massive abnormal laser data mining based on fuzzy proximity algorithm is proposed. A laser data processing domain is defined. The threshold of the fuzzy proximity algorithm is set in the processing domain. Normal and abnormal membership functions are defined according to the threshold, and normal and abnormal fuzzy data sets are obtained. In laser database, using the average abnormal laser data characteristic expression data dynamic characteristics, according to the fuzzy membership degree constraint conditions and the network parameters implemente the transformation from fuzzy data to precise data . Through the analysis of dynamic characteristics, the data location prediction of massive abnormal laser data. The experimental results show that the method is very efficient, and the memory consumption is small and the data aggregation is good.