A wind turbine gearbox fault diagnosis method and system

Abstract

The present invention is applicable to the field of intelligent identification of gearbox faults, and is specifically a method and system for diagnosing faults of gearboxes of wind turbines, the method comprising: collecting vibration signals of key components in the gearbox of the wind turbine; initializing parameters of a variational modal decomposition algorithm, using the variational modal decomposition algorithm to perform modal decomposition on the vibration signals of the key components to obtain multiple IMF components, using a wavelet packet analysis method to extract energy entropy features of each modal IMF component of the optimal decomposition modal number, obtaining a fault feature vector of the gearbox vibration signal, inputting the vector into an SVM for training and identification, and realizing fault diagnosis of key components in the gearbox of the wind turbine. The fault feature vector extracted by the present invention can fully reflect the true state of the gearbox vibration signal, improve the accuracy of gearbox fault diagnosis, and solve the problem of insufficient effective signal components in the decomposed IMF components caused by problems such as modal aliasing and endpoint effect in the traditional EMD method.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent identification of gearbox faults, and in particular relates to a method and system for diagnosing gearbox faults of a wind turbine generator set. Background Art

[0002] The gearbox is an important component in the transmission system of a wind turbine, used to connect the generator and the main shaft; the interior of the gearbox is composed of a sun gear, three planetary gears, an inner ring gear, a planetary carrier and a two-stage parallel gear transmission. The gearbox has the advantages of compact structure, high power density and high transmission efficiency.

[0003] The internal structure and stress of the gearbox are complex, especially when running under variable working conditions and loads, which makes it easy to fail. The gearbox is not easy to disassemble and assemble. With the increase of operating load and the harsh working environment, once a failure occurs, it will have a great impact on the generator set. Therefore, realizing the status monitoring and fault diagnosis function of the gearbox is of great significance to the safe operation of wind turbines.

[0004] In the related technologies, the vibration signal used in gearbox fault diagnosis is basically measured by a sensor installed on the surface of the gearbox. The signal contains gear meshing vibration signals and a large amount of noise interference components, which increases the difficulty of extracting gearbox fault features. The traditional empirical mode decomposition (EMD) method has attracted widespread attention from researchers in the field of fault diagnosis due to its advantages such as suitability for the decomposition of nonlinear and non-stationary signals and strong adaptive decomposition capabilities. Some scholars perform EMD on rolling bearing vibration signals, extract multiple entropy values ​​of intrinsic mode function (IMF) components, fuse them, and input them into support vector machines for fault diagnosis. Another technology performs EMD denoising on the collected rolling bearing fault signals, and uses spectral kurtosis to extract features from the denoised signals, thereby achieving fault diagnosis.

[0005] However, problems such as modal aliasing and endpoint effect in the above-mentioned EMD method will lead to insufficient effective signal components in the decomposed IMF components, which will in turn affect the extraction of fault feature information in the gearbox vibration signal. The inability to extract effective vibration signal fault feature information will lead to poor accuracy of subsequent gearbox fault diagnosis. Summary of the invention

[0006] The purpose of the embodiments of the present invention is to provide a wind turbine gearbox fault diagnosis method and system, aiming to solve the technical problems of poor diagnostic accuracy and low accuracy in traditional wind turbine gearbox fault diagnosis technology.

[0007] To achieve the above object, the present invention provides the following technical solutions.

[0008] In one embodiment of the present disclosure, a wind turbine gearbox fault diagnosis method is provided, the diagnosis method comprising the following steps:

[0009] Collect vibration signals of key components in wind turbine gearboxes;

[0010] Initialize the parameters of the variational modal decomposition algorithm, use the variational modal decomposition algorithm to perform modal decomposition on the vibration signals of key components to obtain multiple IMF components, use the multi-scale linear fitting method to analyze each IMF component, evaluate the linear characteristics of each IMF component, and determine the preset K value of the decomposition based on the changes in the linear characteristics;

[0011] Calculate the sparsity corresponding to each K value based on the value interval, and take the K value corresponding to the maximum sparsity as the optimal decomposition mode number, where the value interval is between 2 and the preset K value;

[0012] The wavelet packet analysis method is used to extract the energy entropy characteristics of each modal IMF component of the optimal decomposition modal number, and the fault feature vector of the gearbox vibration signal is obtained;

[0013] A fault feature dataset is constructed and input into SVM for training and identification to realize fault diagnosis of key components in wind turbine gearbox.

[0014] As a further limitation of the technical solution of the present invention, the step of performing modal decomposition on the vibration signal of the key component using the variational modal decomposition algorithm to obtain multiple IMF components includes:

[0015] Perform VMD decomposition on the wind turbine gearbox vibration signal according to the initialization parameters to obtain k IMF original components;

[0016] The analytical signal of each IMF original component is solved by Hilbert transform to obtain a one-sided spectrum:

[0017] (1);

[0018] (2);

[0019] In formula (1), represents the modal function, k is a positive integer, k=1,2,...,K; K represents the number of modes to be decomposed;

[0020] According to the mixed estimated center frequency of each modal analysis signal, the single-side frequency is modulated to the corresponding baseband to obtain the modulated signal:

[0021] (3);

[0022] In formula (3), represents the exponential signal, represents the center frequency;

[0023] Calculate the square norm of the modulated signal gradient, estimate the bandwidth of each IMF component, and construct a constrained variational model;

[0024] By introducing penalty parameters and Lagrange multipliers to solve the variational model, the frequency domain expression of the original component of the IMF, the update formula of the center frequency, and the update formula of the penalty factor are obtained;

[0025] According to the frequency domain expression of the original component of the IMF, the update formula of the center frequency, and the update formula of the penalty factor, the updated frequency domain value, the updated center frequency, and the updated Lagrange multiplier are obtained;

[0026] According to the updated frequency domain value, the updated center frequency and the updated Lagrange multiplier, the k IMF original components are converted into amplitude modulation-frequency modulation signals to obtain k IMF components.

[0027] As a further limitation of the technical solution of the present invention, in the step of constructing a constrained variational model, the constrained variational model is expressed as follows:

[0028] (4);

[0029] In formula (4), represents the unit pulse function, j represents the imaginary unit, represents the convolution operation, It means to find the partial derivative of the function; represents the k IMF components after decomposition, ; represents the center frequency of each component, .

[0030] As a further limitation of the technical solution of the present invention, in the step of calculating the sparsity corresponding to each K value, the sparsity of each IMF component is expressed as:

[0031] (5);

[0032] In formula (5), represents the energy weight factor, which is defined as the ratio of the maximum amplitude of the unilateral spectrum of the IMF component to the maximum amplitude of the marginal spectra of the K components;

[0033] The unilateral spectrum sparseness of each IMF is taken as the overall sparsity under the decomposition mode number K:

[0034] (6);

[0035] when When , the optimal decomposition mode number is determined to be K-1.

[0036] As a further limitation of the technical solution of the present invention, in the step of analyzing each IMF component using a multi-scale straight line fitting method, evaluating the linear characteristics of each IMF component, and determining the preset K value of the adaptive decomposition according to the change of the linear characteristics, for each IMF component, according to the initial scale, a multi-segment fitting straight line is used to replace it, wherein the calculation formula for the selection of the initial scale is as follows:

[0037] ;

[0038] Among them, each IMF component is divided into m segments, For each component sequence, the least squares method is used to fit the straight line to minimize the residual sum of squares (RSS). The slope corresponding to the fitting straight line of each component sequence is obtained and the slopes are compared. If the slope characteristics of adjacent components are similar, the adjacent components are merged into a wider mode. If the slope characteristics of adjacent components are significantly different, the adjacent components are split into multiple modes until the slope characteristics of the components corresponding to the split modes are similar. The splitting is stopped to obtain the preset K value.

[0039] As a further limitation of the technical solution of the present invention, the process of solving the wavelet packet energy entropy of the modal IMF component signal includes:

[0040] First, the modal IMF component is decomposed by wavelet packets to obtain multiple sub-signals of different frequency bands. Furthermore, each frequency band is reconstructed by wavelet packets, and then the wavelet packet energy value of each frequency band is calculated. After that, the wavelet packet energy value of each frequency band is normalized, and the wavelet packet energy entropy is calculated to obtain the fault feature vector of the gearbox vibration signal.

[0041] As a further limitation of the technical solution of the present invention, in the step of constructing a fault feature data set, inputting it into SVM for training and identification, and realizing fault diagnosis of key components in the gearbox of the wind turbine, since the fault feature vectors contained in the constructed fault feature data set are indivisible samples, a kernel function is used to map the fault feature data set to a high-dimensional space;

[0042] The kernel function in the embodiment of the present invention is expressed as:

[0043] (7);

[0044] In formula (7), Represents the kernel function, which is used to calculate the two samples in the fault feature dataset , similarities between; Represents a mapping function, which is used to transform samples of the fault feature dataset , Mapped to high-dimensional space; T represents vector transpose operation;

[0045] Find a hyperplane in high-dimensional space so that the distance between each data point and the hyperplane is the largest. The hyperplane prediction model constructed is expressed as:

[0046] (8);

[0047] In formula (8), represents the decision function of the SVM model, which is used to predict the category of new sample signal data of gearbox fault; represents a mapping function; represents the weight vector; b represents the bias term, which is used to control the position of the hyperplane along the normal direction; T represents the vector transposition operation;

[0048] The model is used to perform fault diagnosis and identification on the newly input gearbox vibration signal fault feature vector.

[0049] In another embodiment of the present disclosure, a wind turbine gearbox fault diagnosis system is provided, the diagnosis system comprising:

[0050] Signal acquisition unit, used to collect vibration signals of key components in the wind turbine gearbox;

[0051] The signal modal decomposition unit is used to initialize the parameters of the variational modal decomposition algorithm, use the variational modal decomposition algorithm to perform modal decomposition on the vibration signals of key components to obtain multiple IMF components, use the multi-scale linear fitting method to analyze each IMF component, evaluate the linear characteristics of each IMF component, and determine the preset K value of the decomposition according to the change of the linear characteristics;

[0052] A decomposition mode number optimization unit is used to calculate the sparsity corresponding to each K value based on a value interval, and take the K value corresponding to the maximum sparsity as the optimal decomposition mode number, and the value interval is between 2 and a preset K value;

[0053] A fault feature extraction unit is used to extract energy entropy features of each modal IMF component of the optimal decomposition modal number by using a wavelet packet analysis method to obtain a fault feature vector of the gearbox vibration signal;

[0054] The fault diagnosis unit is used to construct a fault feature data set, input it into SVM for training and identification, and realize fault diagnosis of key components in the wind turbine gearbox.

[0055] As a further limitation of the technical solution of the present invention, the signal mode decomposition unit includes:

[0056] The signal decomposition module is used to perform VMD decomposition on the wind turbine gearbox vibration signal according to the initialization parameters to obtain k IMF original components;

[0057] The signal analysis module is used to solve the analytical signal of each IMF original component through Hilbert transform to obtain a single-sided spectrum;

[0058] The signal modulation module is used to modulate the single-side frequency to the corresponding baseband according to the mixed estimated center frequency of each modal analysis signal, so as to obtain the modulated signal:

[0059] Model construction module, used to calculate the norm square of the modulated signal gradient, estimate the bandwidth of each IMF component, and construct a constrained variational model;

[0060] The model solving module is used to solve the variational model by introducing penalty parameters and Lagrange multipliers to obtain the frequency domain expression of the original component of the IMF, the update formula of the center frequency, and the update formula of the penalty factor;

[0061] A parameter updating module is used to obtain an updated frequency domain value, an updated center frequency and an updated Lagrange multiplier according to the frequency domain expression of the original component of the IMF, the update formula of the center frequency and the update formula of the penalty factor;

[0062] The component conversion module is used to convert the k IMF original components into amplitude modulation-frequency modulation signals according to the updated frequency domain value, the updated center frequency and the updated Lagrange multiplier to obtain k IMF components.

[0063] In a third aspect of the present invention, a diagnostic device is provided. The diagnostic device includes a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the wind turbine gearbox fault diagnosis method provided in the first aspect is implemented.

[0064] Compared with the prior art, the wind turbine gearbox fault diagnosis method and system of the present invention have the following beneficial effects:

[0065] First, the present invention uses a variational modal decomposition algorithm to perform modal decomposition on the vibration signal of key components to obtain a solution of multiple IMF components. Through Hilbert transform, each IMF component can be converted into an analytical signal, and calculating its single-sided spectrum helps to reveal the instantaneous frequency characteristics of the signal and provide richer frequency domain information for fault diagnosis; by modulating the mixed estimated center frequency of each modal analytical signal, the single-sided frequency can be adjusted to the corresponding baseband, thereby improving the accuracy of signal processing; by calculating the norm square of the modulated signal gradient to estimate the broadband of each IMF component, the frequency variation range of the signal can be effectively quantified;

[0066] Second, the present invention automatically determines the modal number by solving the variational problem, so that the modal decomposition algorithm has good adaptability; wherein, each IMF component is analyzed by the multi-scale straight line fitting method, the linear characteristics of each component can be evaluated, and the preset K value of the decomposition is determined according to the change of the linear characteristics, and the appropriate modal number is selected to further ensure the effectiveness and accuracy of the signal decomposition; the wavelet packet analysis method is used to extract the energy entropy characteristics of each modal IMF component of the optimal decomposition modal number, and the fault feature vector of the gearbox vibration signal is obtained through the energy entropy feature extraction, and then the fault characteristics can be effectively extracted from the complex vibration signal, thereby improving the accuracy of fault diagnosis, and solving the problems of modal aliasing and endpoint effect existing in the traditional EMD method, which will lead to insufficient effective signal components in the decomposed IMF components. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention.

[0068] Figure 1 This is a flow chart of a wind turbine gearbox fault diagnosis method according to the present invention;

[0069] Figure 2 A sub-flow chart of a wind turbine gearbox fault diagnosis method according to the present invention;

[0070] Figure 3 A structural block diagram of a wind turbine gearbox fault diagnosis system provided by the present invention;

[0071] Figure 4 This is a structural block diagram of the signal modal decomposition unit provided by the present invention. DETAILED DESCRIPTION

[0072] In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0073] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" in the specification and claims of this application and the above-mentioned figure descriptions and any variations thereof are intended to cover non-exclusive inclusions.

[0074] In the description of the embodiments of the present application, the technical terms "first", "second", etc. are only used to distinguish different objects, and cannot be understood as indicating or implying relative importance or implicitly indicating the number, specific order or primary and secondary relationship of the indicated technical features. In the description of the embodiments of the present application, the meaning of "multiple" is more than two, unless otherwise clearly and specifically defined.

[0075] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0076] At present, the rolling bearing vibration signal is subjected to EMD, and multiple entropy values ​​of the intrinsic mode function (IMF) components are extracted and fused before inputting into the support vector machine for fault diagnosis. Another technology performs EMD denoising on the collected rolling bearing fault signal, and uses spectral kurtosis to extract features from the denoised signal to achieve fault diagnosis. Problems such as modal aliasing and endpoint effect in the above EMD method will lead to insufficient effective signal components in the decomposed IMF components, which will in turn affect the extraction of fault features and easily lead to poor accuracy of gearbox fault diagnosis.

[0077] To solve the above problems, the embodiments of the present disclosure provide a wind turbine gearbox fault diagnosis method and system. In the wind turbine gearbox fault diagnosis method provided by the present invention, a variational modal decomposition algorithm is used to perform modal decomposition on the vibration signal of the key components to obtain multiple IMF components, and a multi-scale linear fitting method is used to analyze each IMF component, and the linear characteristics of each IMF component are evaluated, and the preset K value of the decomposition is determined according to the change of the linear characteristics; the sparsity corresponding to each K value is calculated based on the value interval, and the K value corresponding to the maximum sparsity is taken as the optimal decomposition modal number, and the wavelet packet analysis method is used to extract the energy entropy characteristics of each modal IMF component of the optimal decomposition modal number to obtain the fault feature vector of the gearbox vibration signal. The fault feature vector extracted in this way can fully reflect the true state of the gearbox vibration signal and effectively improve the accuracy of gearbox fault diagnosis.

[0078] The specific implementation of the wind turbine gearbox fault diagnosis method and system of the present invention is described in detail below in conjunction with specific embodiments.

[0079] like Figure 1 As shown, in one embodiment of the present invention, a wind turbine gearbox fault diagnosis method is provided, the diagnosis method comprising the following steps:

[0080] Step S101: collecting vibration signals of key components in the gearbox of a wind turbine;

[0081] Specifically, in the process of collecting vibration signals of key components in the gearbox of a wind turbine generator system, the key components that need to be monitored in the gearbox, such as gears, bearings, shafts, etc., are first determined, because they are common sources of potential faults; according to the different transmission speeds of the gearbox, the gearbox can be divided into three states: low-speed stage, intermediate stage, and high-speed stage; wherein, in order to improve the accuracy of obtaining vibration signal data;

[0082] Furthermore, the present invention installs vibration sensors, such as accelerometers, speedometers or displacement meters, in three states respectively, and arranges them at key positions of the gearbox to capture the vibration response of components; in order to comprehensively monitor the vibration signal, two vibration sensors at 90 degrees to each other can be installed at the same level to monitor the axial vibration signal and radial vibration signal of the gearbox respectively.

[0083] Please continue reading Figure 1 The wind turbine gearbox fault diagnosis method of the present invention further includes:

[0084] Step S102: Initialize the parameters of the variational modal decomposition algorithm, use the variational modal decomposition algorithm to perform modal decomposition on the vibration signal of the key components to obtain multiple IMF components, use the multi-scale linear fitting method to analyze each IMF component, evaluate the linear characteristics of each IMF component, and determine the preset K value of the decomposition according to the change of the linear characteristics;

[0085] Among them, in the step of initializing the parameters of the variational mode decomposition algorithm of the present invention, the parameters include a penalty factor for controlling the bandwidth of the mode, a larger Values ​​of will result in narrower modal bandwidths, The selection of the value usually needs to be determined by experiments or optimization algorithms to ensure that the mode has both an appropriate bandwidth and does not overfit the signal. The parameter also includes a Lagrange multiplier, which is used to introduce constraints in the optimization problem. The update factor of the Lagrange multiplier of the present invention is set to 0.01.

[0086] Please continue reading Figure 1 The wind turbine gearbox fault diagnosis method of the present invention further includes:

[0087] Step S103: Calculate the sparsity corresponding to each K value based on the value interval, and take the K value corresponding to the maximum sparsity as the optimal decomposition mode number, wherein the value interval is between 2 and the preset K value;

[0088] Step S104: using wavelet packet analysis to extract energy entropy features of each modal IMF component of the optimal decomposition modal number, and obtaining a fault feature vector of the gearbox vibration signal;

[0089] Step S105: construct a fault feature data set and input it into SVM for training and identification to achieve fault diagnosis of key components in the wind turbine gearbox.

[0090] Optionally, in the embodiment of the present disclosure, sensors are installed to collect vibration signals, temperature, pressure and other data of the gearbox; the collected data is cleaned to remove noise and outliers, and normalized or standardized.

[0091] In an embodiment of the present disclosure, a support vector machine (SVM) algorithm is used to perform fault diagnosis on key components in a wind turbine gearbox; wherein, in the construction of a fault feature data set, it is necessary to collect operating data of the wind turbine gearbox, including but not limited to parameters such as vibration signals, temperature, and pressure; by preprocessing and extracting features from these data, a fault feature data set for training an SVM model can be obtained; during training, the data set is divided into a training set and a test set, and the SVM model is then trained using the training set and the test set.

[0092] Furthermore, in the disclosed embodiment, before using the above-mentioned fault feature data set to train the SVM model, the data is labeled according to the actual detection situation, such as normal, worn, broken, etc.; the SVM model distinguishes different categories based on finding the optimal hyperplane, i.e., the normal state and other fault states of the gearbox.

[0093] Please refer to Figure 2 In a preferred embodiment of the present disclosure, the present invention uses a variational modal decomposition algorithm to perform modal decomposition on the vibration signal of a key component, comprising:

[0094] Step S1021: performing VMD decomposition on the wind turbine gearbox vibration signal according to the initialization parameters to obtain k IMF original components;

[0095] Among them, in the embodiment of the present disclosure, VMD decomposition is performed on the vibration signal of the wind turbine gearbox, and the original signal is decomposed into intrinsic mode functions IMFs of limited bandwidth with specific center frequencies. These IMF components can be used for signal feature extraction and noise reduction processing after analysis and reconstruction;

[0096] The step S1022 is also included: solving the analytical signal of each IMF original component by Hilbert transform to obtain a single-sided spectrum:

[0097] (1);

[0098] (2);

[0099] In formula (1), represents the modal function, k is a positive integer, k=1,2,...,K; K represents the number of modes to be decomposed, K is also a positive integer;

[0100] The Hilbert transform of the embodiment of the present disclosure converts the IMF component into an analytical signal, so that it can be represented on the complex plane, so that the instantaneous frequency and amplitude at each moment can be calculated; the analytical signal obtained by the Hilbert transform of the embodiment of the present disclosure only contains the positive frequency part, thereby avoiding frequency aliasing and being able to observe and analyze the spectrum more intuitively;

[0101] When processing non-stationary signals in the gearbox fault diagnosis of the present invention, the single-sided spectrum obtained by Hilbert transform can more clearly display the frequency components and energy distribution of the signal than the double-sided spectrum of the traditional Fourier transform, and the advantage is more obvious.

[0102] Furthermore, the present invention uses a variational modal decomposition algorithm to perform modal decomposition on the vibration signal of the key component, and further includes step S1023: after obtaining the single-sided spectrum, modulate the single-sided frequency to the corresponding baseband according to the mixed estimated center frequency of each modal analysis signal, so as to obtain a modulated signal:

[0103] (3);

[0104] In formula (3), represents the exponential signal, represents the center frequency;

[0105] In the disclosed embodiment, modulation is performed according to the mixed estimated center frequency of each modal analysis signal, ensuring the accurate correspondence of the signal in the frequency domain, thereby avoiding possible frequency aliasing problems; therefore, in the embodiment of the present invention, VMD and Hilbert transform are combined to adaptively process various non-stationary and nonlinear signals;

[0106] Furthermore, the method further includes step S1024: calculating the square norm of the modulated signal gradient, estimating the bandwidth of each IMF component, and constructing a constrained variational model, wherein the constrained variational model is expressed as follows:

[0107] (4);

[0108] In formula (4), represents the unit pulse function, j represents the imaginary unit, represents the convolution operation, It means to find the partial derivative of the function; represents the k IMF components after decomposition, ; represents the center frequency of each component, where .

[0109] Step S1025: solving the variational model by introducing penalty parameters and Lagrange multipliers to obtain the frequency domain expression of the original component of the IMF, the update formula of the center frequency, and the update formula of the penalty factor;

[0110] Step S1026: according to the frequency domain expression of the original IMF component, the update formula of the center frequency, and the update formula of the penalty factor, an updated frequency domain value, an updated center frequency, and an updated Lagrange multiplier are obtained;

[0111] Step S1027: According to the updated frequency domain value, the updated center frequency and the updated Lagrange multiplier, the k IMF original components are converted into amplitude modulation-frequency modulation signals to obtain k IMF components.

[0112] In the disclosed embodiments, the present invention can more accurately control the bandwidth of each modal component in the variational model by introducing a penalty parameter, thereby obtaining a more accurate frequency resolution; further, the present invention utilizes Lagrange multipliers to adaptively find the optimal center frequency of each mode, thereby effectively improving the accuracy of decomposition.

[0113] It can be understood that the VMD algorithm of the embodiment of the present disclosure shows a high degree of robustness in sampling and noise, and can effectively process noisy signals; and the VMD algorithm of the embodiment of the present disclosure overcomes the endpoint effect problem in the EMD method and improves the stability of signal processing; in addition, the VMD algorithm of the present disclosure is suitable for sequences of non-stationary signals, and can be decomposed into relatively stable subsequences containing multiple different frequency scales.

[0114] Further, in an embodiment of the present invention, in the step of calculating the sparsity corresponding to each K value, the sparsity of each IMF component is expressed as:

[0115] (5);

[0116] In formula (5), represents the energy weight factor, which is defined as the ratio of the maximum amplitude of the unilateral spectrum of the IMF component to the maximum amplitude of the marginal spectra of the K components;

[0117] Furthermore, the present invention takes the unilateral spectrum sparseness of each IMF as the overall sparsity under the decomposition mode number K:

[0118] (6);

[0119] when When , the optimal decomposition mode number is determined to be K-1.

[0120] Furthermore, in the step of analyzing each IMF component using a multi-scale straight line fitting method, evaluating the linear characteristics of each IMF component, and determining the preset K value of the adaptive decomposition according to the change of the linear characteristics, for each IMF component, a multi-segment fitting straight line is used to replace it according to the initial scale, wherein the calculation formula for the selection of the initial scale is as follows: ; Divide each IMF component into m segments, For each component sequence, the least squares method is used to fit the straight line to minimize the residual sum of squares (RSS). The slope corresponding to the fitting straight line of each component sequence is obtained and the slopes are compared. If the slope characteristics of adjacent components are similar, the adjacent components are merged into a wider mode. If the slope characteristics of adjacent components are significantly different, the adjacent components are split into multiple modes until the slope characteristics of the components corresponding to the split modes are similar. The splitting is stopped to obtain the preset K value.

[0121] The embodiment of the present disclosure uses the least squares method to perform straight line fitting, and can accurately calculate the slope of each component sequence, thereby ensuring the accuracy of signal decomposition; the least squares method is used to optimize the fitting effect by minimizing the residual square sum, effectively controlling the error and improving the overall signal processing accuracy;

[0122] In addition, the embodiments of the present disclosure can adaptively merge or split modes according to the similarity or difference of slope features, thereby more accurately reflecting the characteristics of the signal. By comparing the slope features, the number of modal decompositions can be adaptively determined, avoiding waste of resources.

[0123] Further, in the step of extracting energy entropy features from each modal IMF component of the optimal decomposition modal number using the wavelet packet analysis method to obtain the fault feature vector of the gearbox vibration signal, the process of solving the wavelet packet energy entropy of the modal IMF component signal is described as follows:

[0124] First, the modal IMF component is decomposed by wavelet packets to obtain sub-signals of multiple different frequency bands. Further, each obtained frequency band is reconstructed by wavelet packets, and then the wavelet packet energy value of each frequency band is calculated. Thereafter, the wavelet packet energy value of each frequency band is normalized, and the wavelet packet energy entropy is calculated to obtain the fault feature vector of the gearbox vibration signal. Among them, the present invention can accurately decompose the signal into sub-signals of different frequency bands through VMD algorithm and wavelet packet decomposition. In addition, the wavelet packet energy value of each frequency band is calculated and normalized to generate high-dimensional feature vectors, which can more comprehensively reflect the characteristics of the signal, thereby improving the accuracy of fault diagnosis.

[0125] Furthermore, in the step of constructing a fault feature data set, inputting it into SVM for training and identification, and realizing fault diagnosis of key components in the gearbox of a wind turbine, since the fault feature vectors contained in the constructed fault feature data set are indivisible samples, a kernel function is used to map the fault feature data set to a high-dimensional space;

[0126] The kernel function in the embodiment of the present invention is expressed as:

[0127] (7);

[0128] In formula (7), Represents the kernel function, which is used to calculate the two samples in the fault feature dataset , similarities between; Represents a mapping function, which is used to transform samples of the fault feature dataset , Mapped to high-dimensional space; T represents vector transpose operation;

[0129] Furthermore, a hyperplane is found in the high-dimensional space so that the distance between each data point and the hyperplane is maximized. The hyperplane prediction model constructed is expressed as:

[0130] (8);

[0131] In formula (8), represents the decision function of the SVM model, which is used to predict the category of new sample signal data of gearbox fault; represents a mapping function; represents the weight vector; b represents the bias term, which is used to control the position of the hyperplane along the normal direction, and T represents the vector transpose operation.

[0132] Furthermore, in the disclosed embodiment, the model is used to perform fault diagnosis and identification on the newly input gearbox vibration signal fault feature vector.

[0133] like Figure 3 As shown, in another embodiment of the present disclosure, a wind turbine gearbox fault diagnosis system is provided, and the wind turbine gearbox fault diagnosis system includes:

[0134] The signal acquisition unit 201 is used to collect vibration signals of key components in the gearbox of the wind turbine;

[0135] The signal modal decomposition unit 202 is used to initialize the parameters of the variational modal decomposition algorithm, use the variational modal decomposition algorithm to perform modal decomposition on the vibration signal of the key component to obtain multiple IMF components, use the multi-scale linear fitting method to analyze each IMF component, evaluate the linear characteristics of each IMF component, and determine the preset K value of the decomposition according to the change of the linear characteristics;

[0136] The decomposition mode number optimization unit 203 is used to calculate the sparsity corresponding to each K value based on the value interval, and take the K value corresponding to the maximum sparsity as the optimal decomposition mode number, and the value interval is between 2 and the preset K value;

[0137] The fault feature extraction unit 204 is used to extract the energy entropy feature of each modal IMF component of the optimal decomposition modal number by using the wavelet packet analysis method to obtain the fault feature vector of the gearbox vibration signal;

[0138] The fault diagnosis unit 205 is used to construct a fault feature data set and input it into the SVM for training and identification to achieve fault diagnosis of key components in the wind turbine gearbox.

[0139] Further, such as Figure 4 As shown, the signal mode decomposition unit 202 provided in the embodiment of the present disclosure includes:

[0140] The signal decomposition module 2021 is used to perform VMD decomposition on the wind turbine gearbox vibration signal according to the initialization parameters to obtain k IMF original components;

[0141] The signal analysis module 2022 is used to solve the analytical signal of each IMF original component through Hilbert transform to obtain a single-sided spectrum;

[0142] The signal modulation module 2023 is used to modulate the single-side frequency to the corresponding baseband according to the mixed estimated center frequency of each modal analysis signal, so as to obtain a modulated signal:

[0143] A model construction module 2024 is used to calculate the norm square of the modulated signal gradient, estimate the bandwidth of each IMF component, and construct a constrained variational model;

[0144] The model solving module 2025 is used to solve the variational model by introducing penalty parameters and Lagrange multipliers to obtain the frequency domain expression of the original component of the IMF, the update formula of the center frequency, and the update formula of the penalty factor;

[0145] A parameter updating module 2026 is used to obtain an updated frequency domain value, an updated center frequency and an updated Lagrange multiplier according to the frequency domain expression of the original IMF component, the update formula of the center frequency and the update formula of the penalty factor;

[0146] The component conversion module 2027 is used to convert the k IMF original components into amplitude modulation-frequency modulation signals according to the updated frequency domain value, the updated center frequency and the updated Lagrange multiplier to obtain k IMF components.

[0147] In another embodiment of the present disclosure, a diagnostic device is provided. The diagnostic device of the embodiment of the present disclosure includes a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the wind turbine gearbox fault diagnosis method of any of the above embodiments is implemented. The diagnostic device may also include: an input system and an output system. The processor, memory, input system and output system may be connected via a bus or other means. The input system may receive input digital or character information and generate signal input related to the migration of wind turbine gearbox fault diagnosis. The output system may include display devices such as display screens.

[0148] Furthermore, the present invention also provides a computer-readable storage medium, on which computer-readable instructions are stored. When the computer-readable instructions are executed by a processor, the steps of the wind turbine gearbox fault diagnosis method provided above are implemented.

[0149] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0150] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0151] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0152] In a typical configuration of an embodiment of the present invention, the terminal, the device of the service network and the computing device include one or more processors (CPU), an input / output interface, a network interface and a memory.

[0153] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0154] Computer readable media include permanent and non-permanent, removable and non-removable media that can store information by any method or technology. The information can be computer readable instructions, data structures, program modules or other data.

[0155] Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include non-transitory media such as modulated data signals and carrier waves.

[0156] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example for illustration. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. The specific working process of the device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0157] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some or all of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention.

[0158] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A wind turbine gearbox fault diagnosis method, characterized in that: The following steps are involved: Collect vibration signals of key components in wind turbine gearboxes; Initialize the parameters of the variational modal decomposition algorithm, use the variational modal decomposition algorithm to perform modal decomposition on the vibration signals of key components to obtain multiple IMF components, use the multi-scale linear fitting method to analyze each IMF component, evaluate the linear characteristics of each IMF component, and determine the preset K value of the decomposition according to the changes in the linear characteristics; Calculate the sparsity corresponding to each K value based on the value interval, and take the K value corresponding to the maximum sparsity as the optimal decomposition mode number, where the value interval is between 2 and the preset K value; The wavelet packet analysis method is used to extract the energy entropy characteristics of each modal IMF component of the optimal decomposition modal number, and the fault feature vector of the gearbox vibration signal is obtained; A fault feature dataset was constructed and input into SVM for training and identification to perform fault diagnosis on key components in the wind turbine gearbox.

2. A wind turbine gearbox fault diagnosis method according to claim 1, characterized in that: The step of using a variational modal decomposition algorithm to perform modal decomposition on the vibration signal of the key component to obtain multiple IMF components includes: Perform VMD decomposition on the wind turbine gearbox vibration signal according to the initialization parameters to obtain k IMF original components; The analytical signal of each IMF original component is solved by Hilbert transform to obtain a one-sided spectrum: （1）； （2）； In formula (1), represents the modal function, k is a positive integer, k=1,2,...,K; K represents the number of modes to be decomposed; According to the mixed estimated center frequency of each modal analysis signal, the single-side frequency is modulated to the corresponding baseband to obtain the modulated signal: （3）； In formula (3), represents the exponential signal, represents the center frequency, represents the unit pulse function, j represents the imaginary unit, Represents the convolution operation; Calculate the square norm of the modulated signal gradient, estimate the bandwidth of each IMF component, and construct a constrained variational model; By introducing penalty parameters and Lagrange multipliers to solve the variational model, the frequency domain expression of the original component of the IMF, the update formula of the center frequency, and the update formula of the penalty factor are obtained; According to the frequency domain expression of the original component of the IMF, the update formula of the center frequency, and the update formula of the penalty factor, the updated frequency domain value, the updated center frequency, and the updated Lagrange multiplier are obtained; According to the updated frequency domain value, the updated center frequency and the updated Lagrange multiplier, the k IMF original components are converted into amplitude modulation-frequency modulation signals to obtain k IMF components.

3. A wind turbine gearbox fault diagnosis method according to claim 2, characterized in that: In the step of constructing a constrained variational model, the constrained variational model is expressed as follows: （4）； In formula (4), It means to find the partial derivative of the function; represents the k IMF components after decomposition, ; represents the center frequency of each component, where .

4. A wind turbine gearbox fault diagnosis method according to claim 3, characterized in that: In the step of calculating the sparsity corresponding to each K value, the sparsity of each IMF component is expressed as: （5）； In formula (5), represents the energy weight factor, which is defined as the ratio of the maximum amplitude of the unilateral spectrum of the IMF component to the maximum amplitude of the marginal spectra of the K components; The unilateral spectrum sparseness of each IMF is taken as the overall sparsity under the decomposition mode number K: （6）； when When K>2, the optimal decomposition mode number is determined to be K-1.

5. A wind turbine gearbox fault diagnosis method according to claim 4, characterized in that: In the step of analyzing each IMF component using the multi-scale straight line fitting method, evaluating the linear characteristics of each IMF component, and determining the preset K value of the adaptive decomposition according to the change of the linear characteristics: For each IMF component, multiple segments of fitting straight lines are used to replace it according to the initial scale. For the selection of the initial scale, the calculation formula is as follows: ; Divide each IMF component into m segments, ; For each component sequence, the least square method is used to fit the straight line so that the residual sum of squares (RSS) is minimized. The slope corresponding to the fitting straight line of each component sequence is calculated and the slopes are compared. According to the similarity or difference of the slope characteristics, the modes are merged or split to obtain the preset K value.

6. A wind turbine gearbox fault diagnosis method according to claim 5, characterized in that: The process of solving the wavelet packet energy entropy of the modal IMF component signal includes: Firstly, the modal IMF components are decomposed by wavelet packets to obtain multiple sub-signals in different frequency bands; Perform wavelet packet reconstruction on each obtained frequency band, and then calculate the wavelet packet energy value of each frequency band; The wavelet packet energy value of each frequency band is normalized, the wavelet packet energy entropy is calculated, and the fault feature vector of the gearbox vibration signal is obtained.

7. A wind turbine gearbox fault diagnosis method according to claim 6, characterized in that: In the steps of constructing a fault feature data set, inputting it into SVM for training and identification, and realizing fault diagnosis of key components in the wind turbine gearbox, a kernel function is used to map the fault feature data set to a high-dimensional space; Among them, the kernel function is expressed as: （7）； In formula (7), Represents the kernel function, which is used to calculate the two samples in the fault feature dataset , similarities between; Represents a mapping function, which is used to transform samples of the fault feature dataset , Mapped to high-dimensional space; T represents vector transpose operation; Find a hyperplane in high-dimensional space so that the distance between each data point and the hyperplane is the largest. The hyperplane prediction model constructed is expressed as: （8）； In formula (8), represents the decision function of the SVM model, which is used to predict the category of new sample signal data of gearbox fault; represents a mapping function; represents the weight vector; b represents the bias term, which is used to control the position of the hyperplane along the normal direction; T represents the vector transposition operation; The model is used to perform fault diagnosis and identification on the newly input gearbox vibration signal fault feature vector.

8. A fault diagnosis system for implementing a wind turbine gearbox fault diagnosis method as claimed in any one of claims 1 to 7, characterized in that: The fault diagnosis system comprises: Signal acquisition unit, used to collect vibration signals of key components in the wind turbine gearbox; The signal modal decomposition unit is used to initialize the parameters of the variational modal decomposition algorithm, use the variational modal decomposition algorithm to perform modal decomposition on the vibration signals of key components to obtain multiple IMF components, use the multi-scale linear fitting method to analyze each IMF component, evaluate the linear characteristics of each IMF component, and determine the preset K value of the decomposition according to the change of the linear characteristics; A decomposition mode number optimization unit is used to calculate the sparsity corresponding to each K value based on a value interval, and take the K value corresponding to the maximum sparsity as the optimal decomposition mode number, and the value interval is between 2 and a preset K value; A fault feature extraction unit is used to extract energy entropy features of each modal IMF component of the optimal decomposition modal number by using a wavelet packet analysis method to obtain a fault feature vector of the gearbox vibration signal; The fault diagnosis unit is used to construct a fault feature data set, input it into SVM for training and identification, and realize fault diagnosis of key components in the wind turbine gearbox.

9. A diagnostic device, characterized in that: The diagnostic device comprises a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, a wind turbine gearbox fault diagnosis method according to any one of claims 1 to 7 is implemented.

Images