A Prognosis Method Evaluating Fault Tendency Percentage for Key Equipment in Urban Rail Transit Signal System Based on Enhanced Naive Bayesian Classifying

Signal System is the most important partof urban rail transit system and serves as the neural nervous part of the network. It is a thorny issue to enhance the surveillance of the key equipment in the urban rail transit signal system, and to do the accurate fault diagnosis and prognosis for them so as to realize active and accurate system equipment maintenance. In order to realize that, certain fault analysis methods, which should be not only able to determine whether certain kind of faults would take place, but be able to forecast the tendency of the fault as well, is needed. Naive Bayesian Classifying (NBC), which is widely used in big data analysis and data mining, is regarded as the basic method in fault diagnosis for the key equipment in urban rail transit; however, it is rarely used in fault prognosis, or it is only used in it combined with other data mining and analysis methods. By thoroughly analyzing the steps in Naive Bayesian Classifying, an intermediate parameter appears in the processing steps that can characterize the tendency if a certain class has been found. Utilizing this intermediate parameter Enhanced Naive Bayesian Classifying (ENBC) method proposed in this paper, the fault tendency percentage for the key equipment in urban rail transit signal system can be realized. By integrating ENBC into the maintenance and management system (MMS) of Shanghai Metro Line 8, and by analyzing a set of one week-long operating data sets, the effectiveness of the fault prognosis method has been proved.