Fault diagnosis method and device, product, equipment and medium

Abstract

The invention discloses a fault diagnosis method and device, a product, equipment and a medium. The method comprises the following steps: acquiring a fault signal of a target wind power gear box; determining an optimal parameter of signal resonance sparse decomposition based on a first swarm intelligence optimization algorithm, and obtaining a target high resonance component and a target low resonance component obtained by performing signal resonance sparse decomposition on the target wind power gear box fault signal based on the optimal parameter of signal resonance sparse decomposition; determining an optimal parameter of blind deconvolution based on a second swarm intelligence optimization algorithm, and obtaining the optimal parameter based on the blind deconvolution to perform blind deconvolution processing on the target high-resonance component and the target low-resonance component to obtain a high-resonance component after target blind deconvolution processing and a low-resonance component after target blind deconvolution processing; and performing envelope demodulation on the high-resonance component after target blind deconvolution processing and the low-resonance component after target blind deconvolution processing so as to determine the fault type of the wind power gear box. Therefore, the fault diagnosis accuracy of the wind power gear box can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of fault diagnosis, and in particular to a fault diagnosis method, device, product, equipment and medium. Background Art

[0002] Gearbox is a key component in wind turbines. Once a failure occurs, it will cause the wind turbine to shut down, resulting in serious economic losses. As the service life of wind power equipment increases, it often faces multiple component failures. These failures will introduce multiple excitation sources, making the original detection signal show the characteristics of multi-vibration source coupling. Therefore, it is crucial to study the separation method of wind turbine gearbox composite faults and accurately extract the fault characteristics and diagnosis of gearboxes for mechanical equipment operation and maintenance.

[0003] Therefore, how to improve the fault diagnosis accuracy of wind turbine gearboxes is a technical problem that technical personnel in this field need to solve. Summary of the invention

[0004] In view of this, the purpose of this application is to provide a fault diagnosis method, device, product, equipment and medium, which can improve the fault diagnosis accuracy of wind turbine gearboxes. The specific scheme is as follows:

[0005] In a first aspect, the present application discloses a fault diagnosis method, comprising:

[0006] Obtaining a target wind turbine gearbox fault signal;

[0007] Determine the optimal parameters of signal resonance sparse decomposition based on the first group intelligent optimization algorithm, and obtain the target high resonance component and target low resonance component obtained by performing signal resonance sparse decomposition on the target wind turbine gearbox fault signal based on the optimal parameters of signal resonance sparse decomposition;

[0008] Determine the optimal parameters of blind deconvolution based on the second group intelligent optimization algorithm, and obtain the optimal parameters based on blind deconvolution to perform blind deconvolution processing on the target high resonance component and the target low resonance component, to obtain the target high resonance component after blind deconvolution processing and the target low resonance component after blind deconvolution processing;

[0009] Envelope demodulation is performed on the target high resonance component after blind deconvolution processing and the target low resonance component after blind deconvolution processing to determine the fault type of the wind power gearbox.

[0010] Optionally, obtaining a target wind turbine gearbox fault signal includes:

[0011] Collect initial wind turbine gearbox fault signals;

[0012] The initial wind turbine gearbox fault signal is periodically enhanced to obtain a target wind turbine gearbox fault signal.

[0013] Optionally, determining optimal parameters for signal resonance sparse decomposition based on the first group intelligent optimization algorithm includes:

[0014] Maximizing a first indicator based on a first group intelligent optimization algorithm, and determining a first target parameter corresponding to when the first indicator is maximized as an optimal parameter for signal resonance sparse decomposition;

[0015] Among them, the first target parameter is a parameter required for signal resonance sparse decomposition determined from a first preset parameter search range, and the first index is calculated by performing signal resonance sparse decomposition of the target wind turbine gearbox fault signal based on the first target parameter to obtain high resonance components and low resonance components.

[0016] Optionally, the first index is calculated based on the envelope spectrum kurtosis, the root mean square of the maximum value of the autocorrelation function, the center frequency ratio, and the correlation coefficient between the high resonance component and the low resonance component;

[0017] The envelope spectrum kurtosis, the root mean square of the maximum value of the autocorrelation function, the center frequency ratio, and the correlation coefficient between the target high resonance component and the target low resonance component are determined based on the high resonance component and the low resonance component.

[0018] Optionally, determining optimal parameters of blind deconvolution based on a second group intelligent optimization algorithm includes:

[0019] Maximizing the second index based on the second group intelligent optimization algorithm, and determining the second target parameter corresponding to the maximum value of the second index as the optimal parameter of the signal resonance sparse decomposition;

[0020] The second target parameter is a parameter required for blind deconvolution determined from a second preset parameter search range, and the second index is calculated by performing blind deconvolution processing on the target high resonance component and the target low resonance component based on the second target parameter, thereby obtaining a high resonance component after blind deconvolution and a low resonance component after blind deconvolution.

[0021] Optionally, the second indicator is calculated based on the envelope spectrum kurtosis and the center frequency ratio, and the envelope spectrum kurtosis and the center frequency ratio are determined based on the high resonance component after blind deconvolution processing and the low resonance component after blind deconvolution processing.

[0022] In a second aspect, the present application discloses a fault diagnosis device, comprising:

[0023] A signal acquisition module, used to acquire a target wind turbine gearbox fault signal;

[0024] A signal decomposition module, used to determine the optimal parameters of signal resonance sparse decomposition based on the first group intelligent optimization algorithm, and obtain the target high resonance component and the target low resonance component obtained by performing signal resonance sparse decomposition on the target wind turbine gearbox fault signal based on the optimal parameters of the signal resonance sparse decomposition;

[0025] A blind deconvolution module, used to determine the optimal parameters of blind deconvolution based on the second group intelligent optimization algorithm, and obtain the optimal parameters based on the blind deconvolution to perform blind deconvolution processing on the target high resonance component and the target low resonance component, to obtain the target high resonance component after blind deconvolution processing and the target low resonance component after blind deconvolution processing;

[0026] The envelope demodulation module is used to perform envelope demodulation on the high resonance component of the target after blind deconvolution processing and the low resonance component of the target after blind deconvolution processing to determine the fault type of the wind power gearbox.

[0027] In a third aspect, the present application discloses a computer program product, including a computer program / instructions, which are executed by a processor to implement the aforementioned fault diagnosis method.

[0028] In a fourth aspect, the present application discloses an electronic device, including a memory and a processor, wherein:

[0029] The memory is used to store the computer program;

[0030] The processor is used to execute the computer program to implement the aforementioned fault diagnosis method.

[0031] In a fifth aspect, the present application discloses a computer-readable storage medium for storing a computer program, wherein the computer program implements the aforementioned fault diagnosis method when executed by a processor.

[0032] It can be seen from the above scheme that the present application provides a fault diagnosis method, including: obtaining a target wind turbine gearbox fault signal; determining the optimal parameters of signal resonance sparse decomposition based on a first group intelligent optimization algorithm, and obtaining the optimal parameters based on signal resonance sparse decomposition to perform signal resonance sparse decomposition on the target wind turbine gearbox fault signal to obtain a target high resonance component and a target low resonance component; determining the optimal parameters of blind deconvolution based on a second group intelligent optimization algorithm, and obtaining the optimal parameters based on blind deconvolution to perform blind deconvolution on the target high resonance component and the target low resonance component to obtain a target high resonance component after blind deconvolution and a target low resonance component after blind deconvolution; performing envelope demodulation on the target high resonance component after blind deconvolution and the target low resonance component after blind deconvolution to determine the type of wind turbine gearbox fault.

[0033] It can be seen that the beneficial effects of the present application are: determining the optimal parameters of signal resonance sparse decomposition based on the first group intelligent optimization algorithm, obtaining high resonance components and low resonance components obtained by performing signal resonance sparse decomposition of the target wind turbine gearbox fault signal based on the optimal parameters, and determining the optimal parameters of blind deconvolution based on the second group intelligent optimization algorithm, and obtaining the enhanced signal obtained by blind deconvolution of the target high resonance component and the target low resonance component with the optimal parameters, and then performing envelope demodulation. In this way, the optimal parameters can be obtained intelligently, thereby avoiding feature omissions, thereby improving the fault diagnosis accuracy of the wind turbine gearbox.

[0034] Correspondingly, a fault diagnosis device, product, equipment and readable storage medium provided by the present application also have the above-mentioned technical effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0036] Figure 1 A flow chart of a fault diagnosis method provided in an embodiment of the present application;

[0037] Figure 2 A fault diagnosis flowchart provided for an embodiment of the present application;

[0038] Figure 3 A measured data diagram for testing provided in an embodiment of the present application;

[0039] Figure 4 A TSA-RSSD separation result and envelope spectrum provided in an embodiment of the present application;

[0040] Figure 5 A deconvolution result and envelope spectrum of CYCBD provided in an embodiment of the present application;

[0041] Figure 6 A schematic diagram of the structure of a fault diagnosis device provided in an embodiment of the present application;

[0042] Figure 7 A structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0044] As the core component of the power transmission of the unit, the reliability of the wind turbine gearbox is directly related to the operating efficiency and economy of the power generation system. According to statistics, downtime accidents caused by gearbox failures account for 20%-30% of the total failure rate of wind turbines, and a single unplanned downtime can cause hundreds of thousands of yuan in economic losses. As the average service life of wind power equipment worldwide exceeds 10 years, the incidence of multi-component composite failures has increased significantly, and its vibration signal presents new characteristics of multi-excitation source coupling and mutual modulation of fault characteristics. Especially under low signal-to-noise ratio conditions, the superposition of strong background noise (extreme climate loads, internal wear impact, etc.) and non-stationary signals caused by wind speed fluctuations makes it difficult for traditional methods to effectively separate fault components, which seriously restricts fault warning and precise operation and maintenance capabilities. Therefore, how to improve the fault diagnosis accuracy of wind turbine gearboxes is a technical problem that technicians in this field need to solve. To this end, the present application provides a fault diagnosis scheme that can improve the fault diagnosis accuracy of wind turbine gearboxes.

[0045] See also Figure 1 As shown, the embodiment of the present application discloses a fault diagnosis method, including:

[0046] Step S11: Acquire a target wind turbine gearbox fault signal.

[0047] In an optional implementation, an initial wind turbine gearbox fault signal may be collected; the initial wind turbine gearbox fault signal may be periodically enhanced to obtain a target wind turbine gearbox fault signal.

[0048] It should be pointed out that in the case of complex working conditions and strong noise, the effect of direct sparse decomposition is not good. The embodiment of the present application first periodically enhances the initial wind turbine gearbox fault signal to enhance the periodic fault characteristic signal, while suppressing random noise and non-periodic interference.

[0049] The periodic enhancement may include the following steps:

[0050] Signal preprocessing: Preprocess the collected initial wind turbine gearbox fault signal, including filtering and denoising, to reduce the impact of non-periodic components on subsequent analysis.

[0051] Signal segmentation: Split a continuous signal into multiple equal-length signal segments, each of which contains a complete gear rotation cycle.

[0052] Phase Adjustment: Phase adjustment is performed on each signal segment to ensure that all signal segments are aligned at the same phase.

[0053] Synchronous averaging: Synchronous averaging is performed on all phase-adjusted signal segments to eliminate random noise and periodic signals that are not completely synchronized with the gear of interest. The result of synchronous averaging is a signal that enhances the fault pulse, in which the periodic component of the fault pulse is significantly enhanced and the random noise is suppressed. The signal after synchronous averaging is the target wind turbine gearbox fault signal. This step can significantly enhance the periodic fault characteristic signal while suppressing random noise and non-periodic interference.

[0054] Step S12: determining optimal parameters of signal resonance sparse decomposition based on the first group intelligent optimization algorithm, and obtaining target high resonance components and target low resonance components by performing signal resonance sparse decomposition on the target wind turbine gearbox fault signal based on the optimal parameters of signal resonance sparse decomposition.

[0055] In the embodiment of the present application, the first swarm intelligence optimization algorithm and the second swarm intelligence optimization algorithm can be the same swarm intelligence optimization algorithm or different swarm intelligence optimization algorithms. The embodiment of the present application does not limit the swarm intelligence optimization algorithm used, and the first swarm intelligence optimization algorithm and the second swarm intelligence optimization algorithm can include but are not limited to ant colony algorithm, particle swarm optimization algorithm, etc.

[0056] The embodiment of the present application can maximize the first index based on the first group intelligent optimization algorithm, and determine the first target parameter corresponding to the maximum of the first index as the optimal parameter of the signal resonance sparse decomposition; wherein the first target parameter is the parameter required for the signal resonance sparse decomposition determined from the first preset parameter search range, and the first index is calculated by the high resonance component and the low resonance component obtained by performing the signal resonance sparse decomposition on the target wind power gearbox fault signal based on the first target parameter. The first target parameter is the quality factor Q and the redundancy factor r of the signal resonance sparse decomposition algorithm, which can include Q and r corresponding to the high resonance component and Q and r corresponding to the low resonance component. The high resonance component is mainly composed of multiple simultaneous continuous oscillation signals, which usually have stable periodic fluctuations. The low resonance component is composed of multiple transient impact signals without specific status and duration, which show non-periodic and sudden fluctuations in the time domain. The high resonance signal has a higher quality factor, which means that the frequency components of the signal are relatively concentrated and the bandwidth is narrow. It mainly reflects the periodic characteristics of the signal, such as the stable fluctuation of mechanical vibration, the periodic change of the sound wave signal, etc. Low resonance component: has a lower quality factor, the frequency components of the signal are relatively dispersed, the bandwidth is wide, and it mainly reflects the transient impact characteristics of the signal, such as sudden fluctuations caused by noise, interference, mutation, etc. High resonance components mainly reflect gear faults, and low resonance components mainly reflect bearing faults.

[0057] In an optional embodiment, the first indicator is calculated based on the envelope spectrum kurtosis, the root mean square of the maximum value of the autocorrelation function, the center frequency ratio, and the correlation coefficient of the high resonance component and the low resonance component; wherein the envelope spectrum kurtosis, the root mean square of the maximum value of the autocorrelation function, the center frequency ratio, the correlation coefficient of the target high resonance component and the target low resonance component are determined based on the high resonance component and the low resonance component.

[0058] The calculation formula of the first indicator is as follows:

[0059] ;

[0060] Among them, R is the first index, AFM represents the root mean square of the maximum value of the autocorrelation function, CFR represents the center frequency ratio, EKS represents the envelope spectrum kurtosis, HLcc is the correlation coefficient of the high-low component, which represents the correlation between the high resonance component and the low resonance component; the subscript high corresponds to the high resonance component, and low corresponds to the low resonance component, that is, is the root mean square of the maximum value of the autocorrelation function corresponding to the high resonance component, and the autocorrelation function is calculated for the high resonance component. is the envelope spectrum kurtosis of the low resonance component. g corresponds to the characteristic frequency of gear fault, and b corresponds to the characteristic frequency of bearing fault. Indicates the center frequency ratio corresponding to the characteristic frequency of gear fault, It represents the center frequency ratio corresponding to the characteristic frequency of the bearing fault. CFR changes with the signal and reflects the energy ratio of this frequency in the full frequency band of the signal.

[0061] It should be pointed out that kurtosis is an important indicator that reflects the deviation of the signal from the normal distribution. If the signal contains abundant fault impact components, there will be significant fault-related characteristic frequencies and their corresponding frequency multiples in the signal envelope spectrum. Therefore, its envelope spectrum deviates greatly from the normal distribution, resulting in a significant increase in the spectrum kurtosis value. Therefore, the envelope spectrum kurtosis is used to judge the richness of fault information in the low resonance component. The root mean square of the maximum value of the autocorrelation function is used to reflect the useful weak periodic fault influence component in the signal. The core of this indicator is the autocorrelation function. The autocorrelation function of noise decays rapidly to zero, and its AFM value is very small. If the high resonance component contains useful periodic fault impact components, its autocorrelation function is periodic, and the more periodic fault pulse components it contains, the more its AFM value will increase significantly. The same components may exist in high resonance and low resonance components. In order to effectively suppress this phenomenon and decompose the fault information of different components into different components as much as possible, the embodiment of the present application introduces the characteristic frequency ratio CFR, which is a dimensionless parameter that characterizes the energy ratio of the fault characteristic frequency in the full frequency band of the envelope spectrum. The larger the value of CFR, the more fault information there is in the sub-band. At the same time, the embodiment of the present application also introduces the cross-correlation coefficient HLcc to minimize the same components between the high resonance component and the low resonance component. The envelope spectrum kurtosis (EKS) is used to judge the richness of the fault information in the low resonance component. Because when the signal contains a fault shock, the corresponding envelope spectrum will have obvious fault characteristic frequencies and their harmonic components, causing the envelope spectrum to seriously deviate from the normal distribution.

[0062] Step S13: Determine the optimal parameters of blind deconvolution based on the second group intelligent optimization algorithm, and obtain the optimal parameters based on blind deconvolution to perform blind deconvolution on the target high resonance component and the target low resonance component, to obtain the target high resonance component after blind deconvolution and the target low resonance component after blind deconvolution.

[0063] This embodiment can maximize the second index based on the second group intelligent optimization algorithm, and determine the second target parameter corresponding to the maximum of the second index as the optimal parameter for signal resonance sparse decomposition; wherein the second target parameter is a parameter required for blind deconvolution determined from a second preset parameter search range, and the second index is calculated by performing blind deconvolution processing on the target high resonance component and the target low resonance component based on the second target parameter, and the high resonance component after blind deconvolution processing and the low resonance component after blind deconvolution processing are obtained. The second target parameter may include a filter length and a characteristic frequency.

[0064] In an optional embodiment, the second index is calculated based on the envelope spectrum kurtosis and the center frequency ratio, and the envelope spectrum kurtosis and the center frequency ratio are determined based on the high resonance component after blind deconvolution processing and the low resonance component after blind deconvolution processing.

[0065] The calculation formula of the second indicator is as follows:

[0066] ;

[0067] in, is the parameter of blind deconvolution, which characterizes the fault characteristic frequency of the component (which is a priori information). In other words, to extract the high resonance component, is the characteristic frequency of gear fault; low resonance component, is the bearing fault characteristic frequency, express The corresponding center frequency ratio, It represents the high resonance component after blind deconvolution processing or the low resonance component after blind deconvolution processing, that is, in the embodiment of the present application, there is a second index corresponding to the high resonance component and a second index corresponding to the low resonance component, which can be calculated separately.

[0068] Step S14: performing envelope demodulation on the target high resonance component after blind deconvolution processing and the target low resonance component after blind deconvolution processing to determine the fault type of the wind turbine gearbox.

[0069] In the embodiment of the present application, the fault type is determined by observing the existence and significance of characteristic frequencies corresponding to gear and bearing faults in the envelope spectrum after envelope demodulation.

[0070] It can be seen that the embodiment of the present application determines the optimal parameters of the signal resonance sparse decomposition based on the first group intelligent optimization algorithm, obtains the high resonance component and the low resonance component obtained by performing the signal resonance sparse decomposition on the target wind turbine gearbox fault signal based on the optimal parameters, and determines the optimal parameters of the blind deconvolution based on the second group intelligent optimization algorithm, and obtains the enhanced signal obtained by blind deconvolution processing of the target high resonance component and the target low resonance component with the optimal parameters, and then performs envelope demodulation, so that the optimal parameters can be obtained intelligently, thereby avoiding feature omissions, thereby improving the fault diagnosis accuracy of the wind turbine gearbox. The embodiment of the present application realizes the composite fault diagnosis of the wind turbine gearbox and improves the composite fault diagnosis accuracy.

[0071] For further information, see Figure 2 As shown, Figure 2 A fault diagnosis flow chart provided in the embodiment of the present application. Wind turbine gearbox fault diagnosis is implemented based on TSA, RSSD, and CYCBD, including the following steps:

[0072] Period enhancement of fault pulses: The collected weak composite fault signals of the gearbox are preliminarily processed using time synchronous averaging (TSA) to reduce noise and highlight periodic pulses.

[0073] Use resonance sparse decomposition to decouple composite faults: The optimal parameters are obtained through the swarm intelligence optimization algorithm. According to the optimal parameters Q and r, signal resonance sparse decomposition (RSSD) is performed. According to the different resonance properties of bearing and gear faults, the composite fault is decoupled into high resonance (mainly reflecting gear faults) and low resonance (mainly reflecting bearing faults) components.

[0074] Use CYCBD to enhance and extract weak fault features: The optimal parameters are obtained through the swarm intelligence optimization algorithm, and blind deconvolution (CYCBD) is performed according to the optimal parameters: filter length and characteristic frequency to enhance the fault features of high resonance (mainly reflecting gear faults) and low resonance (mainly reflecting bearing faults) components.

[0075] Envelope demodulation and fault diagnosis: Envelope demodulation is performed on the high-resonance and low-resonance signals enhanced by CYCBD. The specific fault type of the gear or bearing can be identified by observing the fault characteristic frequency and its multiple frequency in the envelope spectrum.

[0076] Fault diagnosis under strong noise conditions is often interfered by noise, and it is difficult to achieve accurate fault feature separation and extraction. To this end, the embodiment of the present application introduces time domain synchronous averaging (TSA) before resonant sparse decomposition, and averages the vibration signal by synchronizing with the rotation frequency of the mechanical equipment, effectively suppressing the non-periodic noise components and enhancing the periodic signal related to the equipment speed. The embodiment of the present application uses resonant sparse decomposition to solve the problem of frequency band overlap when processing composite faults through adjustable Q factor wavelet transform and morphological component analysis, adopts the RSSD method of parallel dual parameter optimization, and uses multi-objective optimization to automatically determine key parameters, thereby improving the separation effect of weak composite faults of planetary gearboxes. The CYCBD algorithm solves the problem that fault signals are easily interfered by noise under low signal-to-noise ratio conditions, and uses multi-objective optimization to automatically determine key parameters, solving the problem that its filter optimization process may not reach the global optimum, so as to fully reveal the fault information. The example of the present application can be applied to extract weak but important fault feature information from complex vibration signals. In an optional embodiment, deconvolution methods such as minimum entropy deconvolution (MED) can also be used to enhance the fault feature extraction capability, but the CYCBD algorithm is more effective.

[0077] The embodiment of the present application can separate and diagnose weak composite faults, and can realize composite fault diagnosis of wind turbine gearboxes under strong noise conditions, and has broad application prospects. The technical solution proposed by the present invention is practical and innovative, and can be widely used in the fields of wind turbine fault diagnosis, and has important economic and social value.

[0078] In a specific implementation, the present application may include the following steps:

[0079] The collected gearbox composite fault signal is input into TSA to enhance the period of fault pulse;

[0080] The steps to obtain the optimal RSSD parameters through the swarm intelligence optimization algorithm are as follows: set the search range of the parameters, as well as the SSA parameters, and calculate the number of decomposition layers; the optimized index R is calculated by EKS (envelope spectrum kurtosis), AFM (root mean square of the maximum value of the autocorrelation function), CFR (center frequency ratio), and HLcc (high-low correlation coefficient). The optimal RSSD parameters are determined by maximizing R, and then the signal resonance sparse decomposition (RSSD) is performed according to the optimal parameters Q and r to decompose the periodic enhanced signal into different high-resonance and low-resonance components, or the high-resonance components and low-resonance components corresponding to the optimal index R are obtained from the various high-resonance components and low-resonance components determined in the process of optimizing the index R by the swarm intelligence optimization algorithm, and the target high-resonance component and the target low-resonance component are obtained.

[0081] The steps to obtain the optimal parameters of CYCBD through the swarm intelligence optimization algorithm are as follows: set the parameter search range and SSA parameters for the high resonance and low resonance components respectively; the optimized index P is calculated by EKS (envelope spectrum kurtosis) and CFR (center frequency ratio). The optimal CYCBD parameters are determined by maximizing P, and then blind deconvolution (CYCBD) is performed according to the optimal parameters: filter length and characteristic frequency to achieve fault feature enhancement of high resonance (mainly reflecting gear faults) and low resonance (mainly reflecting bearing faults) components. Alternatively, the enhanced high resonance component and enhanced low resonance component corresponding to the optimal index P are obtained from each enhanced high resonance component and enhanced low resonance component determined in the process of optimizing the index P by the swarm intelligence optimization algorithm.

[0082] The envelope demodulation is performed on the two blind deconvolved signals. The fault type is determined by observing the presence and significance of characteristic frequencies corresponding to gear and bearing faults in the envelope spectrum. The indicator construction process is as follows:

[0083] ;

[0084] .

[0085] In this way, the separation and diagnosis of weak compound faults of wind turbine gearboxes can be realized under strong noise conditions, and a parameter-adaptive TSA-RSSD-CYCBD fusion framework is proposed to overcome the following technical difficulties:

[0086] Parameter adaptability: Construct a multi-objective optimization model to automatically search for the optimal RSSD Q factor and redundancy factor combination, CYCBD filter length and characteristic frequency combination, and solve the problem of feature missed detection caused by empirical parameter adjustment;

[0087] Noise suppression mechanism: Design a TSA-RSSD composite fault enhancement separation structure to enhance the visibility of weak impact components under low signal-to-noise ratio through a periodic pulse enhancement strategy;

[0088] Composite feature decoupling: In the process of optimizing the parameters of signal resonance sparse decomposition, the root mean square comprehensive optimization index of the envelope spectrum kurtosis-autocorrelation function maximum value is designed to achieve the physical separation of composite faults such as gear spalling and bearing cracks.

[0089] Experiments show that this method can still achieve separation diagnosis of compound faults under -12dB Gaussian noise interference and actual working conditions, providing a new technical tool for wind turbine gearbox weak compound fault diagnosis. Figures 3 to 5 As shown, Figure 3 A measured data diagram for testing provided in an embodiment of the present application; Figure 4 A TSA-RSSD separation result and envelope spectrum provided in an embodiment of the present application; Figure 5 The deconvolution result and envelope spectrum of a CYCBD provided in an embodiment of the present application.

[0090] The present application embodiment uses the MCC5-THU gearbox multi-mode composite fault data set, Figure 3 The waveform (a), Fourier spectrum (b) and envelope spectrum (c) of the composite fault signal of gear tooth breakage and bearing inner ring damage obtained by the vertically mounted accelerometer sensor are shown. Figure 3 (a) As can be seen from the waveform, the regular impact component caused by the broken tooth has been masked by the strong background noise. This shows that in the time domain, it is extremely difficult to identify the fault characteristics directly from the waveform. In the frequency domain, Figure 3 (b) The Fourier spectrum shows that the main frequency components of the signal are concentrated in the range of 800 Hz to 1500 Hz. However, the fault characteristic frequencies related to the gear and the inner race of the bearing are not seen in the low frequency band. This indicates that the traditional frequency domain analysis method may not be sufficient to effectively extract the fault characteristics in this case. Figure 3 (c) shows the envelope demodulation processing result of the actual fault signal. Although there are subtle peaks at the gear fault characteristic frequency f and its double frequency 2f, the characteristic frequency and harmonic components related to the bearing inner race fault are not clearly identified. Instead, noise and other irrelevant interference frequencies dominate, resulting in unsatisfactory envelope demodulation results.

[0091] Figure 4 (a) and (c) depict the waveform and envelope spectrum of the high resonance component, respectively. It can be observed that the waveform of the high resonance component shows obvious periodic impact characteristics, and the sun gear fault characteristic frequency f and harmonic components (2fs, 3fs) appear in its corresponding envelope spectrum. The waveform and envelope spectrum of the low resonance component are shown in Figure 2. Figure 4(b) and (d). It can be seen that the waveform of the low resonance component shows obvious transient impact characteristics. In addition, the outer ring fault characteristic frequency f0 and its multiple components (2f0, 3f0 and 5f0) can be identified in its corresponding envelope spectrum. Therefore, the results show that the RSSD method can successfully separate the weak coupling fault characteristics composed of sun gear tooth cracks and planetary bearing outer ring damage. However, the high and low resonance components are still severely interfered by noise, and the characteristic frequencies corresponding to different fault types and their multiple components do not dominate in the envelope spectrum, which reduces the accuracy of feature extraction.

[0092] Figure 5 shows the filtering results obtained after CYCBD. Figure 5 (a) The filtered waveform of the high resonance component shows that the periodic fault effect hidden in the composite fault signal is clearly revealed, and the noise component is effectively suppressed. The fault characteristic frequency of the sun gear and its first 4 times (dominate in its corresponding filtered envelope spectrum, such as Figure 5 (b) shows the extracted characteristic frequency information directly indicating the sun gear fault. Figure 5 It can be found in (c) that the filtered waveform of the low resonance component clearly shows the fault influence at equal intervals. Figure 5 The corresponding envelope spectrum shown in (d) shows that the characteristic frequency of the outer race fault and its first 7 multiples are obvious. Therefore, these features intuitively characterize the damage of the planetary bearing.

[0093] See also Figure 6 As shown, an embodiment of the present application provides a fault diagnosis device, comprising:

[0094] A signal acquisition module 11 is used to acquire a target wind turbine gearbox fault signal;

[0095] The signal decomposition module 12 is used to determine the optimal parameters of the signal resonance sparse decomposition based on the first group intelligent optimization algorithm, and obtain the target high resonance component and the target low resonance component obtained by performing the signal resonance sparse decomposition on the target wind turbine gearbox fault signal based on the optimal parameters of the signal resonance sparse decomposition;

[0096] A blind deconvolution module 13 is used to determine the optimal parameters of blind deconvolution based on the second group intelligent optimization algorithm, and obtain the optimal parameters based on the blind deconvolution to perform blind deconvolution processing on the target high resonance component and the target low resonance component, to obtain the target high resonance component after blind deconvolution processing and the target low resonance component after blind deconvolution processing;

[0097] The envelope demodulation module 14 is used to perform envelope demodulation on the target high resonance component after blind deconvolution processing and the target low resonance component after blind deconvolution processing to determine the fault type of the wind turbine gearbox.

[0098] In an optional implementation, the signal acquisition module 11 is specifically used to collect an initial wind turbine gearbox fault signal; and periodically enhance the initial wind turbine gearbox fault signal to obtain a target wind turbine gearbox fault signal.

[0099] In an optional embodiment, the signal decomposition module 12 is specifically used to maximize the first indicator based on the first group intelligent optimization algorithm, and determine the first target parameter corresponding to the maximum value of the first indicator as the optimal parameter for signal resonance sparse decomposition; wherein the first target parameter is a parameter required for signal resonance sparse decomposition determined from a first preset parameter search range, and the first indicator is calculated by performing signal resonance sparse decomposition of the target wind turbine gearbox fault signal based on the first target parameter to obtain a high resonance component and a low resonance component.

[0100] Optionally, the first index is calculated based on the envelope spectrum kurtosis, the root mean square of the maximum value of the autocorrelation function, the center frequency ratio, and the correlation coefficient between the high resonance component and the low resonance component;

[0101] The envelope spectrum kurtosis, the root mean square of the maximum value of the autocorrelation function, the center frequency ratio, and the correlation coefficient between the target high resonance component and the target low resonance component are determined based on the high resonance component and the low resonance component.

[0102] In an optional embodiment, the blind deconvolution module 13 is used to maximize the second indicator based on the second group intelligent optimization algorithm, and determine the second target parameter corresponding to the maximum of the second indicator as the optimal parameter of the signal resonance sparse decomposition;

[0103] The second target parameter is a parameter required for blind deconvolution determined from a second preset parameter search range, and the second index is calculated by performing blind deconvolution processing on the target high resonance component and the target low resonance component based on the second target parameter, thereby obtaining a high resonance component after blind deconvolution and a low resonance component after blind deconvolution.

[0104] Optionally, the second indicator is calculated based on the envelope spectrum kurtosis and the center frequency ratio, and the envelope spectrum kurtosis and the center frequency ratio are determined based on the high resonance component after blind deconvolution processing and the low resonance component after blind deconvolution processing.

[0105] It can be seen that the embodiment of the present application determines the optimal parameters of signal resonance sparse decomposition based on the first group intelligent optimization algorithm, obtains the high resonance component and the low resonance component obtained by performing signal resonance sparse decomposition on the target wind turbine gearbox fault signal based on the optimal parameters, and determines the optimal parameters of blind deconvolution based on the second group intelligent optimization algorithm, and obtains the enhanced signal obtained by blind deconvolution of the target high resonance component and the target low resonance component with the optimal parameters, and then performs envelope demodulation. In this way, the optimal parameters can be obtained intelligently, thereby avoiding feature omissions, thereby improving the fault diagnosis accuracy of the wind turbine gearbox.

[0106] Furthermore, the present application discloses a computer program product, including a computer program / instruction, which is executed by a processor to implement the aforementioned fault diagnosis method.

[0107] For the specific process of the above fault diagnosis method, reference may be made to the corresponding contents disclosed in the above embodiments, which will not be described in detail here.

[0108] See also Figure 7 As shown, an embodiment of the present application discloses an electronic device 20, including a processor 21 and a memory 22; wherein the memory 22 is used to store a computer program; and the processor 21 is used to execute the computer program, the fault diagnosis method disclosed in the above embodiment.

[0109] For the specific process of the above fault diagnosis method, reference may be made to the corresponding contents disclosed in the above embodiments, which will not be described in detail here.

[0110] Furthermore, the memory 22 as a carrier for storing resources may be a read-only memory, a random access memory, a magnetic disk or an optical disk, etc., and the storage method may be temporary storage or permanent storage.

[0111] In addition, the electronic device 20 also includes a power supply 23, a communication interface 24, an input / output interface 25 and a communication bus 26; wherein the power supply 23 is used to provide working voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and an external device, and the communication protocol it follows is any communication protocol that can be applied to the technical solution of the present application, and is not specifically limited here; the input / output interface 25 is used to obtain external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0112] Furthermore, an embodiment of the present application also discloses a computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the fault diagnosis method disclosed in the aforementioned embodiment.

[0113] For the specific process of the above fault diagnosis method, reference may be made to the corresponding contents disclosed in the above embodiments, which will not be described in detail here.

[0114] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0115] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0116] The above is a detailed introduction to a fault diagnosis method, device, product, equipment and medium provided by the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for general technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

[0117] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not limiting. Although the present application has been described by referring to the preferred embodiments of the present application, a person skilled in the art should understand that various changes can be made in form and details without departing from the spirit and scope of the present application as defined in the appended claims.

Claims

1. A fault diagnosis method, characterized in that: include: Obtaining a target wind turbine gearbox fault signal; Determine the optimal parameters of signal resonance sparse decomposition based on the first group intelligent optimization algorithm, and obtain the target high resonance component and target low resonance component obtained by performing signal resonance sparse decomposition on the target wind turbine gearbox fault signal based on the optimal parameters of signal resonance sparse decomposition; Determine the optimal parameters of blind deconvolution based on the second group intelligent optimization algorithm, and obtain the optimal parameters based on blind deconvolution to perform blind deconvolution processing on the target high resonance component and the target low resonance component, to obtain the target high resonance component after blind deconvolution processing and the target low resonance component after blind deconvolution processing; Envelope demodulation is performed on the target high resonance component after blind deconvolution processing and the target low resonance component after blind deconvolution processing to determine the fault type of the wind power gearbox.

2. The fault diagnosis method according to claim 1, characterized in that: Obtain the target wind turbine gearbox fault signal, including: Collect initial wind turbine gearbox fault signals; The initial wind turbine gearbox fault signal is periodically enhanced to obtain a target wind turbine gearbox fault signal.

3. The fault diagnosis method according to claim 1, characterized in that: The optimal parameters of signal resonance sparse decomposition are determined based on the first group intelligent optimization algorithm, including: Maximizing a first indicator based on a first group intelligent optimization algorithm, and determining a first target parameter corresponding to when the first indicator is maximized as an optimal parameter for signal resonance sparse decomposition; Among them, the first target parameter is a parameter required for signal resonance sparse decomposition determined from a first preset parameter search range, and the first index is calculated by performing signal resonance sparse decomposition of the target wind turbine gearbox fault signal based on the first target parameter to obtain high resonance components and low resonance components.

4. The fault diagnosis method according to claim 3, characterized in that: The first index is calculated based on the envelope spectrum kurtosis, the root mean square of the maximum value of the autocorrelation function, the center frequency ratio, and the correlation coefficient between the high resonance component and the low resonance component; The envelope spectrum kurtosis, the root mean square of the maximum value of the autocorrelation function, the center frequency ratio, and the correlation coefficient between the target high resonance component and the target low resonance component are determined based on the high resonance component and the low resonance component.

5. The fault diagnosis method according to claim 1, characterized in that: The optimal parameters of blind deconvolution are determined based on the second group intelligent optimization algorithm, including: Maximizing the second index based on the second group intelligent optimization algorithm, and determining the second target parameter corresponding to the maximum value of the second index as the optimal parameter of the signal resonance sparse decomposition; The second target parameter is a parameter required for blind deconvolution determined from a second preset parameter search range, and the second index is calculated by performing blind deconvolution processing on the target high resonance component and the target low resonance component based on the second target parameter, thereby obtaining a high resonance component after blind deconvolution and a low resonance component after blind deconvolution.

6. The fault diagnosis method according to claim 5, characterized in that: The second index is calculated based on the envelope spectrum kurtosis and the center frequency ratio, and the envelope spectrum kurtosis and the center frequency ratio are determined based on the high resonance component after blind deconvolution processing and the low resonance component after blind deconvolution processing.

7. A fault diagnosis device, characterized in that: include: A signal acquisition module, used to acquire a target wind turbine gearbox fault signal; A signal decomposition module, used to determine the optimal parameters of signal resonance sparse decomposition based on the first group intelligent optimization algorithm, and obtain the target high resonance component and the target low resonance component obtained by performing signal resonance sparse decomposition on the target wind turbine gearbox fault signal based on the optimal parameters of the signal resonance sparse decomposition; A blind deconvolution module, used to determine the optimal parameters of blind deconvolution based on the second group intelligent optimization algorithm, and obtain the optimal parameters based on the blind deconvolution to perform blind deconvolution processing on the target high resonance component and the target low resonance component, to obtain the target high resonance component after blind deconvolution processing and the target low resonance component after blind deconvolution processing; The envelope demodulation module is used to perform envelope demodulation on the high resonance component of the target after blind deconvolution processing and the low resonance component of the target after blind deconvolution processing to determine the fault type of the wind power gearbox.

8. A computer program product, characterized in that The method comprises a computer program / instruction, wherein the computer program / instruction is executed by a processor to implement the fault diagnosis method according to any one of claims 1 to 6.

9. An electronic device, characterized in that: comprising a memory and a processor, wherein: The memory is used to store the computer program; The processor is used to execute the computer program to implement the fault diagnosis method according to any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that: Used to store a computer program, wherein 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