Low adhesion detection and identification in a railway vehicle system using traction motor behaviour

摘要: It is important to monitor the wheel-rail friction coefficient in railway vehicles improve their traction and braking performance as well reduce number of incidents caused by low friction. Model based fault detection identification (FDI) methods, especially state observers have been commonly used previous research However, methods cannot provide an accurate value few them validated using experiments. A Kalman filter estimator proposed this project. The developed uses signals from motor provides a new more efficient approach monitoring condition contact condition. 1/5 scaled test rig has built evaluate method. This comprises 2 axle-hung induction motors driving both wheelsets bogie through pairs spur gears. DC generators are load rollers timing pulleys. independently controlled inverters. Motor parameters such voltage, current speed measured wheel roller output generator incremental encoders Hall-effect clamps. A LabVIEW code designed process all collected data send control commands communication between PC inverters realized Profibus (Process Field Bus) OPC (Object Linking Embedding (OLE) for Process Control) protocol. 3 different estimators were first computer simulations. its two nonlinear developments: extended (EKF) unscented (UKF) these 3 methods. results show that UKF can best case. requirement measuring also studied UKF. it essential measure but absence measurement does not significant impact on estimation accuracy. re-adhesion algorithm, which reduces excessive creepage rail, estimator. Accurate helps work at optimum point. As largest creep force generated, accelerating time distance be reduced minimum values. controller avoid hence potentially wear rail. The development evaluated experiments conducted rig. Three conditions tested: base without contamination, water contamination oil contamination. was varied cover large range creepage. importance studied. shown reliable most tested conditions. confirm necessary give good agreement with simulation results. With experiment work, capability coefficient.