Wind driven generator bearing quantitative diagnosis method and system based on CPO-VMD
By using the crown porcupine optimization algorithm to optimize the parameters of the variational mode decomposition algorithm in wind turbine bearings, the problem of low diagnostic accuracy of wind turbine bearings is solved, and the accurate extraction and diagnosis of characteristic signals of weak defects and faults is achieved, which improves the Diagnostic efficiency and accuracy.
Patent Information
- Application Number
- CN202410977701.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-08
AI Technical Summary
In the health assessment of wind turbine bearings, the parameter setting of the variational modal decomposition algorithm is difficult to meet high-precision diagnosis, resulting in the inability to accurately extract the characteristic signals of weak defects and faults, affecting the diagnostic accuracy and efficiency.
The parameters of the variational modal decomposition algorithm (VMD) are optimized by using the crown porcupine optimization algorithm (CPO), and the CPO-VMD method is constructed. The wind turbine bearings are extracted through standardization and optimization of the number of modal components and punishment factors. Frequency signal of weak defect fault characteristics.
It improves the accuracy and efficiency of quantitative diagnosis of wind turbine bearings, reduces the adverse consequences caused by artificially set parameters, and can more accurately detect weak defect signals to ensure the safe service of wind turbines.
Smart Images

Figure CN118913691A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of mechanical fault diagnosis, and in particular to a CPO-VMD-based quantitative diagnosis method and system for a wind turbine bearing. Background Art
[0002] Under the interaction of simple harmonic vibration and random vibration generated when the wind turbine is working, the vibration fault signal mechanism of the wind turbine bearing becomes extremely complicated, which makes it difficult to conduct an effective health assessment of the wind turbine bearing, unable to accurately judge potential faults in a timely manner, and unable to ensure the safe service status of the wind turbine. The number of modal decompositions and penalty factors in the parameters of variational modal decomposition in the related technology must be pre-set based on experience. The selection of the two parameters determines whether the fault information can be accurately extracted. In order to reduce the adverse consequences brought about by artificially setting parameters, the setting of the parameters of variational modal decomposition in the related technology is difficult to meet the requirements of quantitative high-precision diagnosis of wind turbine bearings. Summary of the invention
[0003] In order to solve the above technical problems, the purpose of the present invention is to provide a quantitative diagnosis method and system for wind turbine bearings based on CPO-VMD, which can extract accurate acoustic emission signals and vibration signals of characteristic frequencies of weak defect faults, thereby improving the accuracy and efficiency of quantitative diagnosis of wind turbine bearings.
[0004] The first technical solution adopted by the present invention is: a quantitative diagnosis method for wind turbine bearings based on CPO-VMD, comprising the following steps:
[0005] Acquire the fault data of the wind turbine bearing and perform standardization processing to obtain the standardized bearing fault data, wherein the bearing fault data includes a defect acoustic emission signal and a vibration characteristic signal;
[0006] The parameters of the variational mode decomposition (VMD) algorithm were optimized by using the crown porcupine optimization algorithm (CPO), and a quantitative detection algorithm for wind turbine bearings based on CPO-VMD was constructed.
[0007] The wind turbine bearing quantitative detection algorithm of CPO-VMD is used to perform feature detection on the standardized bearing fault data to obtain the quantitative diagnosis results of the wind turbine bearing.
[0008] Furthermore, the expression of the standardization process is specifically as follows:
[0009]
[0010] In the above formula, X′ z represents the normalized bearing failure data, X zrepresents the data of bearing fault, N represents the number of fault data, μ and σ represent the standardized parameters, μ is the mean value, and σ is the standard deviation value.
[0011] Furthermore, the step of optimizing the parameters of the variational mode decomposition algorithm by using the crown porcupine optimization algorithm to construct a quantitative detection algorithm for wind turbine bearings based on CPO-VMD specifically includes:
[0012] Based on the Hilbert spectral function, the parameters of the variational mode decomposition algorithm are optimized by the crested porcupine optimization algorithm to obtain the optimal modal component value and the optimal penalty factor value;
[0013] The optimized modal component values and the optimized penalty factor values are embedded in the variational mode decomposition algorithm, and combined with the crested porcupine optimization algorithm, a quantitative detection algorithm for wind turbine bearings based on CPO-VMD is obtained.
[0014] Further, the step of optimizing the parameters of the variational modal decomposition algorithm based on the Hilbert spectral function by using the crown porcupine optimization algorithm to obtain the optimal modal component value and the optimal penalty factor value specifically includes:
[0015] Initialize the crested porcupine population, set the value range of the modal component and the value range of the penalty factor, and set the crested porcupine distance threshold;
[0016] The initialized crested porcupine population is activated by a cyclic population reduction method to obtain an activated crested porcupine population;
[0017] If the distance between the predator and the crested porcupine is greater than the crested porcupine distance threshold, the activated crested porcupine population performs a global search process through the first defense strategy to obtain a global search defense factor;
[0018] If the distance between the predator and the crested porcupine is less than the crested porcupine distance threshold, the activated crested porcupine population performs local search processing through the second defense strategy to obtain a local search defense factor;
[0019] Based on the global search defense factor and the local search defense factor, the envelope entropy value of the parameters of the variational mode decomposition algorithm is calculated through the variational mode update algorithm, and the optimal modal component value and the optimal penalty factor value are determined according to the calculation results.
[0020] Furthermore, the first defense strategy includes a crested porcupine visual exploration strategy and a crested porcupine sound exploration strategy, wherein the crested porcupine visual exploration strategy is specifically expressed as follows:
[0021]
[0022] In the above formula, represents the optimal solution for the function evaluation t, represents the vector generated between the current crested porcupine and the random crested porcupine, represents the position of the predator at iteration t, τ1 represents a random number based on a normal distribution, τ2 represents a random value in the interval [0, 1], represents the position of the i-th individual at iteration t, represents the position of the i-th individual at iteration t+1;
[0023] The expression of the crested porcupine sound exploration strategy is specifically as follows:
[0024]
[0025] In the above formula, r1 and r2 represent two random integers between [1, N], τ3 represents a random value generated between 0 and 1, and the randomly generated binary vector containing 0 and 1 is To cover all probabilities, represents the predator position, represents the position of the r1th element in the tth iteration, Indicates the position of the r2th element in the tth iteration.
[0026] Furthermore, the second defense strategy includes a crested porcupine odor attack strategy and a crested porcupine physical attack strategy, wherein the crested porcupine odor attack strategy is specifically expressed as follows:
[0027]
[0028] In the above formula, τ3 represents a random number between the interval [1, N], δ represents the parameter controlling the search direction, and γ t represents the defense factor, represents the odor diffusion factor, Indicates the situation where the strategy is simulated. represents the position of the r1th element in the tth iteration, represents the position of the r2th element in the tth iteration, r1 and r2 represent two random integers between [1, N];
[0029] The expression of the crested porcupine physical attack strategy is specifically as follows:
[0030]
[0031] In the above formula, represents the optimal solution, α represents the convergence speed factor, represents the position of the i-th individual during iteration, τ4 and τ5 represent random values in the interval [0,1], represents the average force exerted by the crested porcupine on the i-th predator.
[0032] Furthermore, the expression of the variational mode update algorithm is specifically as follows:
[0033]
[0034] In the above formula, The Wiener filter represents the residual quantity, represents the Fourier transform of the signal f(t), represents the ith modal update expression, represents the ith modal expression, represents the modal bandwidth corresponding to frequency ω, represents the center frequency of the kth mode in the nth iteration process, τ represents the fidelity coefficient, n represents an integer, k represents the initial mode number, i represents different modes or components, and ω represents the angular frequency.
[0035] Further, the step of performing feature detection processing on the standardized bearing fault data by using the CPO-VMD wind turbine bearing quantitative detection algorithm to obtain the wind turbine bearing quantitative diagnosis result specifically includes:
[0036] The standardized bearing fault data is modally decomposed by using the CPO-VMD wind turbine bearing quantitative detection algorithm to obtain the bearing fault data modal components.
[0037] The kurtosis value is set to filter the modal components of the bearing fault data to obtain the filtered modal components of the bearing fault data;
[0038] The modal components of the screened bearing fault data are reconstructed and calculated to obtain the quantitative diagnosis results of wind turbine bearings.
[0039] The second technical solution adopted by the present invention is: a wind turbine bearing quantitative diagnosis system based on CPO-VMD, comprising:
[0040] A standardization module is used to obtain the fault data of the wind turbine bearing and perform standardization processing to obtain the standardized bearing fault data, wherein the bearing fault data includes a defect acoustic emission signal and a vibration characteristic signal;
[0041] The optimization module is used to optimize the parameters of the variational mode decomposition algorithm (VMD) through the crown porcupine optimization algorithm (CPO) and build a quantitative detection algorithm for wind turbine bearings based on CPO-VMD;
[0042] The diagnosis module is used to perform feature detection processing on the standardized bearing fault data through the CPO-VMD wind turbine bearing quantitative detection algorithm to obtain the wind turbine bearing quantitative diagnosis result.
[0043] The beneficial effects of the method and system of the present invention are as follows: the present invention obtains fault data of wind turbine bearings and performs standardized processing, further optimizes the parameters of a variational mode decomposition algorithm (VMD) through a crown porcupine optimization algorithm (CPO), and optimizes the parameters such as the number of decompositions and the penalty factor of the VMD algorithm using a global optimization algorithm to make it suitable for the characteristics of weak defect acoustic emission signals and vibration signals of a wind turbine transmission system, and then uses the optimized VMD algorithm to decompose the acoustic emission signals and vibration signals generated by the weak defects, extracts the acoustic emission signals and vibration signals that can represent the characteristic frequencies of the weak defect faults, and performs detection, thereby improving the accuracy and efficiency of quantitative diagnosis of wind turbine bearings. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] Figure 1 It is a flow chart of steps of a quantitative diagnosis method for wind turbine bearings based on CPO-VMD of the present invention;
[0045] Figure 2 It is a structural block diagram of a wind turbine bearing quantitative diagnosis system based on CPO-VMD of the present invention;
[0046] Figure 3 It is a schematic diagram of the implementation process of the quantitative diagnosis of the wind turbine bearing provided by the specific embodiment of the present invention;
[0047] Figure 4 It is a schematic diagram of the process flow of variational mode decomposition algorithm optimization provided by a specific embodiment of the present invention. DETAILED DESCRIPTION
[0048] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The step numbers in the following embodiments are only provided for the convenience of explanation and description, and the order between the steps is not limited in any way. The execution order of each step in the embodiment can be adaptively adjusted according to the understanding of those skilled in the art.
[0049] Reference Figure 1 and Figure 3 The present invention provides a quantitative diagnosis method for wind turbine bearings based on CPO-VMD, the method comprising the following steps:
[0050] S100, acquiring fault data of a wind turbine bearing and performing standardization processing to obtain standardized bearing fault data, wherein the bearing fault data includes a defect acoustic emission signal and a vibration characteristic signal;
[0051] Specifically, data on bearing failures that reflect the health status of wind turbine bearings are collected. The data on the entire life cycle of the bearings collected include: acoustic emission signals and vibration signals. The data on bearing failures are standardized, and the expression is:
[0052]
[0053] In the above formula, X′ z represents the normalized bearing failure data, X z represents the data of bearing fault, N represents the number of fault data, μ and σ represent the standardized parameters, μ is the mean value, and σ is the standard deviation value.
[0054] S200, optimizing the parameters of the variational mode decomposition algorithm (VMD) by using the crown porcupine optimization algorithm (CPO), and constructing a quantitative detection algorithm for wind turbine bearings based on CPO-VMD;
[0055] Specifically, based on the Hilbert spectral function, the parameters of the variational modal decomposition algorithm are optimized by the crown porcupine optimization algorithm to obtain the optimal modal component values and the optimal penalty factor values; the optimized modal component values and the optimized penalty factor values are embedded in the variational modal decomposition algorithm, and combined with the crown porcupine optimization algorithm to obtain a quantitative detection algorithm for wind turbine bearings based on CPO-VMD.
[0056] In this embodiment, a crown porcupine optimization algorithm is established to optimize the [K, α] parameter combination of the VMD algorithm, and then the optimal parameter combination is found. The crown porcupine optimization algorithm is used to accurately obtain the optimal parameters of variational modal decomposition based on the Hilbert spectrum function, and the modal component number K value and penalty factor α value of the acoustic emission data of the wind turbine bearing fault are adaptively determined. At the same time, the crown porcupine is used to optimize VMD to decompose and process the acoustic emission signals of the inner and outer rings and cage faults of the bearing.
[0057] Further, it should be noted that the optimization process of the parameters of the variational mode decomposition algorithm is to initialize the crowned porcupine population, set the value range of the modal component value and the value range of the penalty factor value, and set the crowned porcupine distance threshold; activate the initialized crowned porcupine population through the cyclic population reduction method to obtain the activated crowned porcupine population; if the distance between the predator and the crowned porcupine is greater than the crowned porcupine distance threshold, the activated crowned porcupine population performs a global search process through the first defense strategy to obtain a global search defense factor; if the distance between the predator and the crowned porcupine is less than the crowned porcupine distance threshold, the activated crowned porcupine population performs a local search process through the second defense strategy to obtain a local search defense factor; based on the global search defense factor and the local search defense factor, the envelope entropy value of the parameters of the variational mode decomposition algorithm is calculated through the variational mode update algorithm, and the optimal modal component value and the optimal penalty factor value are determined according to the calculation results.
[0058] Specifically, the defense mechanism of the crown porcupine optimization algorithm is first explained:
[0059] Crested porcupines (CPs) use four different protection mechanisms: vision, sound, smell, and physical attack. The first and second defense strategies (vision and sound) are CPO exploration behaviors, and the third and fourth defense strategies (smell and physical attack) are CPO development behaviors. Based on this. CPO proposed a new method: cyclic population reduction technology (CPR). This strategy simulates that not all CPs activate the defense mechanism, only the threatened CPs will activate the defense mechanism, take out some CPs from the population set to speed up the convergence, and then put them back into the population to avoid falling into the local minimum. The above behavior is numerically modeled and analyzed using the formula for the activation of the crown porcupine defense mechanism, and its expression is:
[0060]
[0061] Among them, T represents the variable that determines the number of loops, t represents the current function evaluation, and T max Indicates the maximum number of function evaluations, N min Indicates the minimum number of individuals in the newly generated population.
[0062] According to the CP defense behavior, there are two defense strategies, namely visual strategy and sound strategy. These strategies investigate different areas and are committed to global exploration. The expression of the crested porcupine visual exploration strategy is as follows:
[0063]
[0064] In the above formula, represents the optimal solution for the function evaluation t, represents the vector generated between the current crested porcupine and the random crested porcupine, represents the position of the predator at iteration t, τ1 represents a random number based on a normal distribution, τ2 represents a random value in the interval [0, 1], represents the position of the i-th individual at iteration t, represents the position of the i-th individual at iteration t+1;
[0065] The formula for generating the position of the predator at iteration t is as follows:
[0066]
[0067] In the above formula, r represents a random number between [1, N].
[0068] CP uses sound methods to make noise and threaten predators. When the predator approaches, the CP's voice becomes louder. The expression of the crested porcupine's sound exploration strategy is as follows:
[0069]
[0070] In the above formula, r1 and r2 represent two random integers between [1, N], τ3 represents a random value generated between 0 and 1, and the randomly generated binary vector containing 0 and 1 is To cover all probabilities, represents the predator position, represents the position of the r1th element in the tth iteration, Indicates the position of the r2th element in the tth iteration.
[0071] According to the defensive behavior of CP, when the predator approaches, it adopts the odor attack strategy and the physical attack strategy, and conducts regional development of the two attack strategies, focusing on local development search. In the odor attack stage, CP secretes a foul odor to prevent being approached. The expression of the crested porcupine's odor attack strategy is as follows:
[0072]
[0073] In the above formula, τ3 represents a random number between the interval [1, N], δ represents the parameter controlling the search direction, and γ t represents the defense factor, represents the odor diffusion factor, Indicates the situation where the strategy is simulated. represents the position of the r1th element in the tth iteration, Represents the position of the r2th element in the tth iteration, r1 and r2 represent two random integers between [1, N].
[0074] The expression for controlling the search direction parameter vector is as follows:
[0075]
[0076] The expression of the defense factor is as follows:
[0077]
[0078] The expression of the odor diffusion factor is as follows:
[0079]
[0080] in, represents the objective function value of each individual i, ∈ is the minimum value to avoid division by zero, is a vector of randomly generated numbers between 0 and 1, rand is a variable containing randomly generated numbers between 0 and 1, N is the population size, t is the current number of iterations, and t max is the maximum number of iterations.
[0081] In the physical attack phase, when the predator is very close to it, the CP will attack it with its spikes. The expression of the crested porcupine's physical attack strategy is as follows:
[0082]
[0083] in, is the optimal solution, representing cp, α is the convergence speed factor, is the position of the i-th individual at the time of iteration, representing the predator at that position, τ4 is a random value in the interval [0, 1], is the average force of cp on the i-th predator.
[0084] In summary, if Figure 4 As shown, the parameter optimization process of the variational mode decomposition algorithm in the embodiment of the present invention is:
[0085] 1) Initialize the population, set the value range of [K, α] and CPO initial population and other parameters;
[0086] 2) Use the cyclic population reduction technique (CPR) to simulate a population of crested porcupines that is threatened and has activated its defense mechanism;
[0087] 3) Crested porcupines conduct long-distance exploration. When the distance is relatively far, the crested porcupine visual exploration strategy or the crested porcupine sound exploration strategy is activated;
[0088] 4) Crested porcupines conduct close-range exploration. When the distance is close, the crested porcupine scent attack strategy or the crested porcupine physical attack strategy is activated;
[0089] 5) VMD decomposes the original signal, calculates the envelope entropy value of each parameter combination [K, α] using the variational mode update algorithm, and records the optimal individual position;
[0090] 6) Select the iterative formula according to the convergence factor and iteratively update until the termination condition is met;
[0091] 7) Finally, output the optimal number of decompositions K and the quadratic penalty factor α.
[0092] S300, performing feature detection processing on the standardized bearing fault data through the CPO-VMD wind turbine bearing quantitative detection algorithm to obtain the wind turbine bearing quantitative diagnosis result.
[0093] Specifically, the standardized bearing fault data is modally decomposed through the CPO-VMD wind turbine bearing quantitative detection algorithm to obtain the modal components of the bearing fault data; the kurtosis value is set to filter the modal components of the bearing fault data to obtain the filtered modal components of the bearing fault data; the filtered modal components of the bearing fault data are reconstructed and calculated to obtain the quantitative diagnosis results of the wind turbine bearing.
[0094] In an embodiment of the present invention, based on the principle of fast kurtosis graph, the modal components of the bearing fault vibration signal processed by the adaptive variational modal decomposition method are calculated, which can quantitatively describe the state change of the system. According to the principle that the kurtosis value is greater than 3 in the fast kurtosis graph algorithm, the modal components containing more fault information are selected, which effectively improves the performance of the model, thereby further improving the extraction accuracy of the wind turbine bearing fault characteristics.
[0095] Bearing fault feature extraction reconstructs and calculates the modal components selected based on the principle of fast kurtosis diagram, calculates its complexity, and comprehensively characterizes the wind turbine bearing fault data to achieve the purpose of quantitative diagnosis. The wind turbine bearing fault feature extraction method based on crown porcupine optimization algorithm-variational modal decomposition (CPO-VMD) achieves the ultimate goal of quantitative diagnosis of wind turbine bearing fault characteristics.
[0096] It should be noted that the variational mode decomposition uses a quadratic penalty factor α and a Lagrange multiplication operator λ(t) in the "variation" process to transform the constrained variation into an unconstrained variation, and then iterates the optimal solution of the variational model through the alternating direction multiplier method. In order to eventually achieve effective decomposition of the signal adaptively, the main method is to update the bandwidth and frequency center of each mode component multiple times. The variational mode decomposition algorithm is mainly to decompose the original signal by presetting the number of decompositions and decomposing the signal into a certain number of center frequencies ω. k modal function, and each mode u kAt the center frequency ω k The bandwidth of the fluctuation can be determined by H 1 Gaussian smoothing estimate.
[0097] When decomposing the signal, the variational mode decomposition method abandons the screening principle in the empirical mode decomposition method and takes a different approach to transform the decomposition process into a constrained variational problem solution process. The signal is decomposed by constructing and solving the constrained variational model. Assuming that eigenmode functions with limited bandwidth and different center frequencies exist in each mode function, then find k mode functions u k This will be the main purpose of solving the variational problem, and it will minimize the sum of the bandwidth estimates between the modes.
[0098] In summary, the process of modal decomposition of data signals by the variational mode decomposition algorithm is as follows:
[0099] 1) First, for each modal function, in order to obtain its analytical signal, the Hilbert transform is used to obtain its single-sided spectrum. The spectrum expression of the analytical signal is:
[0100]
[0101] In the above formula, δ(t) is the Dirac distribution, {u k}={u1,...,u k} represents the set of K BLIMFs after VMD decomposition, and j is an imaginary unit.
[0102] 2) Secondly, by mixing the analytical signals of each mode, a pre-estimated center frequency is obtained: The frequency shift is transformed to the baseband, and the expression after the analytical signal mixing is:
[0103]
[0104] Where δ(t) is the Dirac distribution, {u k}={u1,...,u k} represents the set of K BLIMFs after VMD decomposition, j is an imaginary unit, is the center frequency.
[0105] 3) Finally, the square value of the L2 norm of the time gradient of the demodulated signal is calculated to estimate the bandwidth of the modal component, which is expressed as:
[0106]
[0107] Therefore, the constrained variational model is constructed as follows:
[0108]
[0109] In the above formula, δ(t) represents the Dirichlet function, * represents the convolution operation, {u k}={u1,...,u k} represents the set of K BLIMFs after VMD decomposition; {ω k}={ω1,...,ω k} represents the combination of the center frequencies of K modal components, and f represents the input signal.
[0110] 4) To solve the constrained variational model, the quadratic penalty factor α and the Lagrange multiplication operator λ(t) are introduced to solve the problems that may arise when Gaussian noise exists. The former can ensure the accuracy of signal reconstruction, while the latter can ensure the strictness of the constraints. The expression of the augmented Lagrange multiplier L is as follows:
[0111]
[0112] Among them, δ(t) represents the Dirichlet function, * represents the convolution operation, {u k}={u1,...,u k} represents the set of K BLIMFs after VMD decomposition, {ω k}={ω1,...,ω k} represents the combination of the center frequencies of K modal components, f represents the input signal, α is the quadratic penalty factor, and λ(t) is the multiplication operator.
[0113] 5) The unconstrained variational problem can be solved by using the Alternate Direction Method of Multipliers (ADMM). Using the ADMM algorithm, the update is performed alternately. λ n+1 , find the “saddle point” in the extended Lagrangian expression.
[0114] For modal The update of can be equivalent to the solution of the minimization problem:
[0115]
[0116] In the above formula, δ(t) represents the Dirichlet function, * represents the convolution operation, {u k}={u1,...,u k} represents the set of K BLIMFs after VMD decomposition, {ω k}={ω1,...,ω k} represents the combination of the center frequencies of K modal components, f represents the input signal, α is the quadratic penalty factor, λ(t) is the multiplication operator, In order to seek partial guidance.
[0117] 6) Modal The solution formula of the minimization problem is transformed into the frequency domain by Parseval / Plancherel Fourier isometry transformation, and then the updated expression of the kth mode is obtained, which is expressed as follows:
[0118]
[0119] In the above formula, T is the fidelity coefficient.
[0120] According to the same principle, the center frequency solution is converted to frequency to obtain the update of the center frequency, and its expression is:
[0121]
[0122] In the above formula, Equivalent to the current remaining amount Wiener filter; center frequency is the corresponding modal function power spectrum The center of gravity, through The signal after Wiener filtering is subjected to inverse Fourier transform to obtain the mode u in the time domain signal k (t) and obtain the real part thereof.
[0123] 7) Using the alternating direction method of multipliers (ADMM) algorithm to constrain and simultaneously obtain the optimal solution of the variational model, k modal components decomposed from the original signal can be finally obtained.
[0124] Reference Figure 2 , a wind turbine bearing quantitative diagnosis system based on CPO-VMD, including:
[0125] A standardization module 201 is used to obtain the fault data of the wind turbine bearing and perform standardization processing to obtain the standardized bearing fault data, wherein the bearing fault data includes a defect acoustic emission signal and a vibration characteristic signal;
[0126] An optimization module 202 is used to optimize the parameters of the variational mode decomposition algorithm (VMD) by using a crown porcupine optimization algorithm (CPO) to construct a quantitative detection algorithm for wind turbine bearings based on CPO-VMD;
[0127] The diagnosis module 203 is used to perform feature detection processing on the standardized bearing fault data through the wind turbine bearing quantitative detection algorithm of CPO-VMD to obtain the wind turbine bearing quantitative diagnosis result.
[0128] The contents of the above method embodiments are all applicable to the present system embodiments. The functions specifically implemented by the present system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.
[0129] The above is a specific description of the preferred implementation of the present invention, but the invention is not limited to the embodiments. Those skilled in the art may make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.
Claims
1. A quantitative diagnosis method for wind turbine bearings based on CPO-VMD, characterized in that: The following steps are involved: Acquire the fault data of the wind turbine bearing and perform standardization processing to obtain the standardized bearing fault data, wherein the bearing fault data includes a defect acoustic emission signal and a vibration characteristic signal; The parameters of the variational mode decomposition (VMD) algorithm were optimized by using the crown porcupine optimization algorithm (CPO), and a quantitative detection algorithm for wind turbine bearings based on CPO-VMD was constructed. The wind turbine bearing quantitative detection algorithm of CPO-VMD is used to perform feature detection on the standardized bearing fault data to obtain the quantitative diagnosis results of the wind turbine bearing.
2. The CPO-VMD-based quantitative diagnosis method for wind turbine bearings according to claim 1, characterized in that: The expression of the standardization process is specifically as follows: In the above formula, X' z represents the normalized bearing failure data, X z represents the data of bearing fault, N represents the number of fault data, μ and σ represent the standardized parameters, μ is the mean value, and σ is the standard deviation value.
3. The CPO-VMD-based quantitative diagnosis method for wind turbine bearings according to claim 2 is characterized in that: The step of optimizing the parameters of the variational mode decomposition algorithm by using the crown porcupine optimization algorithm to construct a quantitative detection algorithm for wind turbine bearings based on CPO-VMD specifically includes: Based on the Hilbert spectral function, the parameters of the variational mode decomposition algorithm are optimized by the crested porcupine optimization algorithm to obtain the optimal modal component value and the optimal penalty factor value; The optimized modal component values and the optimized penalty factor values are embedded in the variational mode decomposition algorithm, and combined with the crested porcupine optimization algorithm, a quantitative detection algorithm for wind turbine bearings based on CPO-VMD is obtained.
4. The CPO-VMD-based quantitative diagnosis method for wind turbine bearings according to claim 3 is characterized in that: The step of optimizing the parameters of the variational modal decomposition algorithm based on the Hilbert spectral function by using the crown porcupine optimization algorithm to obtain the optimal modal component value and the optimal penalty factor value specifically includes: Initialize the crested porcupine population, set the value range of the modal component and the value range of the penalty factor, and set the crested porcupine distance threshold; The initialized crested porcupine population is activated by a cyclic population reduction method to obtain an activated crested porcupine population; If the distance between the predator and the crested porcupine is greater than the crested porcupine distance threshold, the activated crested porcupine population performs a global search process through the first defense strategy to obtain a global search defense factor; If the distance between the predator and the crested porcupine is less than the crested porcupine distance threshold, the activated crested porcupine population performs local search processing through the second defense strategy to obtain a local search defense factor; Based on the global search defense factor and the local search defense factor, the envelope entropy value of the parameters of the variational mode decomposition algorithm is calculated through the variational mode update algorithm, and the optimal modal component value and the optimal penalty factor value are determined according to the calculation results.
5. The CPO-VMD-based quantitative diagnosis method for wind turbine bearings according to claim 4 is characterized in that: The first defense strategy includes a crested porcupine visual exploration strategy and a crested porcupine sound exploration strategy, wherein the crested porcupine visual exploration strategy is specifically expressed as follows: In the above formula, represents the optimal solution for the function evaluation t, represents the vector generated between the current crested porcupine and the random crested porcupine, represents the position of the predator at iteration t, τ1 represents a random number based on a normal distribution, τ2 represents a random value in the interval [0,1], represents the position of the i-th individual at iteration t, represents the position of the i-th individual at iteration t+1; The expression of the crested porcupine sound exploration strategy is specifically as follows: In the above formula, r1 and r2 represent two random integers between [1, N], τ3 represents a random value generated between 0 and 1, and the randomly generated binary vector containing 0 and 1 To cover all probabilities, represents the predator position, represents the position of the r1th element in the tth iteration, Indicates the position of the r2th element in the tth iteration.
6. The CPO-VMD-based quantitative diagnosis method for wind turbine bearings according to claim 5, characterized in that: The second defense strategy includes a crested porcupine odor attack strategy and a crested porcupine physical attack strategy, wherein the crested porcupine odor attack strategy is specifically expressed as follows: In the above formula, τ3 represents a random number between the interval [1, N], δ represents the parameter controlling the search direction, and γ t represents the defense factor, represents the odor diffusion factor, Indicates the situation where the strategy is simulated. represents the position of the r1th element in the tth iteration, represents the position of the r2th element in the tth iteration, r1 and r2 represent two random integers between [1,N]; The expression of the crested porcupine physical attack strategy is specifically as follows: In the above formula, represents the optimal solution, α represents the convergence speed factor, represents the position of the i-th individual during iteration, τ4 and τ5 represent random values in the interval [0,1], represents the average force exerted by the crested porcupine on the i-th predator.
7. A wind turbine bearing quantitative diagnosis method based on CPO-VMD according to claim 6, characterized in that: The expression of the variational mode update algorithm is specifically as follows: In the above formula, The Wiener filter represents the residual quantity, represents the Fourier transform of the signal f(t), represents the ith modal update expression, represents the ith modal expression, represents the modal bandwidth corresponding to frequency ω, represents the center frequency of the kth mode in the nth iteration process, τ represents the fidelity coefficient, n represents an integer, k represents the initial mode number, i represents different modes or components, and ω represents the angular frequency.
8. The CPO-VMD-based quantitative diagnosis method for wind turbine bearings according to claim 7, characterized in that: The step of performing feature detection processing on the standardized bearing fault data by using the CPO-VMD wind turbine bearing quantitative detection algorithm to obtain the wind turbine bearing quantitative diagnosis result specifically includes: The standardized bearing fault data is modally decomposed by using the CPO-VMD wind turbine bearing quantitative detection algorithm to obtain the bearing fault data modal components. The kurtosis value is set to filter the modal components of the bearing fault data to obtain the filtered modal components of the bearing fault data; The modal components of the screened bearing fault data are reconstructed and calculated to obtain the quantitative diagnosis results of wind turbine bearings.
9. A wind turbine bearing quantitative diagnosis system based on CPO-VMD, characterized in that: Includes the following modules: A standardization module is used to obtain the fault data of the wind turbine bearing and perform standardization processing to obtain the standardized bearing fault data, wherein the bearing fault data includes a defect acoustic emission signal and a vibration characteristic signal; The optimization module is used to optimize the parameters of the variational mode decomposition algorithm (VMD) through the crown porcupine optimization algorithm (CPO) and build a quantitative detection algorithm for wind turbine bearings based on CPO-VMD; The diagnosis module is used to perform feature detection processing on the standardized bearing fault data through the CPO-VMD wind turbine bearing quantitative detection algorithm to obtain the wind turbine bearing quantitative diagnosis result.
Citation Information
Cited By
Load prediction method, system and device based on CPO-VMD-MLKELM combination model and medium
CN119940980A
Load Forecasting Method, System, Device and Medium Based on CPO-VMD-MLKELM Combined Model
CN119940980B
