Evaluation Method Of Service State Of Tracked Vehicle Based On Security Domain

As one of the most important parts of the rail vehicles, the traveling system supports the weight of the whole vehicle and ensures vehicle running safety on the track. The wheelset, as one of the most important components of the traveling system, is one of the key factors which influence the safe operation of trains. So it’s very significant to develop the online monitoring system for real-time status of wheelset. Currently, the monitoring of wheelset service condition is mainly based on artificial static detection. Based on the research of predecessors, an assessment method of wheelset service status based on safety region is proposed, and the main content of this paper is as follows:Firstly, the main components of the traveling system are introduced. The major faults and their formation mechanism of wheelset are especially introduced. The wheelset service status is assessed by analyzing the track vibration signals. The basic theroy and method of safety region are also introduced.Secondly, a coupling dynamics model of vertical vehicle-track interactions is developed. The model consists of vehicle subsystem, track subsystem and wheel-rail coupling submodel. With wheel flat, wheel out of round and track irregularity as model excitation, the track vibration signals are obtained.Thirdly, an assessment method of wheelset service status based on EMD generalized energy is proposed with only having the normal track vibration signals.The IMFs can be obtained by using EMD on the track vibration signals, and than giving a weight to each IMFs. EMD generalized energy can be obtained by calculating weighted sum of IMFs’ energy. The safety region boundaries of normal wheelset can be ensured by these values of EMD generalized energy. So it can divide the normal wheelset and faulted wheelset through the safety region boundaries.Finally, this paper focuses on the security domain estimation of wheelset service status based on LSSVM and the fault diagnosis of wheelset based on PNN. Different from only using the normal track vibration signals to get the safety region, this method, which contains the signals of normal wheelset and faulted wheelset realizes, realizes the division of normal and faulted wheelset based on LSSVM through getting the eigenvalues from these track vibration signals. PNN is applied to identify the different faults of wheelset. Simulation data and real data all show that the proposed method can be capable of dividing normal and faulted wheelset and identifying different faults of wheelset accurately.