Research On Signal Correction And Feature Extraction In Rolling Bearing Fault Diagnosis

Machine condition monitoring and fault diagnosis are of great significance to ensuring product quality,enhancing productivity and preventing safety accident.Rolling bearing is an important component in rotating machinery and its running state can impact the performance of the whole machine directly.So rolling bearing fault diagnosis is an important issue in the field of equipment fault diagnosis.Our research focuses on removing Doppler distortion for the non-stationary signal and feature extraction of stationary signal under strong background noise.Firstly,the basic structure,failure mode and fault characteristic frequency of rolling bearing are introduced.The development process of bearing defects,the vibration mechanism of failure bearings and the relationship between vibration signal and the bearing state are analyzed.Then,the bearing vibration experiment and the train bearing dynamic acoustic experiment are carried out.In the train bearing dynamic acoustic experiment,a static acoustic experiment platform and a dynamic acoustic experiment platform are designed to simulate the real scene of running trains.Afterwards,this thesis studies the Doppler distortion of the wayside train bearing fault signal.A Doppler shift removal method combining of time-domain interpolation resampling and propagator method is proposed based on the microphone array technology.The wayside train bearing fault signal is first processed by propagator method to estimate the position of train bearing.Then,a resampling time sequence is constructed according to the position of the train bearing.Finally,the Doppler distortion is removed with time-domain interpolation resampling.The simulated signal and experimental signal are used to verify the proposed method,the result shows that the proposed method can reduce the Doppler distortion effectively.Finally,aiming at the feature extraction of the vibration signal under strong background noise,this thesis proposes a time-varying singular value decomposition method to extract the frequency characteristics of bearing faults.The proposed method does not need to reconstruct the signal after decomposition,as in the traditional method based on singular value decomposition.The fault characteristic frequency is directly extracted from the time-varying singular value sequence.The experimental results show that time-varying singular value decomposition method can not only extract fault characteristic frequency effectively,but also has good effect on the separation of compound fault.