Study On Similarity Characterization Of Rolling Bearing Fault Response And Its Influencing Factors

As one of the most common basic parts in mechanical equipment,rolling bearing is widely used in various industries and its working performance often determines the working state of the whole equipment.However,the rolling bearing plays the role of bearing load in the working process and its working environment is very complex,so it is easy to be damaged.Therefore,many scholars at home and abroad have done a lot of research on condition monitoring and fault diagnosis of rolling bearings.The existing fault diagnosis method of bearing is mostly based on the resonance demodulation technique to find out the fault characteristic frequency in the vibration signal and use it as the main basis forjudging the fault.However,there are defects in the method to infer bearing fault by the fault characteristic frequency,which can easily lead to the misinformation of false fault.When the operating condition of the bearing changes dramatically or when the sensor abates to collect the distorted data,this will result in periodic envelopes that are consistent with the fault characteristic frequency.This paper takes the fault response of rolling bearing as the research object,studying the common attributes feature and their influencing factors.The main contents are:Separately using time-domain waveform,piecewise linear representation based on important points,cepstrum and quadratic Fourier transform to realize the characteristic representation of rolling bearing fault response.Separately using uclidean distance,dynamic time bending,symbolic representation,cosine distance and probability statistics to realize the similarity measurement of fault characteristics to the fault response of rolling bearing.And use experimental data to verify the validity of these four feature representation methods and five similarity measures in extracting and measuring the fault information of impulse response.For each characteristic representation method of vibration signal has its own scope of application,this paper proves that quadratic Fourier transform and dynamic time bending are the best feature representation method and measurement method for characterizing and measuring the characteristic information of fault response by comparing and analyzing the similarity results of different similarity measure methods measuring each representation method,obtains the similarity representation of fault response and verifies its correctness.Aiming at the influence of the change of bearing speed,the simulation model of the fault bearing vibration signal under the condition of variable speed is established in this paper.The similarity between the homologous response and the non-homologousresponse at variable rotational speed is analyzed,and the similarity is reduced by the variation of speed which leads to the critical value of dynamic time bending distance of homologous and non-homologous response in the similarity characterization of rolling bearing fault response becoming larger.The effect of rotating speed on the similarity characterization of fault response is verified by the measured data.In view of the influence of signal transmission path,this paper studies the influence of signal transmission path change on the similarity characterization of rolling bearing fault response by changing the location of bearing damage point.When the fault point is located at 12 o'clock in the outer circle,the similarity of the homologous fault response is the highest,but the similarity of the non-homologous fault response is the lowest,and the threshold of dynamic time bending distance between the homologous and non-homologous fault response is the largest in the similarity characterization,so the distinguishing effect is the most obvious.The 3 o'clock direction is next and the 6 o'clock direction is the worst.Aiming at the influence of background noise,the background noise is added to the signal from two angles of simulation model and measured signal,and the influence of background noise on the similarity of fault response is obtained.Secondly,the vibration signal with background noise is processed by network iterative singular spectrum filtering,which completes the effective stripping of background noise and weakens the influence of background noise on the similarity characterization of fault response.Based on the above research,this paper summarizes the similarity characterization of rolling bearing fault response and the influence of off-condition and background noise on the fault response similarity characterization.The method of network iterative singular spectrum analysis is used to effectively peel off the background noise.With the variation of working conditions and the interference of background noise,the dynamic time bending threshold used to distinguish the homologous and non-homologous responses in the similarity characterization of the fault response will change,but the similarity index is still applicable in the characterization.The simulation and experimental results show that the conclusion is accurate and reliable.