A Data Preprocessing Optimization Method based on Sliding Time Window in Communication Alarm Events
Wengang Chen, Wenwei Chen, Wenwei Chen, Wenwei Chen, Bing Fan, Runze Wu, Yanru Wang, Junli Fan · 2016
In communication fault analysis, mining history alarm data to acquire correlation knowledge has become an inevitable trend.Owing to low efficiency of traditional alarm preprocessing methods, a new preprocessing optimization method is proposed.The quality of alarm sequence partition is evaluated by difference alarm time segment using sliding time window, and the optimal partition based on double constraint and K-average is introduce to extract the alarm transactions by sliding time window so as to acquire better preprocessing results.The performance simulation indicates that this method has higher preprocessing efficiency of alarm sequence and overcomes the problem that the traditional methods are not fully applicable to the alarm sequence, which provides support for the analysis of the follow-up alarm data.