MCKD-enhanced under-damping stability-changing situation stochastic resonance fault diagnosis method

Abstract

The invention belongs to the technical field of fault diagnosis, and particularly relates to an MCKD enhanced under-damping stability-changing situation stochastic resonance fault diagnosis method, which comprises the following steps: S1, preprocessing an original signal by using a parameter adaptive MCKD method; s2, performing small parameterization adjustment on the signal after deconvolution; s3, performing feature enhancement processing on the parameterized signal by adopting a parameter self-adaptive under-damping variable-stability situation stochastic resonance system; and S4, analyzing frequency components with prominent amplitudes in the packet spectrum of the processed signal, and determining an equipment fault state. According to the MCKD enhanced under-damping stability changing situation stochastic resonance fault diagnosis method, weak fault features in a rolling bearing can be effectively enhanced, accurate extraction of the fault features is realized, and the accuracy and reliability of fault diagnosis are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of fault diagnosis, and particularly relates to a fault diagnosis method for underdamped variable stability state stochastic resonance enhanced by MCKD. Background Technique

[0002] The operating environment of equipment is complex and the operating conditions are variable. The mechanical system needs to bear irregular variable load effects and is extremely vulnerable to damage. Analyzing and processing the vibration signals of complex mechanical systems is the main means of equipment condition monitoring and fault diagnosis. However, in the actual engineering environment, the collected signals are often interfered by strong noise and the transmission path, making it difficult to detect fault features, easily leading to problems such as missed diagnosis and misjudgment, and seriously affecting the reliability and operating safety of the equipment.

[0003] In response to this problem, many scholars have conducted extensive research on extracting fault features from complex vibration signals, mainly focusing on the following two types of methods: One type of method is to suppress noise to highlight fault features. These methods include wavelet analysis, empirical mode decomposition, variational mode decomposition, and variable multi-scale morphological filtering, etc., which have important applications in fault diagnosis. In addition, kurtosis analysis and deconvolution methods are also widely used to extract periodic fault pulse signals. However, due to the complexity of the working environment, the influence of the signal transmission path, and the mutual coupling of multiple vibration sources, it is difficult to accurately extract periodic impact features. Another type of method is to use stochastic resonance to enhance weak target signals by using noise, so as to extract fault features. The stochastic resonance method shows good results in improving the signal-to-noise ratio, but in most of the existing studies, the input signals are not fully preprocessed, resulting in the aliasing of noise and useful signals not being effectively processed, which limits the diagnostic effect and robustness. Summary of the Invention

[0004] The purpose of the present invention is to provide a fault diagnosis method for underdamped variable stability state stochastic resonance enhanced by MCKD to achieve accurate diagnosis of equipment components under strong noise interference and provide a basic support for the fault prediction and health management of mechanical equipment.

[0005] The technical solution adopted by the present invention to solve its technical problems is: A fault diagnosis method for underdamped variable stability state stochastic resonance enhanced by MCKD, comprising the following steps:

[0006] S1. Preprocess the original signal by using the parameter adaptive MCKD method:

[0007] S2. Perform small parameter adjustment on the signal after deconvolution reconstruction;

[0008] S3. Perform feature enhancement processing on the parameterized signal by using a parameter adaptive underdamped variable stability state stochastic resonance system;

[0009] S4. Analyze the frequency components with prominent amplitudes in the spectrogram to achieve fault diagnosis.

[0010] Preferably, in step S1, the following steps are included:

[0011] S11: Determine the signal y to be analyzed with a length of N i and its sampling frequency Fs and the fault characteristic frequency f;

[0012] S12: Set the optimization intervals for the three parameters M, L, and T in the maximum correlated kurtosis deconvolution (MCKD) method, where M ∈ [1, 5], L ∈ [100, 300], T ∈ [0.8 * F s / f, 1.2 * F s / f];

[0013] S13: Use the reciprocal E of the square of the peak factor of the deconvolved signal as the fitness function, and utilize the adaptive simulated annealing particle swarm optimization algorithm to adaptively search for the optimal combination of parameters M, L, and T to obtain the best deconvolved output signal, thus realizing the preprocessing of the signal. Among them, the definition of the fitness function E is as follows:

[0014]

[0015] Among them, E c is the peak factor

[0016]

[0017] Among them, the smaller the fitness function E, the stronger the periodic impact characteristic and the more obvious the fault characteristics.

[0018] Furthermore, in step S2, the Hilbert single-sideband modulation and variable-scale improved system model are adopted to exchange the frequency information of the characteristic signal and the small-parameter frequency information, and the signal-to-noise ratio SNR of the output signal is calculated using the fourth-order Runge-Kutta equation out :

[0019]

[0020] Among them, A d is the amplitude of the driving frequency, and A n is the amplitude of the strongest interference frequency.

[0021] Furthermore, in step S3, the following steps are included:

[0022] S31: Set the optimization intervals for the system parameters a, b, R, and the damping factor γ of stochastic resonance;

[0023] S32: Take the maximum SNR of the output signal out as the fitness function, and use the adaptive simulated annealing particle swarm optimization algorithm to perform parameter optimization;

[0024] S33: Input and update the optimal parameter combination into the parameter adaptive underdamped variable stability stochastic resonance system, and process the signal processed in the previous step.

[0025] Furthermore, in step S4, analyze the time-domain waveform and envelope spectrum of the final signal, and check whether there is a frequency peak at the fault characteristic frequency of the envelope spectrum, so as to determine the health state of the device.

[0026] The beneficial effect of the present invention is that it proposes an MCKD-enhanced underdamped variable stability stochastic resonance fault diagnosis method, which can effectively remove the background noise and transmission path interference in the signal, highlight the periodic fault impact characteristics, thereby significantly improving the accuracy and robustness of fault diagnosis, and has important engineering application value in the state monitoring and fault diagnosis of complex mechanical systems. Brief Description of the Drawings

[0027] The present invention will be further described below with reference to the drawings and embodiments.

[0028] Figure 1 is the flowchart of the present invention;

[0029] Figure 2 and Figure 3 are respectively the time-domain diagram and envelope spectrum diagram of the embodiment of the present invention;

[0030] Figure 4 and Figure 5 are respectively the time-domain diagram and envelope spectrum diagram of the signal processed by the present invention. Detailed Embodiment

[0031] Now the present invention will be further described in detail with reference to the drawings. These drawings are all simplified schematic diagrams, only illustrating the basic structure of the present invention in a schematic manner, so they only show the components related to the present invention.

[0032] As Figures 1 to 5 shown, it is an embodiment of the present invention, an MCKD-enhanced underdamped variable stability stochastic resonance fault diagnosis method, and the specific steps are as follows:

[0033] S1. Preprocess the original signal by using the parameter adaptive MCKD method;

[0034] As Figure 1 shown is the flowchart of the present invention. Determine the signal y to be analyzed with a length of N iThe sampling frequency Fs and the fault characteristic frequency f; set the optimization intervals of the three parameters M, L, and T in the maximum correlated kurtosis deconvolution (MCKD) method, where M ∈ [1, 5], L ∈ [100, 300], T ∈ [0.8 * F s / f, 1.2 * F s / f]; use the reciprocal E of the square of the peak factor of the deconvolved signal as the fitness function, and use the adaptive simulated annealing particle swarm algorithm to adaptively search for the optimal combination of parameters M, L, and T to obtain the best deconvolved output signal and achieve signal preprocessing. The definition of the fitness function E is as follows:

[0035]

[0036] where E c is the peak factor

[0037]

[0038] The smaller the fitness function E, the stronger the periodic impact characteristic and the more obvious the fault characteristics.

[0039] In this embodiment, the signal length N is 16000, the sampling frequency Fs is 12800 Hz, and the inner race fault characteristic frequency f is 28.42 Hz. The time domain diagram of the embodiment is as Figure 2 shown, and the envelope diagram is as Figure 3 shown. No clear and unique spectral peak can be found in both the time domain and the envelope spectrum, and the frequency components are very complex. The amplitudes at 203.1 Hz and 1424 Hz are significantly higher than those at other frequencies. The optimization intervals of the three parameters M, L, and T in the maximum correlated kurtosis deconvolution method of this embodiment are M ∈ [1, 5], L ∈ [100, 300], T ∈ [360.3, 540.46]

[0040] S2. Perform small parameter adjustment on the signal after deconvolution reconstruction;

[0041] Adopt Hilbert single-sideband modulation and variable scale improved system model to exchange the frequency information of the characteristic signal and the small parameter frequency information, and use the fourth-order Runge-Kutta equation to calculate the signal-to-noise ratio SNR out :

[0042]

[0043] where A d is the amplitude of the driving frequency, and A n is the amplitude of the strongest interference frequency.

[0044] S3. Use a parameter - adaptive under - damped variable - stability stochastic resonance system to perform feature enhancement processing on the parameterized signal;

[0045] Set the optimization intervals of the system parameters a, b, R and the damping factor γ of stochastic resonance; take the maximum SNR of the output signal out as the fitness function, and use the adaptive simulated annealing particle swarm optimization algorithm to perform parameter optimization; input and update the optimal parameter combination into the parameter - adaptive under - damped variable - stability stochastic resonance system to process the signal processed in the previous step.

[0046] S4. Analyze the frequency components with prominent amplitudes in the envelope spectrum of the processed signal to determine the equipment fault status;

[0047] Analyze the time - domain waveform and envelope spectrum of the final signal to find out whether there is a frequency peak at the fault characteristic frequency, so as to determine whether it is healthy or not.

[0048] The time - domain signal processed by the method of the present invention in the embodiment is as Figure 4 shown, and its corresponding envelope spectrum is as Figure 5 shown. A clear single - harmonic signal can be found from the time - domain diagram. In the envelope spectrum, a very clear and unique peak can be seen at a frequency of 27 Hz, and this frequency is close to the inner - race fault characteristic frequency, so it is determined that the equipment has an inner - race bearing fault.

[0049] It can be seen from the embodiment that the MCKD - enhanced under - damped variable - stability stochastic resonance fault diagnosis method proposed by the method of the present invention can enhance the weak fault characteristics in the original signal, realize the accurate diagnosis of faults, which fully illustrates the important significance and application value of the method of the present invention.

[0050] Taking the above - mentioned ideal embodiment based on the present invention as an inspiration, through the above - mentioned description, relevant staff can completely make various changes and modifications within the scope of not deviating from the technical idea of this invention. The technical scope of this invention is not limited to the content in the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. A MCKD enhanced underdamped variable steady state stochastic resonance fault diagnosis method, characterized in that: The following steps are involved: S1. Preprocess the original signal using the parameter adaptive MCKD method: S2, make small parameter adjustments to the deconvolution reconstructed signal; S3, using a parameter-adaptive underdamped variable-stable state stochastic resonance system to perform feature enhancement processing on the parameterized signal; S4. Analyze the frequency components with prominent amplitudes in the packet spectrum of the processed signal to determine the fault status of the equipment.

2. The MCKD enhanced underdamped variable steady state stochastic resonance fault diagnosis method according to claim 1, characterized in that: In step S1, the following steps are included: S11: Determine the signal y to be analyzed with a length of N i The sampling frequency Fs and fault characteristic frequency f; S12: Set the optimization intervals of the three parameters M, L and T in the maximum correlation kurtosis deconvolution (MCKD) method, and the number of shifts M ∈ [1, 5], filter length L ∈ [100, 300], deconvolution period T ∈ [0.8*F s / f, 1.2*F s / f]; S13: Taking the inverse of the square of the peak factor of the deconvolution signal E as the fitness function, the adaptive simulated annealing particle swarm algorithm is used to adaptively search for the optimal combination of parameters M, L and T to obtain the best deconvolution output signal, thus realizing the preprocessing of the signal. The fitness function E is defined as follows: Among them, E c is the crest factor Among them, the smaller the fitness function E is, the stronger the periodic impact characteristics are and the more obvious the fault characteristics are.

3. The MCKD enhanced underdamped variable steady state stochastic resonance fault diagnosis method according to claim 2, characterized in that: In step S2, Hilbert single-sideband modulation and variable-scaling improved system model are used to exchange characteristic signal frequency information with small parameter frequency information, and the signal-to-noise ratio (SNR) of the output signal is calculated using the fourth-order Runge-Kutta equation. out : Among them, A d is the amplitude of the driving frequency, A n is the amplitude of the strongest interference frequency.

4. The MCKD enhanced underdamped variable steady state stochastic resonance fault diagnosis method according to claim 3, characterized in that: In step S3, the following steps are included: S31: setting the optimization interval of the system parameters a, b, R and damping factor γ of stochastic resonance; S32: Maximum SNR of output signal out is the fitness function, and the adaptive simulated annealing particle swarm algorithm is used to optimize the parameters; S33: Input and update the optimal parameter combination into the parameter adaptive underdamped variable stable state stochastic resonance system, and process the signal processed in the previous step.

5. The MCKD enhanced underdamped variable steady state stochastic resonance fault diagnosis method according to claim 2, characterized in that: In step S4, the time domain waveform and envelope spectrum of the final signal are analyzed to check whether there is a frequency peak at the fault characteristic frequency of the envelope spectrum, so as to determine whether the equipment is in a healthy state.

Images