Research On Train Fault Early Warning Method Based On Data Sigmacompleteness Of Train Control System

By interracting the data of train control system realtimely,many subsystems achieve the normal operation of train in every scene.It is very helpful for finding out the hidden safety troubles of train control system and improving the safety of train control system systematically and comprehensively to study data sigmacompleteness of train control system when the train is operating in the different scenes and various complex environmental conditions,especially in the extreme scene.Data sigmacompleteness of train control system(DSTCS)means the characteristics including completeness,correctness,order,real-time,time-effectiveness,compatibility,stability,periodicity,continuity and repeatability of the control data which can ensure the safety of train operation.In this paper,the warning of train's fault based on train control data completeness is the main body.By predicting and mining the data received from historical fault data and online real-time montoring data in every scene,we can extract effective faults features.After that,we analyze and match sigmacompleteness,montor the states of train real-time to find potential safety risks and alarm early.The main work of this paperis is as follow:1.This paper proposes the new method based on DSTCS on diagnosing train's faults and alarm the risks.In the process of analyzing the data of train operation,interface information of equipments,specific order of transforming information and the constraints of time would be converted to corresponding constraints in terms of sigmacompleteness.Also the data of train's faults and methods of giagnosing faults would be generalized.After the fault component and constraints over data sigmacompleteness can be converted to attributes of conditions and decision making in the level of C4.5 in decision tree,we can find the regulations between fault features and sigmacompleteness.When the effective features be extracted from the online realtime data based on rough set theory,we can detect the similarity for fault features by use of regulations extracted from fault data and draw a matching conclusion.So the faults would be located and reasons would be foud.2.We can establish the model over warning of train fault methods and fault position methods based on DSTCS.After clarifying the formal modeling evidences and targets,the analysis of function requirements would be achieved.The model mainly achieve two functions:One is to establish the fault data diagnostic classifier.It belongs to the non-real-time model.The other is to achieve the detection analysis of sigmacompleteness of the unknown online data,and match the analysis results with the fault modes of fault database.It belongs to the real-time model.3.In this paper,the degradation failure scenario is taken as an example for simulation verifying and analyzing based on MATLAB and C#.For improving persuasion,for a certain railway station example,the received 200 cases over train faults would be trained by use of diagnosis classifier based on DSTCS.We draw a conclusion that 97 per cent of faults can be verifyied the accuracy rate of its classification.Finally,we can prove the feasibility of the method in train fault warning by displaying the regulations of decision tree and potential fault types.