Research On Main Bearing Fault Diagnosis Of Generator Set Based On Vibration Noise

As a core component of operation,the main bearing of an electric generating set is also a component that is very prone to failure,which its safe operation has a vital impact on the whole generator set equipment.Therefore,it is of importance to study the main bearing fault diagnosis method of generator sets.Based on the summary of the existing fault diagnosis technology,this thesis will study the fault diagnosis of main bearing of generator set in terms of its vibration noise.The main research contents of it are as follows.By dynamic analysis,this thesis will explore the vibration characteristics of the crank connecting rod mechanism,which is the main vibration source of the engine part.By using Wokorbench software,the modal calculation and transient dynamics calculation of the engine casing are carried out,so as to obtain the vibration response of the engine casing under the action of reciprocating inertia force.Moreover,by Maxwell software,this thesis will make a further finite element analysis on the radial electromagnetic force which is the main excitation source of the generator part,and finally,the modal analysis is performed on the motor stator model.Through the above analysis,it can be concluded that: The reciprocating inertia force is mainly first-order and second-order;the radial electromagnetic force is larger at the even frequency point;the engine case vibration response frequency is an integer multiple of the engine speed and is related to the intrinsic frequency;the vibration response of the engine case at different locations provides a reference for the placement of acoustic sensors;the bolt connection above the engine case vibrates strongly and can be used to guide the placement of vibration sensors;the reciprocating inertia force and radial electromagnetic force excitation frequencies are different from the modal frequency of the generator set and do not produce resonance.The above conclusions give theoretical support for the following simulation of bearing fault composite signals.In this thesis,the simulation of rolling bearing failure in the display dynamics simulation software LS-DYNA was studied on the failure mechanism of rolling bearings,and the impact of different failures on bearing operation can be seen from the simulation results.On this basis,the fault signal characteristics of rolling bearings are analyzed,focusing on the frequency of the shock generated by surface damage and the causes of vibration noise signal generation.Through the mathematical model of inner and outer ring fault signals,the bearing inner and outer ring fault signals are obtained.The fault signals can be used to verify the effectiveness of the subsequently proposed genetic algorithm-variable modal decomposition algorithm.Concerning to the problems of difficulty in identifying fault features in bearing fault signals,the thesis will present a genetic algorithm-variable modal decomposition method to extract fault feature information.Firstly,the theory of variational modal decomposition algorithm is introduced,which includes parameters such as decomposition scale,penalty factor,noise tolerance and discrimination accuracy,among which the decomposition scale K and penalty factor α have a great influence on the results of variational modal decomposition.Based on the above conclusions,the simulation signal is created to analyze the influence of the number of modes K and penalty α factor on the decomposition effect,and the results show that the variational modal decomposition algorithm can accurately separate each component and effectively eliminate the influence of noise component when the appropriate K and α combination are selected;otherwise,it will generate the phenomenon of modal confusion and introduce too many noise components,or even appear false components,which affects the decomposition accuracy.According to the above analysis,the combination of [K、α] parameters in the variational modal decomposition algorithm is optimized by using genetic algorithm with envelope entropy as the objective function.Finally,the effectiveness of the genetic algorithm-variable modal decomposition algorithm was verified by inputting the bearing fault composite signal and the bearing fault data from Western Reserve University.