Fault Detection Of Oil Pump Bearings Using Embedded Coils

The traction transformer is an important part of the electric locomotive.Its reliable operation is directly related to the stability and safety of the locomotive.The high temperature of locomotive traction transformer will cause many problems,and the oil pump,that is the core of its cooling system,needs more attention.The oil pump motor is generally a three-phase asynchronous motor.This thesis will aim at the online fault detection problem of the oil pump motor.This thesis starts with a theoretical analysis of the motor magnetic field.It reveals that the bearing failure will cause the relative displacement between the stator and rotor of a motor,and the air gap eccentricity will further cause changes in such parameters as air gap permeability,fundamental magnetomotive force,and magnetic flux density.The theory deduces the relationship between the magnetic flux,generated by the fundamental magnetomotive force,and the bearing failure.Then,an electromagnetic analysis model of oil pump is established with Maxwell software;and the modeling,simulation and analysis of the three-phase asynchronous motor are completed.In view of the common eccentricity of the rotor of the motor,the eccentricity fault is simulated and a simulation analysis is performed.The relationship between the eccentric fault of the three-phase asynchronous motor and the change of the internal magnetic field of the motor is obtained.A method to detect the motor faults through the sensing coils embedded in the slot of the motor is studied.The sensing coils measure the relative displacement between the stator and rotor in all directions and serves as the main measures for motor fault diagnosis.The sensing coil positioning and the corresponding sensing principle are analyzed.The output signals of the eight coils are processed by time-division multiplexing to obtain six differential signals,which are then used to sense the radial and axial displacements of the bearings.Additionally,KPCA algorithm is applied to oil pump bearing fault detection.Six sets of induced electromotive force signals of the detection coil pairs are used to form the input vectors of KPCA.Through self-learning of the data under normal conditions,a KPCA model of oil pump bearing failures is established to perform real-time monitoring and evaluation.Finally,field tests are conducted on a test bench with an actual oil pump.Certain shims are intentionally added on the end face of the motor,it is considered as an eccentricity fault.The fault detection method proposed in this thesis is verified under normal condition and faulty condition,respectively.The results show that the combination of sensing coils and KPCA can detect the bearing failure of the oil pump,accurately.The fault detection method based on detection coils to the induction motor bearing not only has a theoretical guiding meaning for oil pump fault diagnosis,but also has applicable value;and it can also provide reference for motor fault detection in other fields.