Research Of Fault Diagnosis Of The Micro-Computer Monitoring System Based On Data Mining

Railway computer monitoring system is important safety equipment for railway, an important part of the modernization of railway equipment. At present, the fault diagnosis of computer monitoring system is still in the initial stage. It judged the run station of traffic signal equipment mainly based on the staff data that the curve collected by the monitoring system, by personal experience to determine, that maintenance efficiency and reliability is relatively low. Therefore, it is the solved task to achieve the the fault diagnosis function of computer monitoring system.Data mining mainly extract the feature in equipment condition monitoring and fault diagnosis, state identification and diagnosis decision. Fault features extraction is one of the most important, critical and the most difficult parts. Meanwhile, the data stream monitored by the computer monitoring system is a time sequence, is rapidly changing, massive and potentially unlimited, and can not store the whole and many times scan, impossible to "see" each element of the data stream. Therefore, the time series data mining excavation is the most challenging one of the top ten researches. The paper combines the own characteristics of computer monitoring system, proposed a new based on fault information extraction of time-based sequence mining. The algorithm estimated by the line density distribution of time-series point sampling value the normal value of equipment, combined the data trends and the comparison situation of the relative changes of data to extrac the fault information, and used the algorithm in three major fault features extraction of switch equimpment, track circuit and power devices under the computer monitor, and through Matlab simulation, the results show that the algorithm applied to extract fault information is feasible.The paper combined the time series data mining of diagnosis petri nets for fault. Firstly, it used the designed data mining algorithms to extract fault information for the the time series that monitored by computer, and then, it inputed the text as diagnostic sample that is extracted data to petri net of customize, analyzed the characteristics of signal equipment failures that monitored by computer, used the petri nets for fault diagnosis method to the corresponding diagnosis, finally, it outputted the diagnosis result from the petri net.The paper analysed the fault symptoms and fault type of railway signal equipment, summarized the characteristics of the signal equipment failures under the computer monitor. It applied the petri nets for fault diagnosis methods to model for the common major equipment fault of track, circuit and the power, and in accordance with the proposed reasoning algorithm that is based on hybrid fault diagnosis of the petri nets, and based on the mining of time series to extract the initial state, and infer possible causes that can cause failure and consequences. By analyzing the monitoring data of the current track, switch current, it obtained the possible degrees which lead to failure causes and the subsequent failure, and reached expected basically purpose from diagnosis result. It proved the correctness and feasibility of fault diagnosis under the monitoring that used the time-series mining methods and petri nets that face the fault diagnosis.