Fault diagnosis method, device, system and storage medium

Abstract

The present invention provides a fault diagnosis method, device, system, and storage medium, belonging to the field of fault diagnosis. The method comprises: performing Fourier transform analysis on a vibration acceleration signal to obtain multiple two-dimensional time-frequency signals; screening multiple target high-time-frequency resolution time-frequency spectra from the multiple two-dimensional time-frequency signals; analyzing all target high-time-frequency resolution time-frequency spectra to obtain an envelope spectrum; and performing fault diagnosis on the envelope spectrum based on all fault characteristic frequency ranges and the fault types corresponding to all fault characteristic frequency ranges to obtain a fault diagnosis result. The present invention can more accurately and effectively extract and separate the impact component of high-speed rotating machinery based on high-resolution time-frequency representation, realizes the pulse feature extraction of the impact component, and can be used for the diagnosis of impact-related faults in high-speed rotating machinery, thereby improving the accuracy of fault diagnosis.

Description

Technical Field

[0001] The present invention mainly relates to the technical field of fault diagnosis, and in particular to a fault diagnosis method, device, system and storage medium. Background Art

[0002] Typical impact-related faults in large, high-speed rotating machinery include rolling bearing defects, gear defects, and friction between moving and static parts. These faults are highly hazardous and can seriously impact the safe operation of the equipment. Impact-related faults generate periodic pulses exhibiting extremely non-stationary characteristics. These pulses can be monitored by vibration sensors installed on high-speed rotating machinery. However, due to the small initial defect size, the small amplitude of the impact component signal, and interference from environmental noise, it is difficult to detect the weak pulse component generated by the initial defect. By extracting and analyzing the impact component in the vibration signal, it is possible to determine the fault type and provide early warning, thereby avoiding serious accidents. Therefore, accurately extracting the impact component and its fault characteristics from non-stationary vibration signals is an important research topic in impact-related fault diagnosis.

[0003] To address the above issues, scholars at home and abroad have proposed various signal processing methods and explored their applications in extracting multiple vibration components, such as empirical mode decomposition and its improved methods, multi-source sparse decomposition, and parametric time-frequency analysis. Empirical mode decomposition and its improved methods exhibit two-dimensional time-frequency bandpass filtering characteristics. Based on the framework of time-frequency analysis, they have been improved and nonlinear mode decomposition has been proposed. This method performs well in processing signals whose instantaneous frequency varies slowly over time, but suffers from problems such as uncertainty in the number of modal separations, endpoint effects, modal aliasing, and low computational efficiency. The key to multi-source sparse decomposition is to construct a dictionary that matches the inherent characteristics of the signal itself. However, in practical applications, the strong frequency-varying characteristics of non-stationary signals are unknown and complex and diverse. Therefore, it is unrealistic to analyze all signals by constructing a dictionary or mathematical model that is consistent with the characteristics of the signal itself.

[0004] Time-frequency analysis is a powerful analytical tool for non-stationary signals. Existing time-frequency analysis methods, such as wavelet transform and short-time Fourier transform, are primarily suitable for analyzing linear, stationary vibration signals. However, due to the influence of Heisenberg uncertainty, their time-frequency extraction results suffer from low time-frequency resolution, unfocused energy dispersion, and fuzzy time-frequency features when dealing with non-stationary signals such as shocks. This makes it difficult to accurately separate and extract the shock component. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a fault diagnosis method, device, system and storage medium in response to the deficiencies in the prior art.

[0006] The present invention solves the above technical problems with the following technical solutions: A fault diagnosis method comprising the following steps:

[0007] Obtaining a vibration acceleration signal of the high-speed rotating machinery to be diagnosed from a vibration acceleration sensor, and performing Fourier transform analysis on the vibration acceleration signal to obtain multiple two-dimensional time-frequency signals;

[0008] Screening out a plurality of target high time-frequency resolution time-spectra from the plurality of two-dimensional time-frequency signals;

[0009] Analyzing all target high time-frequency resolution time-frequency spectra to obtain envelope spectra;

[0010] Import multiple fault characteristic frequency ranges and the fault type corresponding to each of the fault characteristic frequency ranges, perform fault diagnosis on the envelope spectrum according to all the fault characteristic frequency ranges and the fault types corresponding to all the fault characteristic frequency ranges, and obtain a fault diagnosis result.

[0011] Another technical solution of the present invention to solve the above technical problem is as follows: a fault diagnosis device comprising:

[0012] A Fourier transform analysis module is used to obtain a vibration acceleration signal of the high-speed rotating machinery to be diagnosed from a vibration acceleration sensor, and perform Fourier transform analysis on the vibration acceleration signal to obtain multiple two-dimensional time-frequency signals;

[0013] A screening module, configured to screen out a plurality of target high time-frequency resolution time-frequency spectra from the plurality of two-dimensional time-frequency signals;

[0014] An envelope spectrum analysis module is used to analyze all target high time-frequency resolution time-frequency spectra to obtain envelope spectra;

[0015] The fault diagnosis result acquisition module is used to import multiple fault characteristic frequency ranges and the fault type corresponding to each of the fault characteristic frequency ranges, perform fault diagnosis on the envelope spectrum according to all the fault characteristic frequency ranges and the fault types corresponding to all the fault characteristic frequency ranges, and obtain a fault diagnosis result.

[0016] Based on the above-mentioned fault diagnosis method, the present invention also provides a fault diagnosis system.

[0017] Another technical solution of the present invention to solve the above technical problems is as follows: a fault diagnosis system includes a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the computer program, the fault diagnosis method described above is implemented.

[0018] Based on the above-mentioned fault diagnosis method, the present invention also provides a computer-readable storage medium.

[0019] Another technical solution of the present invention to solve the above technical problem is as follows: a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the fault diagnosis method as described above is implemented.

[0020] The beneficial effects of the present invention are: a two-dimensional time-frequency signal is obtained by Fourier transform analysis of a vibration acceleration signal, a target high-time-frequency resolution time-frequency spectrum is screened out from the two-dimensional time-frequency signal, an envelope spectrum is obtained by analyzing the target high-time-frequency resolution time-frequency spectrum, and a fault diagnosis result is obtained by diagnosing the envelope spectrum according to the fault characteristic frequency range and the fault type. The impact component of the high-speed rotating machinery can be more accurately and effectively extracted and separated on the basis of high-resolution time-frequency characterization, and the pulse feature extraction of the impact component is realized. It can be used for the impact fault diagnosis of high-speed rotating machinery, and the accuracy of fault diagnosis is also improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A schematic diagram of a flow chart of a fault diagnosis method provided by an embodiment of the present invention;

[0022] Figure 2 The vibration time domain signal of the outer casing of the high-speed rotating machinery provided by the embodiment of the present invention and its short-time Fourier transform time-frequency distribution diagram;

[0023] Figure 3 The most significant time-frequency amplitude frequency points extracted according to the embodiment of the present invention constitute the envelope spectrum and the pulse feature extraction result thereof;

[0024] Figure 4 A high-resolution time-frequency distribution diagram processed by a local maximum search algorithm and a synchronization extraction algorithm provided in an embodiment of the present invention;

[0025] Figure 5 The reconstructed vibration waveform of the impact component and the vibration waveform of the remaining components after the impact component provided by the embodiment of the present invention;

[0026] Figure 6 The envelope spectrum result of envelope analysis of the reconstructed impact component provided by the embodiment of the present invention;

[0027] Figure 7 This is a module block diagram of a fault diagnosis device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0028] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0029] Figure 1 A flowchart of a fault diagnosis method provided by an embodiment of the present invention.

[0030] like Figure 1 As shown, a fault diagnosis method includes the following steps:

[0031] Obtaining a vibration acceleration signal of the high-speed rotating machinery to be diagnosed from a vibration acceleration sensor, and performing Fourier transform analysis on the vibration acceleration signal to obtain multiple two-dimensional time-frequency signals;

[0032] Screening out a plurality of target high time-frequency resolution time-spectra from the plurality of two-dimensional time-frequency signals;

[0033] Analyzing all target high time-frequency resolution time-frequency spectra to obtain envelope spectra;

[0034] Import multiple fault characteristic frequency ranges and the fault type corresponding to each of the fault characteristic frequency ranges, perform fault diagnosis on the envelope spectrum according to all the fault characteristic frequency ranges and the fault types corresponding to all the fault characteristic frequency ranges, and obtain a fault diagnosis result.

[0035] It should be understood that the vibration acceleration sensor is used to collect the vibration acceleration signal of the high-speed rotating machinery to be analyzed (ie, the vibration acceleration signal).

[0036] Specifically, a BK4519 accelerometer (the vibration acceleration sensor) was used to measure the vibration acceleration signal of high-speed rotating machinery. The sensor was fixed to a designed bracket, which was bolted to the outer housing bearing seat. During the test, the speed was set at 1800 rpm. All bearings tested had outer ring faults, with a fault frequency of 144.96 Hz and a sampling frequency of 16384 Hz.

[0037] Specifically, the vibration acceleration sensor is used to collect the vibration acceleration signal of the high-speed rotating machinery to be analyzed, the vibration acceleration sensor is installed on the bracket, and then the bracket is installed on the housing of the high-speed rotating machinery, and the sampling frequency f of the original signal is set. s The signal after outlier processing (ie, the vibration acceleration signal) is used as the vibration acceleration signal required for time-frequency analysis for analysis.

[0038] It should be understood that the sampling frequency is 16384 Hz, and the outer ring fault bearing is used, and its fault frequency is 144.96 Hz, which is used to determine the accuracy of subsequent diagnosis.

[0039] In the above embodiment, a two-dimensional time-frequency signal is obtained by Fourier transform analysis of the vibration acceleration signal, a target high time-frequency resolution time-frequency spectrum is screened out from the two-dimensional time-frequency signal, the target high time-frequency resolution time-frequency spectrum is analyzed to obtain an envelope spectrum, and the fault diagnosis result of the envelope spectrum is obtained according to the fault characteristic frequency range and the fault type. This can more accurately and effectively extract and separate the impact component of the high-speed rotating machinery on the basis of high-resolution time-frequency representation, realize the pulse feature extraction of the impact component, can be used for the impact fault diagnosis of high-speed rotating machinery, and also improve the accuracy of fault diagnosis.

[0040] Optionally, as an embodiment of the present invention, Figure 1 and 2 As shown, the process of performing Fourier transform analysis on the vibration acceleration signal to obtain multiple two-dimensional time-frequency signals includes:

[0041] intercepting the vibration acceleration signal according to a preset signal length to obtain a signal sequence;

[0042] The signal sequence is time-frequency-space expanded using a short-time Fourier transform algorithm to obtain multiple two-dimensional time-frequency signals.

[0043] It should be understood that the signal interval to be analyzed is intercepted, and the short-time Fourier transform (ie the short-time Fourier transform algorithm) is used to perform time-frequency analysis on the impulse fault signal (ie the signal sequence).

[0044] Specifically, the preprocessed acceleration vibration signal (i.e., the vibration acceleration signal) is used as the signal to be analyzed, the signal sequence of appropriate length is intercepted, a Gaussian window is selected as the window function of the short-time Fourier transform, and the length parameter of the window function is set according to the intercepted signal length. The one-dimensional time series signal is expanded to a two-dimensional time-frequency space to obtain a short-time Fourier transform time-frequency distribution result with lower time-frequency resolution (i.e., the two-dimensional time-frequency signal).

[0045] It should be understood that the interception time is 0.25s and the number of sampling points is 4096. The Gaussian window is selected as the window function of the time-frequency analysis, and the window length parameter hlength is set to 200. The short-time Fourier transform calculation results are as follows Figure 2 As shown, Figure 2 It can be seen that the time-frequency capability of the results is divergent, the time-frequency resolution is low, and it is difficult to accurately extract the impact component.

[0046] In the above embodiment, the vibration acceleration signal is subjected to Fourier transform analysis to obtain multiple two-dimensional time-frequency signals, which can remove abnormal data and accurately obtain useful data. It can more accurately and effectively extract and separate the impact component of high-speed rotating machinery based on high-resolution time-frequency representation, thereby improving the accuracy of fault diagnosis.

[0047] Optionally, as an embodiment of the present invention, Figure 1 and 3 As shown, the process of selecting a plurality of target high time-frequency resolution time-frequency spectra from a plurality of the two-dimensional time-frequency signals includes:

[0048] Calculating estimated values ​​of each of the two-dimensional time-frequency signals respectively to obtain estimated values ​​corresponding to each of the two-dimensional time-frequency signals;

[0049] Calculating the original high time-frequency resolution time spectra of each of the two-dimensional time-frequency signals and the estimated values ​​corresponding to each of the two-dimensional time-frequency signals respectively, to obtain the original high time-frequency resolution time spectra corresponding to each of the two-dimensional time-frequency signals;

[0050] Calculating the envelope spectrum of each of the original high time-frequency resolution time-frequency signals using a Hilbert envelope demodulation algorithm to obtain an envelope spectrum corresponding to each of the two-dimensional time-frequency signals;

[0051] Filtering out a maximum value from all the envelope spectra, and obtaining a maximum envelope spectrum after filtering;

[0052] A difference between the maximum envelope spectrum and a first preset range value and a sum of the maximum envelope spectrum and a second preset range value are used as a screening range, and multiple target high time-frequency resolution time-frequency spectra are screened out from all the original high time-frequency resolution time-frequency spectra according to the screening range.

[0053] It should be understood that the envelope spectrum corresponding to the high-resolution time-frequency spectrum (i.e., the original high-time-frequency resolution time-frequency spectrum) is extracted, and the pulse characteristics (i.e., the target high-time-frequency resolution time-frequency spectrum) are further extracted based on the frequency corresponding to the maximum position of the envelope spectrum (i.e., the maximum envelope spectrum).

[0054] It should be understood that the Hilbert envelope demodulation algorithm is specifically as follows:

[0055] A signal x(t) whose envelope is to be determined is Hilbert transformed to obtain HHT(x(t)), and a signal x(t)+j*HHT(x(t)) is synthesized. Then, the amplitude part y(t)=Amp([x(t)+j*HHT(x(t))]) of this synthesized signal is taken. At this time, y(t) is the obtained envelope result (i.e., the envelope spectrum).

[0056] Specifically, if Figure 3 As shown in , since pulse signals usually have broadband properties, under high-resolution time-frequency spectrum, the frequency points with the most significant time-frequency amplitude under broadband are calculated to form the envelope spectrum. The calculation results of the time-frequency envelope spectrum are as follows: Figure 3 , Figure 3In the filter, the highest peak value in the envelope spectrum (i.e., the maximum envelope spectrum) is 4700 Hz. It can be obtained that the pulse feature is most significant at 4700 Hz. Based on this, the frequency point with the most significant time-frequency amplitude in a wide bandwidth can be accurately calculated. The interval between frequency points in the group delay trajectory can be used to accurately describe the pulse interval and further extract the pulse feature.

[0057] Specifically, Figure 3 By extracting the time-frequency information at 4700 Hz and forming time-frequency slices, the timing pulse information corresponding to the frequency point with the most significant time-frequency amplitude can be obtained. The time interval between two adjacent pulses is calculated to be 0.0068 s, which is converted into a frequency of 147.06 Hz, which is basically consistent with the fault frequency of the outer ring fault bearing and can accurately describe the pulse characteristics of the fault.

[0058] In the above embodiment, by screening out multiple target high-time-frequency resolution time-spectra from multiple two-dimensional time-frequency signals, the group delay information of the signal can be accurately obtained. By extracting the group delay ridge component of the time-spectra, the ideal time-spectra of the impact-type vibration signal is constructed to achieve the purpose of reducing noise interference and improving the energy concentration of the time-spectra. Ultimately, a time-spectra with high time-frequency resolution is obtained, which improves the time-frequency resolution of the traditional short-time Fourier transform, and effectively realizes the extraction of impact-type fault features, thereby guiding the diagnosis of impact-type faults.

[0059] Optionally, as an embodiment of the present invention, the process of respectively calculating the estimated value of each of the two-dimensional time-frequency signals to obtain the estimated value corresponding to each of the two-dimensional time-frequency signals includes:

[0060] The estimated values ​​of each of the two-dimensional time-frequency signals are calculated respectively by the first formula to obtain the estimated values ​​corresponding to each of the two-dimensional time-frequency signals. The first formula is:

[0061]

[0062] Among them, t m (t,ω) is the estimated value of the ωth frequency at the tth moment, t is time, ω is frequency, G(t,ω) is the two-dimensional time-frequency signal of the ωth frequency at the tth moment, |G(t,ω)| is the modulus of the two-dimensional time-frequency signal of the ωth frequency at the tth moment, and Δ is the preset local maximum frequency search range.

[0063] It should be understood that a suitable local maximum search range is set and a local maximum search algorithm is used to estimate the group delay of the impulse signal.

[0064] Specifically, an appropriate local maximum search range is set, and the local maximum search algorithm is used to estimate the group delay of the impulse signal in the short-time Fourier transform time-frequency distribution result. The formula of this process can be expressed as:

[0065]

[0066] Where t represents time, ω represents frequency, and t m (t, ω) represents a group delay estimation operator, |G(t, ω)| represents the energy at the corresponding position of the two-dimensional time-frequency spectrum in the short-time Fourier transform; Δ represents the local maximum frequency search range), so that the group delay estimation value of each signal component (i.e., the estimated value) can be obtained.

[0067] It should be understood that a local maximum search algorithm is used to calculate the group delay estimates of all components in the time-frequency distribution, the local maximum search step is set to 100, and a local maximum search is performed along the time axis to obtain the location of the group delay of all components.

[0068] In the above embodiment, the estimated values ​​of each two-dimensional time-frequency signal are calculated separately to obtain the estimated values, so that the group delay information of the signal can be accurately obtained. By extracting the group delay ridge component of the time-frequency spectrum, the ideal time-frequency spectrum of the impact vibration signal is constructed to achieve the purpose of reducing noise interference and improving the energy concentration of the time-frequency spectrum, and finally obtaining a time-frequency spectrum with high time-frequency resolution.

[0069] Optionally, as an embodiment of the present invention, Figure 1 and 4 As shown, the process of respectively calculating each of the two-dimensional time-frequency signals and the original high time-frequency resolution time-spectra of the estimated value corresponding to each of the two-dimensional time-frequency signals to obtain the original high time-frequency resolution time-spectra corresponding to each of the two-dimensional time-frequency signals includes:

[0070] The original high time-frequency resolution time-frequency spectra corresponding to each of the two-dimensional time-frequency signals and the estimated values ​​corresponding to each of the two-dimensional time-frequency signals are calculated respectively by the second formula to obtain the original high time-frequency resolution time-frequency spectra corresponding to each of the two-dimensional time-frequency signals. The second formula is:

[0071] Ts(t,ω)=G(t,ω)δ(ω-t m (t,ω)),

[0072] Among them, Ts(t,ω) is the original high time-frequency resolution time spectrum of the ωth frequency at the tth moment, ω is the frequency, G(t,ω) is the two-dimensional time-frequency signal of the ωth frequency at the tth moment, t m (t, ω) is the estimated value of the ωth frequency at the tth moment, and δ is the Dirac operation assigner.

[0073] It should be understood that the group delay is estimated using the local maximum search algorithm, and the synchronous extraction algorithm is used to accurately extract the impulse signal (i.e., the two-dimensional time-frequency signal) to obtain a time-frequency spectrum with high time-frequency resolution (i.e., the original high time-frequency resolution time-frequency spectrum).

[0074] Specifically, based on the group delay operator estimated by local maximum search, a synchronous extraction algorithm is used to extract the group delay ridge component of the time-frequency spectrum and construct the ideal time-frequency spectrum of the impact vibration signal, thereby reducing noise interference and improving the energy concentration of the time-frequency spectrum. The process of the synchronous extraction algorithm can be expressed as follows:

[0075] Ts(t,η)=G(t,ω)δ(η-t m (t,ω)),

[0076] Wherein, Ts(t, η) represents the reconstructed high time-frequency resolution time spectrum, δ represents the Dirac operation allocator, and finally the high time-frequency resolution time spectrum (i.e. the original high time-frequency resolution time spectrum) is obtained. The high-resolution time spectrum result is as follows: Figure 4 As shown, Figure 4 In the local magnified image, it can be seen that the time-frequency aggregation of the results is significantly improved, the time-frequency resolution is higher, and it is more conducive to the accurate extraction of the impact component.

[0077] In the above embodiment, the original high-time-frequency resolution time-frequency spectrum of the two-dimensional time-frequency signal and the estimated value are calculated respectively to obtain the original high-time-frequency resolution time-frequency spectrum, and the ideal time-frequency spectrum of the impact vibration signal is constructed, which achieves the effect of reducing noise interference and improving the energy concentration of the time-frequency spectrum, and helps to achieve accurate extraction of the impact component.

[0078] Optionally, as an embodiment of the present invention, Figure 1 and 5 As shown, the process of analyzing all the target high time-frequency resolution time-frequency spectra to obtain envelope spectra includes:

[0079] The reconstructed time series signals of all the target high time-frequency resolution time-frequency spectra are calculated by the third formula to obtain the reconstructed time series signals. The third formula is:

[0080]

[0081] Where s(t) is the reconstructed time series signal, g(0) is the function value of the window function at point 0, Ts(t,ω) is the original high time-frequency resolution time spectrum of the ωth frequency at the tth time, and e is the base of the natural logarithm.

[0082] Extracting a signal from the reconstructed time series signal using a Hilbert transform algorithm to obtain an envelope signal;

[0083] Performing Fourier transform on the envelope signal to obtain an envelope spectrum.

[0084] It should be understood that the high-frequency impact range is determined based on the envelope spectrum, and the inverse Fourier transform is used to separate and reconstruct the impact components.

[0085] It should be understood that the envelope spectrum analysis is performed on the impulse component extracted based on the time-frequency reconstruction (ie, the reconstructed time series signal).

[0086] Specifically, based on the envelope spectrum, it is determined that the high-frequency impact range is approximately 3200 Hz to 5500 Hz. Therefore, the inverse Fourier transform is used to separate and reconstruct the impact component (i.e., the target high time-frequency resolution time-frequency spectrum). The time-frequency reconstruction formula can be expressed as:

[0087] Where s(t) represents a reconstructed impulse component time series signal, g(0) represents the function value of the window function at point 0, Ts(t,η) represents the reconstructed high time-frequency resolution time-spectrum, and e represents the base of the natural logarithm.

[0088] It should be understood that if Figure 5 As shown in the figure, the vibration waveform of the impact component is reconstructed by relying on the time-frequency information such as the energy returned by the high-frequency impact range, and the time domain diagram of the reconstructed vibration waveform of the impact component is obtained. The noise and the remaining components can be obtained by subtracting the reconstructed impact component from the original signal.

[0089] Specifically, the reconstructed vibration waveform of the impact component (ie, the reconstructed time series signal) is first subjected to Hilbert transform to obtain the envelope signal of the signal, and then subjected to Fourier transform (FFT) to obtain the envelope spectrum thereof.

[0090] In the above embodiment, all target high-resolution time-frequency spectra are analyzed to obtain envelope spectra, which can realize separation and reconstruction of shock components and obtain reconstructed time-domain waveforms of shock components, thereby guiding shock-related fault diagnosis.

[0091] Optionally, as an embodiment of the present invention, Figure 1 and 6 As shown, the process of performing fault diagnosis on the envelope spectrum according to all the fault characteristic frequency ranges and the fault types corresponding to all the fault characteristic frequency ranges to obtain a fault diagnosis result includes:

[0092] Extracting the characteristic frequencies of the first N highest peaks from the envelope spectrum to obtain a plurality of characteristic frequencies to be verified;

[0093] Verify whether the plurality of characteristic frequencies to be verified are all within any one of all the fault characteristic frequency ranges; if the verification is successful, take the fault type corresponding to the fault characteristic frequency range as the fault diagnosis result.

[0094] Preferably, the N may be 3.

[0095] It should be understood that the shock-type fault diagnosis can be achieved based on the fault frequency (ie, the characteristic frequency to be verified).

[0096] Specifically, if Figure 6 As shown, the frequency doubling of the fault characteristic frequency is about 148.1H. The fault characteristic frequency (i.e., the characteristic frequency to be verified) is extracted from the envelope spectrum and compared with the theoretical fault characteristic frequency of the faulty component (i.e., the fault characteristic frequency range). The result is basically consistent with the fault frequency of the bearing with outer ring fault, so it can be determined that the impact fault is a bearing outer ring fault.

[0097] In the above embodiment, the envelope spectrum is used for fault diagnosis according to all fault characteristic frequency ranges and fault types to obtain the fault diagnosis results, which can more accurately and effectively extract and separate the impact component of the high-speed rotating machinery on the basis of high-resolution time-frequency characterization, realize the pulse feature extraction of the impact component, and can be used for the impact fault diagnosis of high-speed rotating machinery, and also improve the accuracy of fault diagnosis.

[0098] Optionally, as another embodiment of the present invention, the present invention can more accurately and effectively extract and separate the impact components of high-speed rotating machinery based on high-resolution time-frequency characterization, realize the pulse feature extraction of the impact components, and can be used for impact fault diagnosis of high-speed rotating machinery.

[0099] Alternatively, as another embodiment of the present invention, the present invention can more accurately and effectively extract and separate the impact component of high-speed rotating machinery based on high-resolution time-frequency distribution, achieving pulse feature extraction of the impact component, which can be used for impact-related fault diagnosis of high-speed rotating machinery. The method includes the following steps: using a vibration acceleration sensor to collect the vibration acceleration signal of the high-speed rotating machinery to be evaluated; obtaining the time-frequency distribution of the vibration signal through a short-time Fourier transform; then estimating the group delay parameter of the impact signal through a local maximum search algorithm; from this, accurately extracting the impact signal using a synchronous extraction algorithm to obtain a time-frequency spectrum with high time-frequency resolution; calculating the spectral amplitude of each frequency in the high-resolution time-frequency spectrum, extracting the maximum position of its envelope spectrum as the pulse feature, determining the high-frequency impact range based on the envelope spectrum, and finally separating and reconstructing the impact component using an inverse Fourier transform; and performing envelope spectrum analysis based on the time-frequency reconstructed impact component to determine the fault characteristic frequency, which can be used to guide fault diagnosis of high-speed rotating machinery.

[0100] Optionally, as another embodiment of the present invention, the present invention provides a high-resolution time-frequency analysis method for high-speed rotating machinery impact fault diagnosis, wherein a vibration acceleration sensor is mounted on a bracket, and the bracket is then mounted on the housing of the high-speed rotating machinery, and the housing vibration acceleration signal of the high-speed rotating machinery is collected and obtained, and the signal is used for time-frequency analysis after data interception and abnormal data removal. By combining the local maximum search method and the synchronous extraction transformation, the group delay information of the signal can be accurately obtained, and the ideal time-frequency spectrum of the impact-type vibration signal can be constructed by extracting the group delay ridge component of the time-frequency spectrum to achieve the purpose of reducing noise interference and improving the energy concentration of the time-frequency spectrum, and finally obtaining a time-frequency spectrum with high time-frequency resolution. The present invention completes the high-resolution time-frequency characterization of the impact component, and the fast frequency-changing signal provided by the high-resolution time-frequency spectrum can realize the feature extraction and analysis of the impact-type vibration signal, thereby guiding the extraction of early impact-type fault features.

[0101] Optionally, as another embodiment of the present invention, the present invention can accurately calculate the frequency points with the most significant time-frequency amplitudes under the wide band based on the high-resolution time-frequency spectrum and the wide-band characteristics of the pulse signal, and accurately describe the pulse intervals by using the intervals between the frequency points under the group delay trajectory to characterize the characteristics of impact-type faults; in addition, based on the properties of the inverse Fourier transform, the separation and reconstruction of the impact components can be achieved to obtain the reconstructed time domain waveform of the impact components, thereby guiding the diagnosis of impact-type faults.

[0102] Optionally, as another embodiment of the present invention, the present invention improves the time-frequency resolution of traditional short-time Fourier transform through a local maximum search algorithm and a synchronous extraction algorithm. Under the high-resolution time-frequency spectrum, the frequency points with the most significant time-frequency amplitude under a wide band are calculated to form an envelope spectrum. The envelope spectrum can describe the amplitude capacity of each frequency band under an impact fault. The frequency point with the largest envelope spectrum amplitude is used to characterize the pulse characteristics of the fault, and the properties of the inverse Fourier transform are used to realize the separation and reconstruction of the impact component, thereby effectively realizing the extraction of impact fault characteristics, thereby guiding the diagnosis of impact faults.

[0103] Alternatively, as another embodiment of the present invention, Figures 2 to 6 As shown in the figure, the present invention measures the time domain waveform of the vibration acceleration signal of the outer casing of the high-speed rotating machinery and the short-time Fourier transform time-frequency distribution as shown in the figure. Figure 2 As shown in the time-frequency distribution, it can be clearly seen that the short-time Fourier transform result has the defect of low time-frequency resolution. In the present invention, the short-time Fourier transform result is post-processed, and the local maximum search algorithm and the synchronous extraction algorithm are used to obtain the high-resolution time-frequency spectrum result as shown in the figure. Figure 3 As shown, Figure 3 The results show that the time-frequency aggregation and resolution of the time-frequency distribution are improved, which can be used for the subsequent separation, extraction and reconstruction of shock components. Figure 4 The most significant time-frequency amplitude frequency point envelope spectrum and pulse feature extraction results are shown. Figure 4 It shows that the present invention can realize the pulse feature extraction of the impact component for non-stationary vibration signals. Figure 5 It is the reconstructed vibration waveform of the impact component and the vibration waveform of the remaining components after the component, which can be used to extract and analyze the fault characteristic frequency of the impact component, and then guide the early diagnosis of impact-related faults. Figure 6 It is the envelope spectrum result of reconstructing the impact component and extracting the fault characteristic frequency, which can be used to determine the category of impact type faults, thereby realizing impact type fault diagnosis.

[0104] Figure 7 This is a module block diagram of a fault diagnosis device provided by an embodiment of the present invention.

[0105] Alternatively, as another embodiment of the present invention, Figure 7 As shown, a fault diagnosis device includes:

[0106] A Fourier transform analysis module is used to obtain a vibration acceleration signal of the high-speed rotating machinery to be diagnosed from a vibration acceleration sensor, and perform Fourier transform analysis on the vibration acceleration signal to obtain multiple two-dimensional time-frequency signals;

[0107] A screening module, configured to screen out a plurality of target high time-frequency resolution time-frequency spectra from the plurality of two-dimensional time-frequency signals;

[0108] An envelope spectrum analysis module is used to analyze all target high time-frequency resolution time-frequency spectra to obtain envelope spectra;

[0109] The fault diagnosis result acquisition module is used to import multiple fault characteristic frequency ranges and the fault type corresponding to each of the fault characteristic frequency ranges, perform fault diagnosis on the envelope spectrum according to all the fault characteristic frequency ranges and the fault types corresponding to all the fault characteristic frequency ranges, and obtain a fault diagnosis result.

[0110] Alternatively, another embodiment of the present invention provides a fault diagnosis system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the fault diagnosis method described above is implemented. The system may be a computer or other system.

[0111] Optionally, another embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the fault diagnosis method as described above is implemented.

[0112] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0113] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0114] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is merely a logical functional division. In actual implementation, other division methods may be used, such as combining or integrating multiple units or components into another system, or ignoring or not implementing certain features.

[0115] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected based on actual needs to achieve the objectives of the embodiments of the present invention.

[0116] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0117] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program code.

[0118] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A fault diagnosis method, characterized in that: The steps include: Obtaining a vibration acceleration signal of the high-speed rotating machinery to be diagnosed from a vibration acceleration sensor, and performing Fourier transform analysis on the vibration acceleration signal to obtain multiple two-dimensional time-frequency signals; Screening out a plurality of target high time-frequency resolution time-spectra from the plurality of two-dimensional time-frequency signals; Analyzing all target high time-frequency resolution time-frequency spectra to obtain envelope spectra; Importing multiple fault characteristic frequency ranges and fault types corresponding to each of the fault characteristic frequency ranges, performing fault diagnosis on the envelope spectrum according to all the fault characteristic frequency ranges and the fault types corresponding to all the fault characteristic frequency ranges, and obtaining a fault diagnosis result; The process of selecting a plurality of target high time-frequency resolution time-frequency spectra from the plurality of two-dimensional time-frequency signals includes: Calculating estimated values ​​of each of the two-dimensional time-frequency signals respectively to obtain estimated values ​​corresponding to each of the two-dimensional time-frequency signals; Calculating the original high time-frequency resolution time spectra of each of the two-dimensional time-frequency signals and the estimated values ​​corresponding to each of the two-dimensional time-frequency signals respectively, to obtain the original high time-frequency resolution time spectra corresponding to each of the two-dimensional time-frequency signals; Calculating the envelope spectrum of each of the original high time-frequency resolution time-frequency signals using a Hilbert envelope demodulation algorithm to obtain an envelope spectrum corresponding to each of the two-dimensional time-frequency signals; Filtering out a maximum value from all the envelope spectra, and obtaining a maximum envelope spectrum after filtering; A difference between the maximum envelope spectrum and a first preset range value and a sum of the maximum envelope spectrum and a second preset range value are used as a screening range, and multiple target high time-frequency resolution time-frequency spectra are screened out from all the original high time-frequency resolution time-frequency spectra according to the screening range.

2. The fault diagnosis method according to claim 1, characterized in that: The process of performing Fourier transform analysis on the vibration acceleration signal to obtain multiple two-dimensional time-frequency signals includes: intercepting the vibration acceleration signal according to a preset signal length to obtain a signal sequence; The signal sequence is time-frequency-space expanded using a short-time Fourier transform algorithm to obtain multiple two-dimensional time-frequency signals.

3. The fault diagnosis method according to claim 1, characterized in that: The process of respectively calculating the estimated value of each of the two-dimensional time-frequency signals to obtain the estimated value corresponding to each of the two-dimensional time-frequency signals includes: The estimated values ​​of each of the two-dimensional time-frequency signals are calculated respectively by the first formula to obtain the estimated values ​​corresponding to each of the two-dimensional time-frequency signals. The first formula is: Among them, t m (t,ω) is the estimated value of the ωth frequency at the tth moment, t is time, ω is frequency, G(t,ω) is the two-dimensional time-frequency signal of the ωth frequency at the tth moment, |G(t,ω)| is the modulus of the two-dimensional time-frequency signal of the ωth frequency at the tth moment, and Δ is the preset local maximum frequency search range.

4. The fault diagnosis method according to claim 1, characterized in that: The process of respectively calculating the original high time-frequency resolution time spectra of each of the two-dimensional time-frequency signals and the estimated values ​​corresponding to each of the two-dimensional time-frequency signals to obtain the original high time-frequency resolution time spectra corresponding to each of the two-dimensional time-frequency signals includes: The original high time-frequency resolution time-frequency spectra corresponding to each of the two-dimensional time-frequency signals and the estimated values ​​corresponding to each of the two-dimensional time-frequency signals are calculated respectively by the second formula to obtain the original high time-frequency resolution time-frequency spectra corresponding to each of the two-dimensional time-frequency signals. The second formula is: Ts(t,ω)=G(t,ω)δ(ω-t m (t,ω)), Among them, Ts(t,ω) is the original high time-frequency resolution time spectrum of the ωth frequency at the tth time, ω is the frequency, G(t,ω) is the two-dimensional time-frequency signal of the ωth frequency at the tth time, t m (t,ω) is the estimated value of the ωth frequency at the tth time, and δ is the Dirac operation assigner.

5. The fault diagnosis method according to claim 1, characterized in that: The process of analyzing all target high time-frequency resolution time-frequency spectra to obtain envelope spectra includes: The reconstructed time series signals of all the target high time-frequency resolution time-frequency spectra are calculated by the third formula to obtain the reconstructed time series signals. The third formula is: Where s(t) is the reconstructed time series signal, g(0) is the function value of the window function at point 0, Ts(t,ω) is the original high time-frequency resolution time spectrum of the ωth frequency at the tth time, and e is the base of the natural logarithm. Extracting a signal from the reconstructed time series signal using a Hilbert transform algorithm to obtain an envelope signal; Performing Fourier transform on the envelope signal to obtain an envelope spectrum.

6. The fault diagnosis method according to claim 1, characterized in that: The process of performing fault diagnosis on the envelope spectrum according to all the fault characteristic frequency ranges and the fault types corresponding to all the fault characteristic frequency ranges to obtain a fault diagnosis result includes: Extracting the characteristic frequencies of the first N highest peaks from the envelope spectrum to obtain a plurality of characteristic frequencies to be verified; Verify whether the plurality of characteristic frequencies to be verified are all within any one of all the fault characteristic frequency ranges; if the verification is successful, take the fault type corresponding to the fault characteristic frequency range as the fault diagnosis result.

7. A fault diagnosis device, characterized in that: include: A Fourier transform analysis module is used to obtain a vibration acceleration signal of the high-speed rotating machinery to be diagnosed from a vibration acceleration sensor, and perform Fourier transform analysis on the vibration acceleration signal to obtain multiple two-dimensional time-frequency signals; A screening module, configured to screen out a plurality of target high time-frequency resolution time-frequency spectra from the plurality of two-dimensional time-frequency signals; An envelope spectrum analysis module is used to analyze all target high time-frequency resolution time-frequency spectra to obtain envelope spectra; a fault diagnosis result obtaining module, configured to import a plurality of fault characteristic frequency ranges and a fault type corresponding to each of the fault characteristic frequency ranges, perform fault diagnosis on the envelope spectrum according to all the fault characteristic frequency ranges and the fault types corresponding to all the fault characteristic frequency ranges, and obtain a fault diagnosis result; The screening module is specifically used for: Calculating estimated values ​​of each of the two-dimensional time-frequency signals respectively to obtain estimated values ​​corresponding to each of the two-dimensional time-frequency signals; Calculating the original high time-frequency resolution time spectra of each of the two-dimensional time-frequency signals and the estimated values ​​corresponding to each of the two-dimensional time-frequency signals respectively, to obtain the original high time-frequency resolution time spectra corresponding to each of the two-dimensional time-frequency signals; Calculating the envelope spectrum of each of the original high time-frequency resolution time-frequency signals using a Hilbert envelope demodulation algorithm to obtain an envelope spectrum corresponding to each of the two-dimensional time-frequency signals; Filtering out a maximum value from all the envelope spectra, and obtaining a maximum envelope spectrum after filtering; A difference between the maximum envelope spectrum and a first preset range value and a sum of the maximum envelope spectrum and a second preset range value are used as a screening range, and multiple target high time-frequency resolution time-frequency spectra are screened out from all the original high time-frequency resolution time-frequency spectra according to the screening range.

8. A fault diagnosis system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the fault diagnosis method according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the fault diagnosis method according to any one of claims 1 to 6 is implemented.

Images