Research On Correlation And Intelligent Sorting Algorithm Based On Rail Traffic Alarm Data

The operation control center(OCC),that is,the integrated monitoring and communication center,is the key central system to ensure the normal operation of the rail transit system.However,the challenge facing the current OCC system is that each subsystem generates a large number of events and alarms every day,which greatly exceeds the ability of the operation team to manually handle it.It is necessary to design an effective alarm correlation model to help O & M personnel locate the fault,quickly solve it,and study An intelligent sorting algorithm for alarms helps operators with no industry experience to prioritize alarm events with a higher severity.First,through in-depth mining and exploration of rail traffic warning data,based on the data,analyze the characteristics of the alarm on the warning system,visually analyze the rail traffic warning log data from various dimensions,find some warning rules,and propose corresponding Comments and suggestions.Then,on the basis of theoretical and technical support,an alarm correlation model is designed,using the information features in the alarm log data,and innovatively using the word2 vec model to vectorize each information feature to carry out alarms within a certain time window.Synthetic information clustering based on semantics and fusion of other features,using the Fp-Growth algorithm to perform association mining between different types of alarms,and then calculating the scores of alarm categories based on the number of points to other categories,confidence,promotion,etc.,Obtain the importance of such alarms in a period of time,and define the most important alarms as root alarms.This part is mainly based on the mining of alarm association rules to design a set of root alarm search methods.Finally,some derived variables are constructed,including the confirmation delay of the learning and operation personnel’s behavior,the entropy information of the site and system,the density of alarms,and short-term effects.Innovatively create a generalized linear model and synthesize these data to obtain more detailed alarm levels.