Noise Suppression Of Vibration Signal Based On Group Coordination And Its Application In Mechanical Fault Diagnosis

With the development of science and technology,mechanical equipment towards automation,integration,complicated and systematic development,and the coupling between the equipment increase,once the device a component failure not only itself might be a problem,other equipment associated with the device would also be affected,cause a chain reaction,therefore,timely and accurate fault diagnosis is of great significance.The vibration signal not only contains the operating state information of the equipment,but also is easy to be collected.The analysis of vibration signal can judge the equipment efficiently and accurately whether the equipment is running normally.Rolling bearing is a very important component in mechanical equipment,and its reliability is very important for the normal operation of mechanical equipment and even the whole system.This paper mainly studies two important problems in vibration signal fault diagnosis: firstly,we proposed a patch-matching 2 dimensional denoising(PM2D)method to remove noise.The signal after denoising is diagnosed by empirical mode decomposition and envelope analysis.Then,the method of extracting integrated fault features(IFF)is proposed,and SVM is used to classify the faults.The main research contents are as follows:(1)Aiming at the shortcomings of the existing vibration signal denoising methods,a PM2 D denoising method is proposed.This algorithm is mainly inspired by the three-dimensional block matching(BM3D)method,and is proposed according to the local and non local correlation of vibration signals.Due to the high frequency of vibration signal and more noise,PM2 D is divided into two steps,the first part includes three steps: grouping,collaborative filtering and aggregation;the second part repeats the above steps on the basis of the results of the first part to achieve the ideal denoising effect.After signal denoising,envelope analysis is used to judge whether the rolling bearing has faults from the frequency value of envelope spectrum.The method is verified by numerical simulation data and experimental data respectively.(2)Aiming at the problem that it is difficult for fault diagnosis to meet the requirements of effectiveness and rapidity at the same time,this paper proposes an IFF extraction method based on modern signal processing technology,which can effectively and quickly extract the characteristics of vibration signals in different states.Firstly,Hilbert-Huang transform is used to get the Hilbert spectrum.Hilbert-Huang transform is adaptive and has high computational efficiency.However,the dimension of Hilbert spectrum is high and there are many data,so the computational efficiency of further analysis is low.Therefore,the singular value obtained by singular value decomposition of Hilbert spectrum is one of the signal features.Singular value decomposition(SVD)is used to extract the features of Hilbert spectrum,which can reduce the dimension and improve the efficiency.Due to the end effect and mode aliasing of empirical mode decomposition in Hilbert-Huang,the feature of Hilbert spectrum has errors,so permutation entropy is used as another feature of vibration signal to make up for the errors.Experiments show that the two features can accurately express the characteristics of vibration signals under different conditions.After extracting signal features,fault classification is also an important problem.Because SVM has advantages in dealing with classification problems,this paper uses SVM as fault classifier.Finally,this method is applied to the data of two actual rolling bearing faults,and is compared with the classical method.