Method for diagnosing mechanical rotating component acoustic emission faults based on bidirectional weighted cyclostationarity

Abstract

The application provides a mechanical rotating part acoustic emission fault diagnosis method based on bidirectional weighted cyclic stationarity, which comprises the following steps: collecting an acoustic emission signal to be diagnosed, obtaining a cyclic spectrum coherence function, and obtaining a baseline discrete spectrum coherence mapping and a diagnosis discrete spectrum coherence mapping; based on the baseline discrete spectrum coherence mapping and the diagnosis discrete spectrum coherence mapping, a spectrum frequency weight redistribution vector is obtained; based on the cyclic spectrum coherence function, a difference spectrum coherence matrix is obtained; and based on the spectrum frequency weight redistribution vector and the difference spectrum coherence matrix, a bidirectional reweighted difference spectrum coherence matrix is obtained. The application integrates healthy baseline data as prior information into a cyclic stationarity analysis framework, innovatively combines the difference and weighting ideas, deeply excavates the unique advantages and potential of the acoustic emission technology in the field of fault diagnosis, fully meets the needs of complex and variable industrial application scenarios, and has practical application value.

Description

Technical Field

[0001] This invention belongs to the field of rotating machinery fault diagnosis technology, and in particular, it is a method for diagnosing acoustic emission faults of rotating mechanical components based on bidirectional weighted cyclic stability. Background Technology

[0002] Acoustic emission technology, as an effective tool for monitoring the damage process of rotating machinery components, has demonstrated high sensitivity to the detection of early and subtle faults, providing an indispensable supplement to the current condition monitoring technology system.

[0003] While combining acoustic emission technology with health indicators enables real-time tracking of damage states, its analytical focus leans towards processing time-domain information, with insufficient in-depth mining and utilization of frequency-domain information. When rotating components fail, their acoustic emission responses typically exhibit significant impact and cyclic stationarity characteristics. Therefore, extracting signal components that correspond to these characteristics has become a core research objective. In existing acoustic emission signal spectrum analysis and diagnostic techniques, most methods borrow from vibration acceleration-based algorithm frameworks, with resonance demodulation technology being particularly prominent. This encompasses a series of blind filtering methods, such as 1 / 3 binary tree spectrum analysis, deconvolution processing, and signal decomposition. The core of these methods lies in accurately locating and extracting the target narrowband signal containing the maximum amount of fault information through optimization strategies, followed by envelope analysis to further highlight fault characteristics.

[0004] In practical applications, the accuracy and reliability of diagnostic results are often severely affected when the signal-to-noise ratio of the target frequency band is low or when other non-fault-related information is mixed in. Furthermore, the selection of filter parameters needs to be flexibly adjusted according to the specific application scenario, a process highly dependent on the knowledge and practical experience of domain experts. Another important method utilizes cyclostationary analysis tools, which can effectively detect the statistical periodicity characteristics of signals across the entire carrier frequency domain, thus avoiding the dilemma of frequency band selection. In the field of mechanical condition monitoring, research has successfully introduced cyclostationary technology into the detection of unknown targets, with applications including the monitoring of helicopters and underwater propellers. Cyclostationary algorithms, including spectral correlation analysis and spectral coherence analysis, provide a new perspective for signal processing by converting time-domain signals into a bispectral mapping form of cyclic frequency and carrier frequency.

[0005] However, current cyclic stationary analysis methods that integrate acoustic emission technology still face several challenges: First, it is difficult to simultaneously and effectively extract fault feature information distributed across multiple narrow frequency bands within a wider range of acoustic emission information frequency bands; second, the prior information upon which feature extraction relies is limited by fault features derived from dynamic models, and the noise suppression problem within the preferred frequency band has not been properly solved; finally, acoustic emission monitoring technology based on cyclic stationary analysis has previously focused mainly on single components such as rolling bearings, and reports on its application in monitoring other complex mechanical rotating parts are still scarce.

[0006] Therefore, given the wide frequency response range and rich information content of acoustic emission signals, it is urgent and necessary to seek a method for diagnosing acoustic emission faults of mechanical rotating parts based on bidirectional weighted cyclic stationary signals, so as to effectively extract and identify early weak fault features. Summary of the Invention

[0007] This invention addresses the shortcomings of existing technologies by proposing a bidirectional weighted cyclostationary method for diagnosing acoustic emission faults in rotating mechanical components. The method includes acquiring the acoustic emission signal to be diagnosed, obtaining the cyclic spectrum coherence function and obtaining a baseline discrete spectrum coherence map and a discrete spectrum coherence map to be diagnosed; based on the baseline discrete spectrum coherence map and the discrete spectrum coherence map to be diagnosed, obtaining a spectral frequency weight redistribution vector; based on the cyclic spectrum coherence function, obtaining a differential spectrum coherence matrix; and based on the spectral frequency weight redistribution vector and the differential spectrum coherence matrix, obtaining a bidirectional reweighted differential spectrum coherence matrix. This invention integrates healthy baseline data as prior information into the cyclostationary analysis framework, innovatively combining differential and weighted thinking, deeply exploring the unique advantages and potential of acoustic emission technology in the field of fault diagnosis, fully meeting the needs of complex and ever-changing industrial application scenarios, and possessing practical application value.

[0008] This invention provides a method for diagnosing acoustic emission faults in mechanical rotating parts based on bidirectional weighted cyclic stationarity, comprising the following steps:

[0009] S1. Acquire the acoustic emission signal to be diagnosed, obtain the cyclic spectrum coherence function, and obtain the baseline discrete spectrum coherence mapping and the discrete spectrum coherence mapping to be diagnosed;

[0010] S2. Based on the baseline discrete spectrum coherence mapping and the target discrete spectrum coherence mapping, obtain the spectral frequency weight redistribution vector; based on the maximum mean difference (MMD) metric, evaluate the target-baseline difference for each spectral frequency component in the cyclic spectral coherence function, and calculate the spectral frequency weight redistribution vector R:

[0011]

[0012] in, This represents the baseline discrete spectrum coherence mapping. f represents the discrete spectrum coherence mapping to be diagnosed. n Indicates a resolution of F s / N w The nth spectral component in the spectral frequency; F s N represents the sampling frequency; w The window length set in cyclic coherence spectrum estimation;

[0013] S3. Obtain the difference spectrum coherence matrix based on the cyclic spectrum coherence function;

[0014] S31. Calculate the differential demodulation spectrum: based on the narrowband demodulation spectrum γ of the baseline signal. H (α) and the narrowband demodulation spectrum γ of the signal to be diagnosed F (α), calculate the differential demodulation spectrum γ diff (α):

[0015]

[0016] Among them, ||·|| L1 The L1 norm of the matrix is ​​represented by c; the separating factor is represented by c.

[0017] S32. Determine the spectral ratio. If the spectral ratio is greater than 1, execute c = 0.9c and return to execute step S31 again; if the spectral ratio is less than 1, execute c = 1.1c and return to execute step S31 again; if the spectral ratio is equal to 1, execute step S33.

[0018] S33, Based on the differential demodulation spectrum γ diff (α), the cyclic spectrum coherence function γ of the acoustic emission signal x(t) to be diagnosed is separated. x Fault-inducing factors for each spectral frequency component in (α,f), obtaining the differential spectral coherence matrix:

[0019]

[0020] Where Max(·,·) represents taking the maximum value of the corresponding elements of the two vectors;

[0021] S4. Based on the spectral frequency weight redistribution vector and the differential spectral coherence matrix, obtain the bidirectional reweighted differential spectral coherence matrix: the spectral frequency weight redistribution vector R is used as the differential spectral coherence matrix CSCoh. diff The re-integration strategy yields the bidirectional reweighted differential spectral coherence matrix (DRD):

[0022] DRD=(R×(CSCoh diff ) T ) T (9)

[0023] in,(·) TThis represents the transpose of a matrix.

[0024] Furthermore, step S1 specifically includes the following steps:

[0025] S11. In continuous monitoring scenarios, use an acoustic emission monitoring system to obtain historical health baseline data of the machine. H After (t), the acoustic emission signal x(t) to be diagnosed is acquired;

[0026] S12. Calculate the time-varying autocorrelation function R of the acoustic emission signal x(t) to be diagnosed. xx (t,τ):

[0027]

[0028] Where t represents time; τ represents time delay; x represents the set average operator; * This indicates that the conjugate operation is performed on x;

[0029] S13. Calculate the cyclic spectrum correlation function S of the acoustic emission signal x(t) to be diagnosed. x (α,f), for the time-varying autocorrelation function R in formula (1) xx The bispectral representation of (t,τ) after a two-dimensional Fourier transform is specifically as follows:

[0030]

[0031] Where α represents the cyclic frequency; f represents the spectral frequency; j represents the imaginary number; and e represents the exponential function.

[0032] S14. Calculate the cyclic spectrum coherence function γ of the acoustic emission signal x(t) to be diagnosed. x (α,f), for the cyclic spectrum correlation function S described in formula (2) x (α,f) is represented in terms of energy normalization as follows:

[0033]

[0034] S15, Based on the cyclic spectrum coherence function γ x (α,f), combined with spectral estimation methods, obtain the baseline discrete spectral coherence mapping of the acoustic emission signal x(t) to be diagnosed. Coherence mapping with discrete spectrum of the diagnosis

[0035] Preferably, in step S2, a first probability distribution p and a second probability distribution q are set, assuming that the first group of signal data X follows X~p and the second group of signal data Y follows Y~q, and the maximum mean difference (MMD) is measured. Represented as:

[0036]

[0037] in, The n represents the set of functions in the sample space; f(·) denotes the mapping to the reproducing Hilbert space; p and n q These represent the number of samples in the first group of signal data X and the second group of signal data Y, respectively; x i and y j These represent sample data of the first group of signal data X and the second group of signal data Y, respectively.

[0038] Preferably, the Spectral Ratio in step S32 is defined as:

[0039]

[0040] Wherein, min(γ) diff (α)) and max(γ) diff (α) represents the differential demodulation spectrum γ. diff The minimum and maximum values ​​of (α).

[0041] Preferably, in step S13, the cyclic frequency α and the spectral frequency f correspond to the time t and the time delay τ before mapping, respectively, and the cyclic frequency α and the spectral frequency f indicate the modulation frequency and carrier frequency of the acoustic emission signal x(t) to be diagnosed, respectively.

[0042] Preferably, in step S31, negative values ​​in the differential demodulation spectrum represent basic components, and positive values ​​represent fault-causing factors.

[0043] Preferably, the acoustic emission signal to be diagnosed in step S11 undergoes a preprocessing operation, which includes high-frequency sampling, data truncation, and mean removal.

[0044] Compared with the prior art, the technical effects of the present invention are as follows:

[0045] 1. The present invention designs a method for diagnosing acoustic emission faults in mechanical rotating parts based on bidirectional weighted cyclostationarity. Based on the analysis of acoustic emission monitoring signal characteristics, it integrates healthy baseline data as prior information into the cyclostationarity analysis framework, effectively extracting fault feature components and improving the accuracy of fault diagnosis. It innovatively combines differential and weighted thinking, performing dual optimization on the cyclic spectrum coherence feature plane. On the one hand, it uses the maximum mean difference to measure the distribution difference between the healthy reference and the signal to be diagnosed, achieving precise reweighting in the spectral frequency direction. On the other hand, through differential processing along the cyclic frequency, it significantly enhances the diagnostic capability of the one-dimensional mapping after integration along the spectral frequency, further improving the sensitivity of fault identification.

[0046] 2. The acoustic emission fault diagnosis method for mechanical rotating parts based on bidirectional weighted cyclic stability designed in this invention can be verified using various acoustic emission monitoring signals of mechanical rotating parts to comprehensively evaluate the performance of the proposed method. For example, planetary gearbox fault cases and rolling bearing fault cases fully demonstrate the state monitoring and fault diagnosis capabilities of the proposed method in various acoustic emission monitoring signals of rotating parts. The proposed method deeply explores the unique advantages and potential of acoustic emission technology in the field of fault diagnosis, and realizes high-precision monitoring and fault diagnosis of the state of various mechanical rotating parts, fully meeting the needs of complex and ever-changing industrial application scenarios. Attached Figure Description

[0047] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings.

[0048] Figure 1 This is a flowchart of the acoustic emission fault diagnosis method for mechanical rotating parts based on bidirectional weighted cyclic stabilization of the present invention;

[0049] Figure 2 This is a schematic diagram of the structure of the test bench in a specific embodiment of the present invention;

[0050] Figure 3 This is the acoustic emission waveform stream of the planetary gearbox during healthy operation in the first specific embodiment of the present invention;

[0051] Figure 4 This is the acoustic emission waveform stream of the planetary gearbox under a weak fault in the planetary gears in the first specific embodiment of the present invention;

[0052] Figure 5 This is the spectral kurtosis ratio feature matrix of the fault signal in the first specific embodiment of the present invention;

[0053] Figure 6 According to the first specific embodiment of the present invention Figure 5 The optimal frequency band is used to obtain the envelope spectrum of the filtered signal;

[0054] Figure 7 This is the cyclic spectrum coherence matrix of the fault signal in the first specific embodiment of the present invention;

[0055] Figure 8 According to the first specific embodiment of the present invention Figure 7 The enhanced envelope spectrum obtained by applying a full-band integration strategy;

[0056] Figure 9 This is a histogram of the cyclic spectral coherence weight redistribution vector in the first specific embodiment of the present invention;

[0057] Figure 10According to the first specific embodiment of the present invention Figure 9 The envelope spectrum obtained after bidirectional reweighted difference operation;

[0058] Figure 11 The acoustic emission waveform stream of the harmonic reducer during healthy operation in the second specific embodiment of the present invention;

[0059] Figure 12 This is the acoustic emission waveform stream before the final failure of the harmonic reducer in the second specific embodiment of the present invention;

[0060] Figure 13 This is the spectral kurtosis ratio feature matrix of the fault signal in the second specific embodiment of the present invention;

[0061] Figure 14 According to the second specific embodiment of the present invention Figure 13 The optimal frequency band is used to obtain the envelope spectrum of the filtered signal;

[0062] Figure 15 This is the cyclic spectrum coherence matrix of the fault signal in the second specific embodiment of the present invention;

[0063] Figure 16 According to the second specific embodiment of the present invention Figure 15 The enhanced envelope spectrum obtained by applying a full-band integration strategy;

[0064] Figure 17 This is a histogram of the cyclic spectrum coherence weight redistribution vector in a second specific embodiment of the present invention;

[0065] Figure 18 According to the second specific embodiment of the present invention Figure 17 The envelope spectrum obtained after bidirectional reweighted difference operation. Detailed Implementation

[0066] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other. The present application will now be described in detail with reference to the accompanying drawings and embodiments.

[0067] Figure 1 This invention illustrates a method for diagnosing acoustic emission faults in rotating mechanical components based on bidirectional weighted cyclostationarity. This method fully exploits the difference in cyclostationarity between healthy and faulty states. The method includes the following steps:

[0068] S1. Acquire the acoustic emission signal to be diagnosed, obtain the cyclic spectrum coherence function, and obtain the baseline discrete spectrum coherence mapping and the discrete spectrum coherence mapping to be diagnosed.

[0069] S11. In continuous monitoring scenarios, use an acoustic emission monitoring system to obtain historical health operation baseline data of the machine. H After (t), the acoustic emission signal x(t) to be diagnosed is acquired. To ensure the accuracy and effectiveness of signal acquisition, the acoustic emission sensor should be placed in a flat position close to the rotating mechanical part under test, and fixed by adhesive or magnetic clamp. If magnetic clamp is used, high-vacuum silicone grease or other acoustic coupling agent should be applied to optimize signal transmission. Before formal testing, a lead breakage test is also required to verify the reliability and stability of the sensor connection. The acoustic emission signal to be diagnosed undergoes preprocessing, including high-frequency sampling, data truncation, and mean removal, to ensure the accuracy and efficiency of subsequent analysis.

[0070] S12. Calculate the time-varying autocorrelation function R of the acoustic emission signal x(t) to be diagnosed. xx (t,τ):

[0071]

[0072] Where t represents time; τ represents time delay; x represents the set average operator; * This indicates that the conjugate operation is performed on x.

[0073] S13. Calculate the cyclic spectrum correlation function S of the acoustic emission signal x(t) to be diagnosed. x (α,f), for the time-varying autocorrelation function R in formula (1) xx The bispectral representation of (t,τ) after a two-dimensional Fourier transform is specifically as follows:

[0074] S x (α,f)=∫∫R x (t,τ)e -j2π(αt+fτ) dtdτ (2)

[0075] Where α represents the cyclic frequency; f represents the spectral frequency; j represents the imaginary number; and e represents the exponential function.

[0076] The cyclic frequency α and the spectral frequency f correspond to the time t and time delay τ before mapping, respectively, and also indicate the modulation frequency and carrier frequency of the acoustic emission signal x(t) to be diagnosed.

[0077] S14. Calculate the cyclic spectrum coherence function γ of the acoustic emission signal x(t) to be diagnosed. x (α,f), for the cyclic spectrum correlation function S in formula (2) x(α,f) is represented in energy-normalized form as follows:

[0078]

[0079] S15, Based on the cyclic spectral coherence function γ x (α,f), combined with spectral estimation methods, obtain the baseline discrete spectral coherence mapping of the acoustic emission signal x(t) to be diagnosed. Coherent mapping with discrete spectrum of the diagnosis

[0080] S2. Obtain the weight redistribution vector based on the baseline discrete spectrum coherence mapping and the discrete spectrum coherence mapping to be diagnosed.

[0081] In the spectral frequency dimension of the spectral coherence matrix, this invention dynamically determines the spectral frequency weights by evaluating the deviation of each sub-band monitoring signal relative to the baseline signal, aiming to enhance the diagnostic accuracy and efficiency of the one-dimensional mapping obtained after integrating along the spectral frequency.

[0082] Maximum Mean Discrepancy (MMD), as a metric, effectively quantifies the distance between two statistical distributions in a reproducing kernel Hilbert space, and is particularly suitable for assessing the degree of difference between the sensor signal to be diagnosed and historical health data. Compared with other discrepancy quantification techniques, MMD exhibits superior discrepancy representation capabilities and numerical stability. Furthermore, by appropriately selecting the kernel function, the mapping strategy of the feature space can be flexibly adjusted and optimized.

[0083] Define a first probability distribution p and a second probability distribution q. Assume that the first set of signal data X follows the pattern X ~ p, and the second set of signal data Y follows the pattern Y ~ q. The maximum mean difference (MMD) is used as the metric. Represented as:

[0084]

[0085] in, The n represents the set of functions in the sample space; f(·) denotes the mapping to the reproducing Hilbert space; p and n q These represent the number of samples in the first group of signal data X and the second group of signal data Y, respectively; x i and y j These represent sample data from the first group of signal data X and the second group of signal data Y, respectively. Maximum Mean Difference (MMD) metric. The larger the value, the greater the distance between the first set of signal data X and the second set of signal data Y.

[0086] Based on the maximum mean difference (MMD) metric, the patient-baseline difference for each spectral frequency component in the cyclic spectral coherence function is assessed, and the spectral frequency weight redistribution vector R is calculated.

[0087]

[0088] in, This represents the baseline discrete spectrum coherence mapping. f represents the discrete spectrum coherence mapping to be diagnosed. n Indicates a resolution of F s / N w The nth spectral component in the spectral frequency; F s N represents the sampling frequency; w The window length set in the cyclic coherence spectrum estimation.

[0089] S3. Obtain the difference spectrum coherence matrix based on the cyclic spectrum coherence function.

[0090] In the cyclic axis direction of the spectral coherence matrix, residual signals based on a health reference are used to characterize the cyclic components introduced by the fault.

[0091] S31. Calculate the differential demodulation spectrum: based on the narrowband demodulation spectrum γ of the baseline signal. H (α) and the narrowband demodulation spectrum γ of the signal to be diagnosed F (α), calculate the differential demodulation spectrum γ diff (α):

[0092]

[0093] Among them, ||·|| L1 represents the L1 norm of the matrix; c represents the separation factor, and adjusting its value can achieve positive and negative separation of fault-causing factors and basic components.

[0094] S32. Determine the spectral ratio. If the spectral ratio is greater than 1, execute c = 0.9c and return to execute step S31 again. If the spectral ratio is less than 1, execute c = 1.1c and return to execute step S31 again. If the spectral ratio is equal to 1, execute step S33.

[0095] The Spectral Ratio is defined as follows:

[0096]

[0097] Wherein, min(γ) diff (α)) and max(γ) diff (α) represents the differential demodulation spectrum γ. diff The minimum and maximum values ​​of (α).

[0098] The differential demodulation spectrum calculated by the above algorithm ensures that the absolute amplitudes of the fundamental components and fault-inducing factors are approximately equal. Negative values ​​in the spectrum represent the fundamental components, and positive values ​​represent the fault-inducing factors.

[0099] S33, Based on differential demodulation spectral modulation γ diff (α), the cyclic spectrum coherence function γ of the acoustic emission signal x(t) to be diagnosed is separated. x Fault-inducing factors for each spectral frequency component in (α,f), obtaining the differential spectral coherence matrix:

[0100]

[0101] Max(·,·) represents taking the maximum value of the corresponding elements of the two vectors, which is intended to remove the basic components in the differential demodulation spectrum.

[0102] S4. Based on the spectral frequency weight redistribution vector and the differential spectral coherence matrix, obtain the bidirectional reweighted differential spectral coherence matrix.

[0103] The spectral frequency weight redistribution vector R obtained in step S2 represents the frequency band rich in fault-introduced information.

[0104] The spectral frequency weight redistribution vector R is used as the differential spectral coherence matrix CSCoh. diff The re-integration strategy yields the bidirectional reweighted differential spectral coherence matrix (DRD):

[0105] DRD=(R×(CSCoh diff ) T ) T (9)

[0106] in,(·) T This represents the transpose of a matrix.

[0107] Acoustic emission technology features a wideband response, with sampling frequencies typically set at the MHz level. On one hand, the spectral frequency weighting redistribution vector ensures the effective synchronous extraction of fault information components across multiple frequency bands. On the other hand, differential demodulation spectrum retention preserves fault-inducing components in the acoustic emission signal and suppresses fundamental components that persist throughout the monitoring process. Positive values ​​represent fault-inducing components, while zero values ​​represent fundamental components, i.e., inherent components. Fault-inducing components refer to signal components introduced by the degradation of mechanical rotating parts, while inherent components refer to signal components generated by mechanical rotating parts in a healthy state.

[0108] S5. Based on the bidirectional reweighted differential spectral coherence matrix, fault feature components of mechanical rotating parts are obtained, and faults of mechanical rotating parts are analyzed. Specifically, for the bidirectional reweighted differential spectral coherence matrix, the spectral frequency direction is integrated to form the corresponding envelope spectrum. Fault feature components are searched in the enhanced envelope spectrum to determine the faults of mechanical rotating parts. This method can accurately identify various types of damage, such as tooth wear, tooth cracks, and bearing inner and outer ring faults.

[0109] In one specific embodiment, the method proposed in this invention is applied to a planetary gearbox fault diagnosis embodiment for verification.

[0110] Take a test bench for a certain mechanical transmission system as an example. Figure 2 As shown, the test bench includes a drive motor 1, several couplings 2, a speed and torque measurement module 3, a parallel shaft gearbox 4, an input shaft support module 5, an input shaft speed measurement module 6, a planetary gearbox 7, an acoustic emission sensor 8, an output shaft speed measurement module 9, an output shaft support module 10, and a magnetic powder brake 11. The drive motor 1 has a rated voltage of 380V, a rated power of 3kW, a rated speed of 3000rpm, and a rated torque of 9.55N·m. The couplings 2, input shaft support module 5, and output shaft support module 10 play a crucial connecting and supporting role in the test bench. The speed and torque measurement module 3 is configured to monitor the actual speed and torque values ​​of the drive motor 1. The parallel shaft gearbox 4 serves as the first-stage reduction device in the transmission system, and the planetary gearbox 7 is the second-stage reduction device. The input shaft speed measurement module 6 and the output shaft speed measurement module 9 are used to accurately measure the actual input and output speeds of the planetary gearbox 7, respectively. Furthermore, the acoustic emission sensor 8 is used to collect acoustic emission signals during the test process for fault diagnosis. The magnetic powder brake 11 provides the necessary output load for the entire transmission system.

[0111] In one specific embodiment, the parallel shaft gearbox 4 has a reduction ratio of 1:3.9, the planetary gearbox 7 has a reduction ratio of 1:5, the sun gear has 21 teeth, the planet gears have 31 teeth, and the external ring gear has 84 teeth. The power flow direction is with the sun gear shaft as the input and the planet carrier shaft as the output. A weak fault is implanted into the planet gear tooth surface to simulate its degradation process. The acoustic emission sampling frequency is set to 2MHz, and the preamplifier is selected as 40dB. According to the kinematic model of this mechanical transmission system, the characteristic order of the planetary gear fault is calculated to be 0.139. In the experiment, the motor input speed is set to 3000rpm, the load level is maintained at about 20%, corresponding to a planetary gearbox input speed of 12.82Hz, and the characteristic frequency of the planetary gear fault is 6.95Hz. The acquired acoustic emission waveform of the planetary gearbox during healthy operation is shown below. Figure 3 As shown, the acoustic emission waveform during a minor fault in the planetary gear is as follows: Figure 4 As shown.

[0112] To demonstrate the superiority of the proposed method, a traditional 1 / 3-binary tree strategy is first used to determine the frequency partitioning strategy of different layers of the signal to form a feature matrix. Then, the ratio of the feature matrices of the fault signal and the healthy signal is calculated to form a spectral kurtosis ratio map, highlighting the kurtosis difference between the signal to be diagnosed and the baseline signal. The spectral kurtosis ratio map feature matrix is ​​shown below. Figure 5 As shown, the preferred frequency band has a center frequency of 812,500 Hz and a bandwidth of 41,667 Hz. However, the spectral kurtosis ratio map has limited performance in extracting fault features in this embodiment, based on its guided envelope spectrum. Figure 6 The fault characteristics were almost impossible to identify, and the fault was not detected at the characteristic frequency f corresponding to the planetary gear fault. p The frequency peaks of the fault signal and its harmonics have extremely low signal-to-noise ratios, making it difficult to directly diagnose the fault type of the equipment. Next, the cyclic spectrum coherence matrix of the fault signal is calculated, as follows: Figure 7 As shown, an enhanced envelope spectrum is obtained by applying an integration strategy across its entire frequency band, as follows. Figure 8 As shown. Although the enhanced envelope spectrum contains some fault information, the characteristic frequency f corresponding to the planetary gear fault is... p The frequency peaks of its harmonics are relatively inconspicuous, and directly identifying this weak fault remains challenging in the absence of prior fault characteristic frequencies. Finally, to more effectively identify weak faults, the bidirectional reweighted cyclostationary analysis method proposed in this invention is used. Figure 9 A histogram of the weight redistribution vector used for bidirectional reweighted difference operation is presented. Based on this, the envelope spectrum guided by this method is obtained through envelope analysis. Figure 10 Although the envelope spectrum appears complex due to the weakness of the fault, the five spectral lines with the highest identified peaks successfully contained the fault characteristic frequencies and their harmonic components, verifying the effectiveness of the proposed method in weak fault extraction.

[0113] In another specific embodiment, the harmonic reducer fails after long-term use, exhibiting three failure modes: gear tooth wear, wear of the inner wall of the flex wheel, and fatigue cracking of the flexible thin-walled bearing in the wave generator. The acoustic emission signals collected during component degradation are analyzed, and the acoustic emission waveforms of the harmonic reducer during healthy operation are as follows: Figure 11 As shown, the acoustic emission waveform before final failure is as follows: Figure 12 As shown. First, the spectral kurtosis ratio plot is used to guide the signal processing process, and its feature matrix is ​​as follows. Figure 13 As shown, the preferred frequency band has a center frequency of 104167Hz and a bandwidth of 41667Hz. However, the spectral kurtosis ratio map performs limitedly in extracting fault features in this embodiment, although it guides the extraction of envelope spectrum. Figure 14 In the process, the 2x rotational frequency f caused by gear tooth wear and wear of the inner wall of the flexible gear rIts harmonics can be clearly shown, but the fault characteristic frequency f corresponding to the outer ring of the flexible thin-walled bearing in the wave generator is... or The spectral lines at that location are quite weak, making it difficult to conclude that a fault has occurred. Next, this method calculates the cyclic spectral coherence matrix of the fault signal, as shown below. Figure 15 As shown, an enhanced envelope spectrum is obtained by applying an integration strategy across its entire frequency band, as follows. Figure 16 As shown, its diagnostic effect is similar to that of the method based on spectral kurtosis ratio, but the extraction of fault features corresponding to the outer ring of the flexible thin-walled bearing in the signal is still not ideal. Figure 17 A histogram of the weight redistribution vector used for bidirectional reweighted difference operation is presented. Based on this, the envelope spectrum guided by this method is obtained through envelope analysis. Figure 18 It can not only identify the 2x rotational frequency f caused by two uniform wear phenomena r Furthermore, it can identify the fault characteristic frequency f of the outer ring of flexible thin-walled bearings. or It can accurately identify all three types of faults, ensuring comprehensive capture and accurate identification of fault information.

[0114] This invention presents a method for diagnosing acoustic emission faults in rotating mechanical components based on bidirectional weighted cyclostationarity. By analyzing the characteristics of acoustic emission monitoring signals, it integrates healthy baseline data as prior information into a cyclostationar analysis framework, effectively extracting fault feature components and improving the accuracy of fault diagnosis. It innovatively combines differential and weighted approaches, performing dual optimization on the cyclospectral coherence feature plane. On one hand, it utilizes the maximum mean difference to measure the distribution difference between the healthy reference and the signal to be diagnosed, achieving precise reweighting in the spectral frequency direction. On the other hand, through differential processing at the cyclospectral frequency, it significantly enhances the diagnostic capability of the one-dimensional mapping after integration along the spectral frequency, further improving the sensitivity of fault identification. To comprehensively evaluate the performance of the proposed method, it can be verified using various acoustic emission monitoring signals from rotating mechanical components, such as planetary gearbox fault cases and rolling bearing fault cases, fully demonstrating the proposed method's ability to monitor the state and diagnose faults in various types of rotating component acoustic emission monitoring signals. The proposed method deeply explores the unique advantages and potential of acoustic emission technology in the field of fault diagnosis, achieving high-precision monitoring and fault diagnosis of the state of various rotating mechanical components, fully meeting the needs of complex and ever-changing industrial application scenarios.

[0115] Finally, it should be noted that the above embodiments are for illustration only and not for limiting the technical solutions of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the present invention without departing from the spirit and scope of the present invention. Any modifications or partial substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for diagnosing acoustic emission faults in mechanical rotating parts based on bidirectional weighted cyclic stationarity, characterized in that, It includes the following steps: S1. In the continuous monitoring scenario, after obtaining the historical health operation baseline data of the machine using the acoustic emission monitoring system, the acoustic emission signal to be diagnosed of the mechanical rotating parts is collected, the cyclic spectrum coherence function is obtained, and the baseline discrete spectrum coherence mapping and the discrete spectrum coherence mapping of the acoustic emission signal to be diagnosed are obtained. S2. Based on the baseline discrete spectrum coherence mapping and the target discrete spectrum coherence mapping, obtain the spectral frequency weight redistribution vector; based on the maximum mean difference (MMD) metric, evaluate the target-baseline difference for each spectral frequency component in the cyclic spectral coherence function, and calculate the spectral frequency weight redistribution vector of the target acoustic emission signal. : （5）； in, This represents the baseline discrete spectrum coherence mapping. This represents the discrete spectrum coherence mapping to be diagnosed. Indicates resolution as The first frequency in the spectrum Each spectral component; Indicates the sampling frequency; The window length set in cyclic coherence spectrum estimation; S3. Based on the cyclic spectrum coherence function, obtain the differential spectrum coherence matrix of the acoustic emission signal to be diagnosed; S31. Calculate the differential demodulation spectrum: Narrowband demodulation spectrum based on the baseline signal. Narrowband demodulation spectrum of the signal to be diagnosed Calculate differential detuning spectrum : （6）； in, Representing a matrix Norm; c represents the separating factor; S32. Determine the spectral ratio. If the spectral ratio is greater than 1, then execute... Then return to re-execute step S31; if the spectral ratio is less than 1, then execute... Then return to re-execute step S31; if the spectral ratio is equal to 1, then execute step S33; S33, Based on the differential resolution spectrum The acoustic emission signal to be diagnosed is separated. Cyclic spectral coherence function By identifying the fault-inducing components of each spectral frequency component, the differential spectral coherence matrix of the acoustic emission signal to be diagnosed is obtained. （8）； in, This indicates taking the maximum value of the corresponding elements of the two vectors; the positive value of the difference detuning spectrum represents the fault-causing component of the mechanical rotating parts, and the zero value represents the inherent component of the mechanical rotating parts. S4. Based on the spectral frequency weight redistribution vector and the differential spectral coherence matrix, obtain the bidirectional reweighted differential spectral coherence matrix: based on the spectral frequency weight redistribution vector As the difference spectrum coherence matrix The re-integration strategy yields the bidirectional reweighted differential spectral coherence matrix. : （9）； in, Represents the transpose of a matrix; S5. Based on the bidirectional reweighted differential spectrum coherence matrix, the fault characteristic components of mechanical rotating parts are obtained, and the faults of mechanical rotating parts are analyzed.

2. The method for diagnosing acoustic emission faults of mechanical rotating parts based on bidirectional weighted cyclic stationarity as described in claim 1, characterized in that, Step S1 specifically includes the following steps: S11. In continuous monitoring scenarios, use an acoustic emission monitoring system to obtain historical health operation baseline data of the machine. Then, the acoustic emission signal to be diagnosed was collected. ; S12. Calculate the acoustic emission signal to be diagnosed. Time-varying autocorrelation function : （1）； in, Indicates time; Indicates time delay; Represents the set averaging operator; Indicates to Perform conjugate operations; S13. Calculate the acoustic emission signal to be diagnosed. Cyclic spectrum correlation function Regarding the time-varying autocorrelation function described in formula (1) The bispectral representation after performing a two-dimensional Fourier transform is specifically shown as follows: （2）； in, Indicates the cycle frequency; Indicates spectral frequency; represents an imaginary number; Represents an exponential function; S14. Calculate the acoustic emission signal to be diagnosed. Cyclic spectral coherence function Regarding the cyclic spectrum correlation function described in formula (2) The energy is normalized and specifically represented as follows: （3）； S15, Based on the cyclic spectrum coherence function By combining spectral estimation methods, the acoustic emission signal to be diagnosed is obtained. Baseline discrete spectrum coherence mapping Coherent mapping with discrete spectrum of the diagnosis .

3. The method for diagnosing acoustic emission faults of mechanical rotating parts based on bidirectional weighted cyclic stabilization as described in claim 1, characterized in that, In step S2, the first probability distribution is set. Second probability distribution Assuming the first set of signal data obey The second set of signal data obey The maximum mean difference (MMD) metric Represented as: （4）； in, Represents the set of functions in the sample space; This indicates mapping to the regenerated Hilbert space; and These represent the first group of signal data. Second group of signal data The number of samples; and These represent the first group of signal data. Second group of signal data Sample data.

4. The method for diagnosing acoustic emission faults of mechanical rotating parts based on bidirectional weighted cyclic stabilization as described in claim 1, characterized in that, The spectral ratio mentioned in step S32 Defined as: （7）； in, and They represent the differential resolution spectrum respectively. The minimum and maximum values.

5. The method for diagnosing acoustic emission faults of mechanical rotating parts based on bidirectional weighted cyclic stabilization according to claim 2, characterized in that, Cycle frequency in step S13 Spectral frequencies Corresponding to the time before mapping and latency And the cycle frequency Spectral frequencies Indicate the acoustic emission signal to be diagnosed respectively The modulation frequency and carrier frequency.

6. The method for diagnosing acoustic emission faults of mechanical rotating parts based on bidirectional weighted cyclic stabilization according to claim 1, characterized in that, In step S31, negative values ​​in the differential demodulation spectrum represent basic components, while positive values ​​represent fault-causing factors.

7. The method for diagnosing acoustic emission faults of mechanical rotating parts based on bidirectional weighted cyclic stationarity as described in claim 2, characterized in that, The acoustic emission signal to be diagnosed in step S11 undergoes preprocessing operations, including high-frequency sampling, data truncation, and mean removal.

Images