A bearing fault diagnosis method based on correlation entropy and short-time fourier transform

Abstract

The application provides a bearing fault diagnosis method based on correlation entropy and short-time Fourier transform, and the diagnosis method is as follows: step 1, collecting a vibration signal x(i), the sampling length is N, the signal x(i) is an N*1 column vector, and the kernel matrix M of the signal is calculated x ,M x (i,j) = K[x(i),x(j)], K(·) is a kernel function, e (·) is a natural exponential function, sigma is a kernel length, i, j = 1, 2, 3,..., N, M x is an N*N square matrix, since the traditional short-time Fourier transform is susceptible to interference noise, and the bearing outer ring fault characteristic frequency and the system inherent vibration frequency are coupled with each other, it is difficult to effectively extract the bearing outer ring fault characteristic information under the noise interference, compared with the traditional short-time Fourier transform method, the application can effectively suppress the Gaussian noise and non-Gaussian noise in the signal, has the self-adaptive noise reduction performance, and can highlight the bearing fault characteristics.

Description

Technical Field

[0001] This invention relates to the field of modern signal processing technology, and in particular to a bearing fault diagnosis method based on correlation entropy and short-time Fourier transform. Background Technology

[0002] Short-Time Fourier Transform (STFT) is a commonly used time-frequency signal processing method, widely applied in fault diagnosis of components such as bearings and gears in rotating machinery. However, when the signal-to-noise ratio is low or contains non-Gaussian noise, the performance of traditional STFT signal processing methods degrades or even fails. In actual industrial scenarios, vibration signals collected from electromechanical equipment often contain a large amount of Gaussian and non-Gaussian noise, making it difficult for traditional STFT methods to achieve satisfactory diagnostic results. Therefore, there is an urgent need for a fault diagnosis method that can effectively suppress Gaussian and non-Gaussian noise in the signal, has adaptive noise reduction performance, and can highlight the characteristics of bearing faults. Summary of the Invention

[0003] To address the above technical problems, this invention provides a bearing fault diagnosis method based on correlation entropy and short-time Fourier transform. The diagnosis method of this invention is as follows:

[0004] Step 1: Acquire vibration signal x(i) with a sampling length of N. Signal x(i) is an N×1 column vector. Calculate the kernel matrix M of the signal. x M x (i,j)=κ[x(i),x(j)], where κ(·) is the kernel function. e (·) It is a natural exponential function, where σ is the kernel length, i,j=1,2,3,…,N,M x It is an N×N square matrix.

[0005] Step 2, calculate the correlation entropy V of signal x(i). x (n), V x (n) is an N×1 column vector.

[0006] Step 3, calculate the relevant entropy V x The short-time Fourier transform S of (n) x (t,f), S x (t,f)=STFT[V x [(n)], STFT(·) is the short-time Fourier transform operator, t is time, and f is frequency.

[0007] Step 4, draw the short-time Fourier transform S x The (t,f) graph shows that bearing fault characteristics can be identified from the spectral peaks.

[0008] The beneficial effects of this invention are as follows:

[0009] The bearing fault diagnosis method proposed in this invention, based on correlation entropy and short-time Fourier transform, comprehensively utilizes the advantages of both. Traditional short-time Fourier transform is susceptible to interference noise, and the characteristic frequency of the bearing outer ring fault is coupled with the system's natural vibration frequency, making it difficult to effectively extract bearing outer ring fault feature information under noise interference. Compared with traditional short-time Fourier transform methods, this invention effectively suppresses Gaussian and non-Gaussian noise in the signal, exhibits adaptive noise reduction performance, and highlights bearing fault characteristics. Attached Figure Description

[0010] Figure 1 This is a flowchart of the method described in this invention;

[0011] Figure 2 The time-domain waveform of the bearing outer ring fault vibration signal in Example 2;

[0012] Figure 3 This is a fast Fourier transform diagram of the vibration signal of the bearing outer ring fault in Example 2;

[0013] Figure 4 The vibration signal of the bearing outer ring failure in Example 2 is shown in the relevant entropy diagram at that time.

[0014] Figure 5 The short-time Fourier transform spectrum plane profile diagram of the bearing outer ring fault vibration signal in Example 2.

[0015] Figure 6 The short-time Fourier transform spectrum of the vibration signal of the bearing outer ring fault in Example 2 is shown in three dimensions.

[0016] Figure 7 The short-time Fourier transform spectrum of the bearing outer ring fault vibration signal in Example 2 is the spectrum after integration along the time axis.

[0017] Figure 8 The traditional short-time Fourier transform planar profile of the vibration signal of the bearing outer ring fault in Comparative Example 1 is shown.

[0018] Figure 9 The traditional short-time Fourier transform three-dimensional image of the vibration signal of the bearing outer ring fault in Comparative Example 1 is shown.

[0019] Figure 10 The spectrum of the vibration signal of the bearing outer ring fault in Comparative Example 1 is obtained by integrating the traditional short-time Fourier transform along the time axis. Detailed Implementation

[0020] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings.

[0021] like Figure 1 As shown, this invention discloses a bearing fault diagnosis method based on correlation entropy and short-time Fourier transform, comprising the following steps:

[0022] Step S1: Acquire vibration signal x(i) with a sampling length of N, such as... Figure 2 As shown, signal x(i) is an N×1 column vector. Calculate the kernel matrix M of the signal. x M x (i,j)=κ[x(i),x(j)], where κ(·) is the kernel function. e (·) It is a natural exponential function, where σ is the kernel length, i,j=1,2,3,…,N,M x It is an N×N square matrix.

[0023] Step S2, calculate the correlation entropy V of signal x(i). x (n), V x (n) is an N×1 column vector.

[0024] Step S3, calculate the relevant entropy V x The short-time Fourier transform S of (n) x (t,f), S x (t,f)=STFT[V x [(n)], STFT(·) is the short-time Fourier transform operator, t is time, and f is frequency.

[0025] Step S4, draw the short-time Fourier transform S x The (t,f) graph shows that bearing fault characteristics can be identified from the spectral peaks.

[0026] Example 2

[0027] This embodiment verifies the method of Embodiment 1. The vibration signal x(i) from a bearing outer ring fault is used in this embodiment. The bearing model is a deep groove ball bearing 6205, the shaft speed is 1730 r / min, and the sampling frequency is f. s =12000Hz, number of sampling points N=4096, bearing geometry is: major diameter D=52.0mm; ball diameter d=7.94mm; number of balls z=9; pressure angle α=0°, the bearing outer ring fault characteristic frequency f is calculated. outer =103.36Hz, bearing outer ring fault characteristic period T outer =1 / f outer =0.00967s.

[0028] The time-domain graph of the bearing outer ring fault vibration signal x(i) in Example 2 is shown below. Figure 2 As shown, the Fast Fourier Transform of x(i) is as follows: Figure 3 As shown, according to step 2 of Example 1, the correlation entropy V of signal x(i) is calculated. x (n),

[0029] like Figure 4 As shown, in Figure 4 The data shows regular transient impacts, with the interval between adjacent transient impact peaks being the bearing outer ring fault characteristic period T. outer According to step 3 of Example 1, the relevant entropy V is calculated. x The short-time Fourier transform S of (n) x (t,f), Figure 5 For S x Planar profile of (t,f) Figure 6 For S x A three-dimensional plot of (t,f), Figure 7 for Figure 4 S x The power spectrum after integration along the time axis (t,f) is obtained from... Figure 5 It can be clearly seen that in the low-frequency range of the time-frequency plane (t,f), there are equally spaced spectral peaks, and the interval between adjacent spectral peaks is the bearing outer ring fault characteristic frequency f. outer ,from Figure 7 The characteristic frequency f of the bearing outer ring fault can be clearly seen. outer Its higher harmonics clearly depict the frequency information of bearing outer ring fault characteristics.

[0030] Comparative Example 1

[0031] To compare the bearing fault diagnosis results based on correlation entropy and short-time Fourier transform with those based on traditional short-time Fourier transform, this comparative example uses the traditional short-time Fourier transform method to analyze the bearing fault diagnosis results in Example 2. Figure 2 The vibration signal x(i) of the bearing outer ring fault was analyzed. Figure 8 This is a planar profile diagram based on the traditional short-time Fourier transform. Figure 9 This is a three-dimensional image based on the traditional short-time Fourier transform. Figure 10 for Figure 8 The power spectrum obtained by integrating the time-frequency plot along the time axis, from Figure 8 and Figure 9 It can be seen that, due to the susceptibility of traditional short-time Fourier transform to interference noise, and the coupling between the bearing outer ring fault characteristic frequency and the system's natural vibration frequency, therefore, in Figure 8 No regularly distributed spectral peaks were observed in the low-frequency range. Figure 10 The characteristic frequency f of bearing outer ring failure was not observed in the middle. outerBecause of its high-order harmonics, the traditional short-time Fourier transform method is difficult to effectively extract the fault characteristic information of the bearing outer ring under noise interference.

[0032] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0033] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A bearing fault diagnosis method based on correlation entropy and short-time Fourier transform, characterized in that, Includes the following steps: Step 1, collect vibration signals The sampling length is ,Signal yes Given column vectors, calculate the kernel matrix of the signal. , , It's a kernel function. , It is a natural exponential function. It is the core length, , yes Array; Step 2, Calculate the signal Relevant entropy , , , , yes Column vectors; Step 3, calculate the relevant entropy Short-time Fourier Transform , , It is the short-time Fourier transform operator. It is time. It is frequency; Step 4: Draw the short-time Fourier transform. The figure shows that bearing fault characteristics can be identified by spectral spikes.

Images