A bearing fault diagnosis method and system based on bubble entropy and AOA to optimize SVM model parameters

Abstract

This invention proposes a bearing fault diagnosis method and system based on bubble entropy and AOA-optimized SVM model parameters. The method includes: acquiring vibration signals of bearing faults; adding adaptive white noise to the vibration signals and then decomposing them to obtain initial IMF components, and filtering all initial IMF components to obtain the required IMF components; calculating the bubble entropy corresponding to each filtered IMF component, and reconstructing feature vectors based on all obtained bubble entropies; inputting the feature vectors into a fault diagnosis classification model to obtain the bearing fault type, wherein the fault diagnosis classification model is a support vector machine model optimized using an arithmetic optimization algorithm. This invention achieves effective feature extraction through adaptive noise set empirical mode decomposition and bubble entropy, eliminating noise interference in the signal; and uses an arithmetic optimization algorithm to optimize the penalty factor and kernel parameters, avoiding the SVM model from getting trapped in local optima and improving the accuracy of its fault diagnosis classification results.

Description

Technical Field

[0001] This invention belongs to the field of rolling bearing fault diagnosis and identification, and specifically relates to a bearing fault diagnosis method and system that uses bubble entropy and AOA to optimize SVM model parameters. Background Technology

[0002] Modern machinery is evolving towards higher speeds, heavier loads, and greater precision. Rolling bearings, as core components of these machines, play a crucial role in maintaining equipment operation. However, the prolonged operation under complex conditions and the continuous application of alternating stresses cause bearings to degrade in performance, eventually leading to failure. Therefore, monitoring and identifying rolling bearing failures is of great significance, not only for preventing malfunctions in advance but also for effectively protecting human safety.

[0003] Feature extraction is a hot topic and key area in fault diagnosis, and many scholars both domestically and internationally have conducted in-depth research on it. Existing ensemble empirical mode decomposition (EEMD) adds white noise to empirical mode decomposition (EMD) to assist signal processing, effectively suppressing mode aliasing. However, this phenomenon still cannot be avoided in the low-frequency range. Variational mode decomposition (VMD) avoids the problems of EMD and has good noise resistance, but the VMD decomposition results are easily affected by the number of decomposition layers and the penalty factor setting. Manual parameter optimization relies on prior knowledge, resulting in low computational efficiency and failing to achieve optimal decomposition results. If grid algorithms or similar methods are used for global optimization, the selected objective function is susceptible to external interference and easily gets trapped in local optima. Summary of the Invention

[0004] In view of the shortcomings of the prior art, the purpose of this invention is to provide a bearing fault diagnosis method and system based on bubble entropy and AOA-optimized SVM model parameters, so as to solve the above-mentioned technical problems.

[0005] To achieve the above and other related objectives, this invention proposes a bearing fault diagnosis method using bubble entropy and AOA-optimized SVM model parameters, comprising:

[0006] Collect vibration signals from bearing failures;

[0007] The vibration signal is decomposed after adding adaptive white noise to obtain initial IMF components, and all the initial IMF components are filtered to obtain the required IMF components.

[0008] Calculate the bubble entropy corresponding to each IMF component obtained from screening, and reconstruct the feature vector based on all the obtained bubble entropies;

[0009] The feature vector is input into the fault diagnosis classification model to obtain the fault type of the bearing, wherein the fault diagnosis classification model is a support vector machine model optimized by an arithmetic optimization algorithm.

[0010] In one embodiment of the present invention, the steps of adding adaptive white noise to the vibration signal and then decomposing it to obtain initial IMF components, and then filtering all the initial IMF components to obtain the desired IMF components include:

[0011] After adding white noise to the vibration signal, EMD decomposition is performed to obtain multiple first IMF components;

[0012] The average value of the multiple first IMF components obtained from the decomposition is used to obtain the initial IMF components;

[0013] The initial IMF component is subtracted from the original signal to obtain the residual component;

[0014] Determine whether the residual component is a monotonic function;

[0015] If not, the residual component is used as the vibration signal, and the vibration signal is decomposed and the residual component is recalculated.

[0016] If so, the multiple initial IMF components obtained will be filtered to obtain the IMF component.

[0017] In one embodiment of the present invention, the bubble entropy corresponding to each screened IMF component is calculated, and the feature vector is reconstructed based on all the obtained bubble entropies;

[0018] Each IMF component obtained by screening is reconstructed in phase space to obtain a first matrix and a second matrix, wherein the dimension of the second matrix is ​​one dimension larger than that of the first matrix;

[0019] The first Ruili entropy and the second Ruili entropy corresponding to each IMF component are obtained based on the first matrix and the second matrix;

[0020] Calculate the bubble entropy corresponding to each IMF component based on the first Ruili entropy and the second Ruili entropy.

[0021] In one embodiment of the present invention, the step of obtaining Ruili entropy includes:

[0022] Swap adjacent elements in each row of the matrix until the elements in each row are in ascending order from left to right, and count the number of swaps in each row.

[0023] Obtain a histogram of the number of occurrences of each type of exchange and the number of occurrences of that exchange, normalize the histogram to obtain the probability of each type of exchange occurrence, and calculate its Rayleigh entropy.

[0024] In one embodiment of the invention, the probability of each number of swaps is calculated by the following formula:

[0025] Where, p i n represents the probability of the number of swaps being i; i Let be the number of swaps i; N represents the length of the IMF component; m represents the dimension of the matrix.

[0026] In one embodiment of the present invention, Ruili entropy is determined by the formula... calculate.

[0027] In one embodiment of the present invention, the bubble entropy is calculated according to the following formula:

[0028] Where BE represents bubble entropy; This represents the second Ruili entropy; This represents the first Ruili entropy.

[0029] In one embodiment of the present invention, the step of obtaining a fault diagnosis classification model using a support vector machine model optimized by an arithmetic optimization algorithm includes:

[0030] Set the range of parameter penalty factor and kernel parameter, randomly select a value from the range of parameter penalty factor and kernel parameter to form a particle, and form a particle swarm from multiple particles;

[0031] The parameters of the particle swarm, the penalty factor and the kernel parameter are updated iteratively, and the fitness value of the particle is calculated. If the fitness value of the updated particle is better than the fitness value before the update, then the updated particle is taken as the optimal particle.

[0032] Determine if the number of update iterations is equal to the preset maximum number of iterations;

[0033] If the number of iterations is equal to the preset maximum number of iterations, then the optimization is complete; obtain the optimal particle, and construct the optimal support vector machine model based on the parameter penalty factor and kernel parameters corresponding to the optimal particle.

[0034] Otherwise, the parameter penalty factor and kernel parameters of the particle are repeatedly updated for iteration.

[0035] In one embodiment of the present invention, the fault type includes one or more of the following: normal bearing, single-point fault of rolling element, multi-point fault of rolling element, single-point fault of inner ring, multi-point fault of inner ring, single-point fault of outer ring, multi-point fault of outer ring, combined fault of outer ring and roller, and combined fault of inner ring and roller.

[0036] This invention also proposes a bearing fault diagnosis system that optimizes SVM model parameters using bubble entropy and AOA, comprising:

[0037] The signal acquisition module is used to acquire vibration signals from bearing failures.

[0038] The data processing module is used to add adaptive white noise to the vibration signal and then decompose it to obtain initial IMF components, and to filter all the initial IMF components to obtain the required IMF components; and to calculate the bubble entropy corresponding to each filtered IMF component, and to reconstruct the feature vector based on all the obtained bubble entropies.

[0039] The model optimization module is used to optimize the support vector machine model using arithmetic optimization algorithms to obtain a fault diagnosis classification model.

[0040] The fault diagnosis and classification module is used to input the feature vector into the fault diagnosis and classification model to obtain the fault type of the bearing.

[0041] This invention proposes an electronic device, comprising: one or more processors; and a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the electronic device enables the bearing fault diagnosis method based on bubble entropy and AOA-optimized SVM model parameters as described in any of the above embodiments.

[0042] This invention proposes a computer-readable storage medium storing a computer program thereon. When the computer program is executed by a computer processor, it causes the computer to perform a bearing fault diagnosis method based on bubble entropy and AOA-optimized SVM model parameters as described in any of the above embodiments.

[0043] This invention proposes a bearing fault diagnosis method and system based on bubble entropy and AOA-optimized SVM model parameters. It combines adaptive noise set empirical mode decomposition, bubble entropy, and support vector machine (SVM) for bearing fault diagnosis. Effective feature extraction is achieved through adaptive noise set empirical mode decomposition and bubble entropy, eliminating noise interference in the signal. Arithmetic optimization algorithms are used to optimize the penalty factor and kernel parameters, preventing the SVM model from getting trapped in local optima and improving the accuracy of its fault diagnosis and classification results. Attached Figure Description

[0044] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 This is a flowchart of a bearing fault diagnosis method according to one embodiment of the present invention.

[0046] Figure 2 This is a flowchart illustrating the steps of obtaining the required IMF components in one embodiment of the present invention.

[0047] Figure 3 This is a flowchart illustrating the steps of calculating bubble entropy and reconstructing feature vectors in one embodiment of the present invention.

[0048] Figure 4 This is a flowchart illustrating the steps of optimizing a fault diagnosis classification model in one embodiment of the present invention.

[0049] Figure 5 This is a structural block diagram of a bearing fault diagnosis system according to one embodiment of the present invention.

[0050] Figure 6 This is a time-domain signal diagram of a bearing outer ring fault in one embodiment of the present invention.

[0051] Figure 7 This is a diagram of the first 6 IMF components obtained by fault signal decomposition in one embodiment of the present invention.

[0052] Figure 8 This is an optimization curve of the AOA algorithm in one embodiment of the present invention.

[0053] Figure 9 This is a classification result diagram of a bearing fault diagnosis method in one embodiment of the present invention.

[0054] Figure 10 The image shows the classification results of the existing BE-SVM classification method.

[0055] Figure 11 The image shows the classification results of the existing PE-AOA-SVM classification method.

[0056] Figure 12 This is a schematic diagram of the computer system of an electronic device according to one embodiment of the present invention. Detailed Implementation

[0057] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention.

[0058] It should be noted that the illustrations provided in this embodiment are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0059] Please see Figure 1 As shown in this embodiment, to solve the above-mentioned technical problems, this invention proposes a bearing fault diagnosis method and system based on bubble entropy and AOA-optimized SVM model parameters. This bearing fault diagnosis and classification method is based on a combination of Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN), bubble entropy, and Support Vector Machine (SVM). This method first decomposes the model using CEEMDAN to obtain a series of Intrinsic Mode Function (IMF) components. Then, it selects important IMF components and calculates their bubble entropy values. Based on these bubble entropy values, a fault feature vector is constructed and input into a support vector machine model optimized by Arithmetic Optimization Algorithm (AOA) for training and bearing fault classification. This effectively improves the accuracy of rolling bearing fault diagnosis, achieving an accuracy of up to 97.8%, and can be used for bearing fault classification. For details, please refer to [link to relevant documentation]. Figure 1 As shown, the bearing fault diagnosis method includes:

[0060] Step S101: Collect vibration signals of bearing failure;

[0061] Please see Figure 1 As shown, in this embodiment, an acceleration vibration sensor and a sound sensor are installed on the spindle bearing testing machine to collect vibration signals during bearing operation. It should be noted that the spindle bearing testing machine operates at a test speed of 3000 t / min, with a radial load of 4 kN, and collects vibration signals at a sampling frequency of 20480 Hz for a sampling time of 1 second. In this embodiment, the acceleration vibration sensor and sound sensor installed on the spindle bearing testing machine are used to collect full-life data of the bearing from initial operation to complete failure.

[0062] Step S102: After adding adaptive white noise to the vibration signal, decompose it to obtain initial IMF components, and filter all the initial IMF components to obtain the required IMF components.

[0063] Please see Figure 1 and Figure 2 As shown, in this embodiment, by adding white noise in each decomposition process, the computational efficiency is improved and the reconstruction error is almost zero. Specifically, the steps of adding adaptive white noise to the vibration signal before decomposition to obtain initial IMF components, and filtering all the initial IMF components to obtain the required IMF components include:

[0064] S201. White noise is added to the original signal and then EMD decomposition is performed to obtain multiple first IMF components;

[0065] Please see Figure 1 and Figure 2 As shown, specifically, the first signal obtained by adding white noise to the vibration signal y(t) is: R i (t)=y(t)+β0ω i (t), the first signal obtained after each addition of white noise is subjected to EMD decomposition to obtain multiple first IMF components. i Where i represents the number of EMD decompositions, and i = 1, 2, Λ, I; β0 represents the amplitude of the added white noise; ω i (t) represents the normally distributed Gaussian white noise added during the i-th decomposition; IMF i This represents the IMF component obtained from the i-th EMD decomposition.

[0066] S202, average the multiple first IMF components obtained from the decomposition to obtain the initial IMF component IMF. j Please see. Figure 1 and Figure 2 As shown, among which, IMF j Let IMF represent the initial IMF component of the j-th iteration. For example, in the first iteration, the initial IMF component obtained after EMD decomposition of the vibration signal y(t) as the original signal and taking the average value is the first initial IMF component IMF1, i.e.

[0067] S203. Subtract the initial IMF component from the original signal to obtain the residual component. (See also...) Figure 2 As shown, for example, the residual component obtained in the first iteration is: Z1(t)=y(t)-IMF1.

[0068] S204. Determine whether the residual component is a monotonic function;

[0069] S205. If not, the residual component is used as the vibration signal, and the vibration signal is decomposed and the residual component is calculated again. That is, the process of adding white noise to perform EMD decomposition until the average value is taken to obtain the residual component is repeated. That is, the process of repeating steps S202 to S204 is iterated.

[0070] S206. If so, the multiple initial IMF components that have been saved and obtained will be filtered to obtain the IMF components.

[0071] Please see Figure 1 and Figure 2 As shown, for example, in the first iteration, step S205 uses the residual component Z1(t) as the original signal and repeats steps S201 to S204 to obtain the first initial IMF component IMF2 and the corresponding residual component Z2(t) = Z1(t) - IMF2. In the remaining iterations, the initial IMF component IMF2 is obtained. j and the corresponding residual component Z j (t)=Z j-1 (t)-IMF j Among them, Z j (t) represents the residual component obtained in the j-th iteration, j = 1, 2, ..., n. This process is repeated until the obtained residual component is a monotonic function r(t), and multiple initial IMF components are obtained simultaneously. j .

[0072] Filter all the initial IMF components to obtain the required IMF components; see [link to relevant documentation]. Figure 1 and Figure 2 As shown, specifically, for example, a first signal Y(t) is constructed based on all the initial IMF components obtained from the decomposition and the monotonic function r(t), where Then, the required IMF components are selected based on the first signal.

[0073] Step S103: Calculate the bubble entropy corresponding to each IMF component obtained from the screening, and reconstruct the feature vector based on all the obtained bubble entropies; where, please refer to Figure 1 and Figure 3 As shown, the steps for calculating the bubble entropy corresponding to each screened IMF component include:

[0074] S301, each IMF component obtained by screening is reconstructed in phase space to obtain a first matrix and a second matrix, wherein the dimension of the second matrix is ​​one dimension larger than the dimension of the first matrix;

[0075] S302, obtain the first Ruili entropy and the second Ruili entropy corresponding to each IMF component based on the first matrix and the second matrix;

[0076] S303, calculate the bubble entropy corresponding to each IMF component based on the first Ruili entropy and the second Ruili entropy.

[0077] The steps for obtaining the entropy of Ruili include:

[0078] S3021, swap the order of elements in each row of the first matrix until the order of elements in each row increases from left to right, and count the number of swaps in each row.

[0079] S3022, obtain a histogram of the number of exchanges that occur, normalize the histogram to obtain the probability of each number of exchanges that occur, and calculate its first Ruili entropy.

[0080] For an explanation using the first matrix as an example, please refer to [link / reference]. Figure 1 and Figure 3 As shown, for example, a phase space reconstruction is performed on a certain IMF component obtained after screening to obtain an m-dimensional first matrix S, S={S(1),S(2),Λ,S(N-m+1)}, where S(i)={s(i),s(i+1),Λ,S(m)}, where N represents the length of the IMF component.

[0081] Please see Figure 1 and Figure 3 As shown, for example, the order of elements in each row of the first matrix S is swapped until the order of elements in each row increases from left to right. The number of swaps in each row is counted. For example, if the number of elements in each row is m, starting from the first element, the following m-1 elements need to be judged. If the left element is smaller than the right element, the position remains unchanged; otherwise, the position is swapped. The maximum number of swaps in each row is...

[0082] Please see Figure 1 and Figure 3 As shown, for example, the number of swaps required to achieve an increasing left-to-right order for elements in each row is indeterminate. A histogram is obtained by counting the number of each swap and its corresponding count, and the probability of each swap is calculated, for example, using the following formula:

[0083] Where, p i n represents the probability of the number of swaps being i; i Let be the number of swaps i; N represents the length of the IMF component; m represents the dimension of the matrix.

[0084] After calculating the probability of each type of exchange, the formula is used. Calculate the corresponding first Ruili entropy.

[0085] Please see Figure 1 and Figure 3 As shown, specifically, the bubble entropy BE is calculated using the following formula:

[0086] Where BE represents bubble entropy; This represents the second Ruili entropy; This represents the first Ruili entropy.

[0087] Please see Figure 1 and Figure 3 As shown, after calculating the bubble entropy corresponding to each IMF component required in step S102 according to steps S301 to S303 above, a feature vector is constructed based on its bubble entropy.

[0088] Step S104: Input the feature vector into the fault diagnosis classification model to obtain the fault type of the bearing, wherein the fault diagnosis classification model is a support vector machine model optimized by an arithmetic optimization algorithm.

[0089] Please see Figure 1 and Figure 4 As shown, in this embodiment, the steps for obtaining the fault diagnosis classification model using the support vector machine model optimized by the arithmetic optimization algorithm include:

[0090] S401, Set the range of the parameter penalty factor and the kernel parameter, randomly select a value from the range of the parameter penalty factor and the kernel parameter to form particles, and form a particle swarm from multiple particles; Please refer to Figure 1 and Figure 4 As shown, for example, by pre-setting the parameter to be optimized, namely the parameter penalty factor and the parameter value range of the kernel parameter, and then randomly selecting a value from their respective value ranges to form a particle, a particle swarm is formed by multiple such particles, thus completing the initialization of the particle positions in the particle swarm.

[0091] S402, update the parameter penalty factor and kernel parameter of the particle iteratively, and calculate the fitness value of the particle. If the fitness value of the updated particle is better than the fitness value before the update, then the updated particle is taken as the optimal particle.

[0092] Please see Figure 1 and Figure 4 As shown, specifically, the method for updating the particle's parameter penalty factor and kernel parameters, i.e., updating the particle's position, includes:

[0093] The Attribute Optimizer Probability (MOA) is calculated using the formula... The calculation yielded that, MOA stands for Attribute Optimizer Probability; MOP max and MOP min represents the maximum and minimum values ​​of the acceleration function, respectively; t is the current iteration number; M is the maximum iteration number; α represents the accuracy of local exploration during the iteration process, and in this embodiment, α is set to 5.

[0094] The value of r1 is randomly selected between [0,1]. When r1 < MOA, the exploration phase begins.

[0095] Randomly select the value of r2 between [0,1]. When r2 < 0.5, explore and update using the following formula.

[0096]

[0097] When r² ≥ 0.5, the following formula is used for exploratory updates:

[0098]

[0099] When r1 > MOA, the development phase begins;

[0100] Randomly select the value of r2 between [0,1]. When r3 < 0.5, update the value using the following formula.

[0101]

[0102] When r3 ≥ 0.5, the development and update are performed using the following formula.

[0103]

[0104] In the formula, X t+1 This represents the parameter penalty factor or nuclear parameter of the particle during the (t+1)th iteration; Let denot be the optimal particle in the population at the t-th iteration, ε be a random minimum, and ub j and lb j These are the upper and lower bounds of the parameter penalty factor or kernel parameter, respectively. ω is the adjustment parameter, and r1, r2, and r3 all represent random numbers between [0,1].

[0105] Please see Figure 1 and Figure 4 As shown, the particle's position is updated using the above formula, i.e., the parameter penalty factor and kernel parameters are updated, and the particle's fitness value is calculated. If the updated particle's fitness value is better than the unupdated fitness value, then the updated particle is taken as the optimal particle. It should be noted that in each update of the particle's parameter penalty factor or kernel parameters, the parameter penalty factor is updated first, and then the particle's kernel parameters are updated using the above embodiment.

[0106] S403, determine whether the number of update iterations is equal to the preset maximum number of iterations M. In this embodiment, the maximum number of iterations M is, for example, 100.

[0107] S404, if the number of iterations is equal to the preset maximum number of iterations M, then the optimization is complete; obtain the optimal particle, and construct the optimal support vector machine model based on the parameter penalty factor and kernel parameters corresponding to the optimal particle.

[0108] Otherwise, continue updating the particle's parameter penalty factor and nuclear parameters for iteration until the number of iterations is equal to the preset number of iterations M;

[0109] Please see Figure 1 and Figure 4 As shown, the optimized support vector machine model, i.e., the fault diagnosis classification model, is obtained through the above optimization method. The feature vector obtained in step S103 is used as the input of the fault diagnosis classification model to output the bearing fault type. The fault type includes one or more of the following: normal bearing, single-point rolling element fault, multi-point rolling element fault, single-point inner ring fault, multi-point inner ring fault, single-point outer ring fault, multi-point outer ring fault, combined outer ring and roller fault, and combined inner ring and roller fault.

[0110] Please see Figure 5 As shown, this invention also proposes a bearing fault diagnosis system based on bubble entropy and AOA-optimized SVM model parameters. The fault diagnosis classification system 100 includes a signal acquisition module 10, a data processing module 20, a model optimization module 30, and a fault diagnosis classification module 40. The signal acquisition module 10 is used to acquire vibration signals of bearing faults. The data processing module 20 is used to add adaptive white noise to the vibration signals and then decompose them to obtain initial IMF components, and to filter all the initial IMF components to obtain the required IMF components. The data processing module 20 is also used to calculate the bubble entropy corresponding to each filtered IMF component and reconstruct the feature vector based on all the obtained bubble entropies. The model optimization module 30 is used to optimize the support vector machine model using an arithmetic optimization algorithm to obtain a fault diagnosis classification model. The fault diagnosis classification module 40 is used to input the feature vector into the fault diagnosis classification model to obtain the fault type of the bearing.

[0111] In one specific embodiment, vibration signals of a bearing were collected. The bearing model was an NSK NU1010 single-row cylindrical roller bearing. The experimental conditions were a rotational speed of 3000 r / min, a radial load of 4 kN, and a sampling frequency of 20480 Hz.

[0112] To obtain bearings with different faults, laser processing was used to obtain bearings with inner ring faults, while wire cutting was used to obtain bearings with outer ring and roller faults. Based on the location and degree of damage to the rolling bearings, the experimental bearings were divided into six fault types: normal, outer ring fault, inner ring fault, roller fault, outer ring-roller fault, and inner ring-roller fault. Labels were then defined for each fault type.

[0113] Because the bearing test bench is subject to background noise interference during operation, the vibration signals under the six states acquired in step S101 are first decomposed using the decomposition method described in step S102 of the above embodiment. Specifically, the standard deviation of the added white noise is 0.2, the noise is added 100 times, and the sampling points are 2000, which can reduce noise interference. The time-domain signal of the bearing outer ring fault is as follows: Figure 6 As shown, after performing the decomposition method described in step S102 of the above embodiments, a series of IMF components for each state are obtained. In this embodiment, only the first 6 IMF components are shown, and the first 6 IMF components are as follows: Figure 7 As shown.

[0114] Please see Figure 7 As stated, from Figure 7 It can be seen that IMF1 and IMF2 contain significant impact components and a large amount of fault information, while IMF3 to IMF5 contain fewer impact components. The waveform of IMF5 is similar to a sine wave, and the waveforms of IMF6 to IMF5 are... 10 The amplitude is much smaller than the first five orders, so it is not considered. Two IMF components with a high degree of correlation with the original signal, namely the first two IMF components, are selected for feature extraction. Some feature values ​​of the rolling bearing under the six states are shown in Table 1.

[0115] Table 1. Partial characteristic values ​​of rolling bearings under six conditions.

[0116]

[0117] Feature vectors are constructed based on the bubble entropy of each IMF classification, and these feature vectors are used as samples to input into an AOA-optimized SVM model for fault diagnosis and classification. Initial parameters are set as follows: population size N = 20, upper boundary ub = 100, lower boundary lb = 0.0001, maximum number of iterations M = 100, and the number of parameters to be optimized dim = 2, i.e., parameter penalty factor c and kernel parameter g. After iterative updates, the optimal parameter penalty factor c and optimal kernel parameter g are output. An SVM model is then constructed based on these optimal parameters to obtain the fault diagnosis and classification module. Finally, the feature vectors are input into the fault diagnosis and classification module for fault classification. The optimization iteration curve of AOA-SVM is shown below. Figure 8 As shown, the final classification result is as follows: Figure 6As shown, where Figure 6 The vertical axis category labels 1, 2, 3, 4, 5, and 6 correspond to the normal condition of the bearing, outer ring failure, inner ring failure, roller failure, outer ring roller failure, and inner ring roller failure, respectively.

[0118] To verify the superiority of the method, it was compared with the BE-SVM and PE-AOA-SVM diagnostic models respectively, and the classification results are as follows: Figure 9 , Figure 10 As shown in Table 2, the fault identification accuracy of the three fault diagnosis models is as follows.

[0119] Table 2. Accuracy of Fault Classification Model Recognition

[0120]

[0121] From Table 2 and Figure 9 , Figure 10 and Figure 11 As can be seen, the method presented in this paper (BE-AOA-SVM method) achieves an accuracy of 97.8%, with only 2 samples not correctly identified, representing a 2.2% improvement over BE-SVM. In contrast, the accuracy of PE-AOA-SVM is only 77.8%, demonstrating the superiority of the proposed method. Experimental results on fault diagnosis classification proposed in this invention show that the classification accuracy of the method reaches 97.8%, verifying its superiority.

[0122] Embodiments of this application also provide an electronic device 200, including: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, enable the electronic device to implement the bearing fault diagnosis method based on bubble entropy and AOA-optimized SVM model parameters provided in the above embodiments.

[0123] Figure 12 This is a schematic diagram illustrating the structure of a computer system suitable for implementing the electronic devices of the present application. It should be noted that... Figure 12 The computer system of the electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0124] like Figure 12 As shown, the computer system includes a central processing unit (CPU), which can perform various appropriate actions and processes based on a program stored in read-only memory (ROM) or a program loaded from storage into random access memory (RAM), such as performing the methods described in the above embodiments.

[0125] It should be noted that the computer-readable medium shown in the embodiments of this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer program contained on a computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0126] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. Each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0127] The units described in the embodiments of this application can be implemented in software or hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the specific unit itself.

[0128] Another aspect of this application provides a computer-readable storage medium storing a computer program thereon, which, when executed by a computer's processor, causes the computer to perform the bearing fault diagnosis method using bubble entropy and AOA-optimized SVM model parameters as described above. This computer-readable storage medium may be included in the electronic device described in the above embodiments, or it may exist independently and not assembled into the electronic device.

[0129] Another aspect of this application provides a computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the bearing fault diagnosis method based on bubble entropy and AOA-optimized SVM model parameters provided in the various embodiments above.

[0130] This invention proposes a bearing fault diagnosis method and system based on bubble entropy and AOA-optimized SVM model parameters. It combines adaptive noise set empirical mode decomposition, bubble entropy, and support vector machine (SVM) for bearing fault diagnosis. Effective feature extraction is achieved through adaptive noise set empirical mode decomposition and bubble entropy, eliminating noise interference in the signal. Arithmetic optimization algorithms are used to optimize the penalty factor and kernel parameters, preventing the SVM model from getting trapped in local optima and improving the accuracy of its fault diagnosis and classification results.

[0131] The above description is merely a preferred embodiment of this application and an explanation of the technical principles used. Those skilled in the art should understand that the scope involved in this application is not limited to the technical solutions formed by a specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the inventive concept. For example, technical solutions formed by replacing the above-mentioned features with technical features with similar functions disclosed in this application (but not limited to) each other.

[0132] Apart from the technical features described in the specification, the other technical features are known to those skilled in the art. To highlight the innovative features of this invention, the other technical features will not be described in detail here.

Claims

1. A bearing fault diagnosis method using bubble entropy and AOA to optimize SVM model parameters, characterized in that, include: Collect vibration signals from bearing failures; The vibration signal is decomposed after adding adaptive white noise to obtain initial IMF components, and all the initial IMF components are filtered to obtain the required IMF components. Calculate the bubble entropy corresponding to each IMF component obtained from the screening, and reconstruct the feature vector based on all the obtained bubble entropies. The feature vector is input into the fault diagnosis classification model to obtain the fault type of the bearing, wherein the fault diagnosis classification model is a support vector machine model optimized by an arithmetic optimization algorithm; The calculation of the bubble entropy corresponding to each screened IMF component includes: Each IMF component obtained by screening is reconstructed in phase space to obtain a first matrix and a second matrix, wherein the dimension of the second matrix is ​​one dimension larger than that of the first matrix; The first Ruili entropy and the second Ruili entropy corresponding to each IMF component are obtained based on the first matrix and the second matrix; The bubble entropy corresponding to each IMF component is calculated based on the first Ruili entropy and the second Ruili entropy, and the bubble entropy is calculated according to the following formula: Where BE represents bubble entropy; This represents the second Ruili entropy; Represents the first Ruili entropy; The steps to obtain Ruili entropy include: Swap adjacent elements in each row of the matrix until the elements in each row are in ascending order from left to right, and count the number of swaps in each row. Obtain a histogram of the number of occurrences of each type of exchange and the number of occurrences of that exchange, normalize the histogram to obtain the probability of each type of exchange occurrence, and calculate its Rayleigh entropy.

2. The bearing fault diagnosis method based on bubble entropy and AOA-optimized SVM model parameters according to claim 1, characterized in that, The steps of adding adaptive white noise to the vibration signal and then decomposing it to obtain initial IMF components, and then filtering all the initial IMF components to obtain the desired IMF components, include: After adding white noise to the vibration signal, EMD decomposition is performed to obtain multiple first IMF components; The average value of the multiple first IMF components obtained from the decomposition is used to obtain the initial IMF components; The initial IMF component is subtracted from the original signal to obtain the residual component; Determine whether the residual component is a monotonic function; If not, the residual component is used as the vibration signal, and the vibration signal is decomposed and the residual component is recalculated. If so, the multiple initial IMF components obtained will be filtered to obtain the IMF component.

3. The bearing fault diagnosis method based on bubble entropy and AOA-optimized SVM model parameters according to claim 1, characterized in that, The probability of each number of exchanges is calculated using the following formula: Where, p i n represents the probability of the number of swaps being i; i Let be the number of swaps i; N represents the length of the IMF component; m represents the dimension of the matrix.

4. The bearing fault diagnosis and classification method according to claim 1, characterized in that, Ruili entropy through formula calculate.

5. The bearing fault diagnosis method based on bubble entropy and AOA-optimized SVM model parameters according to claim 1, characterized in that, The steps for obtaining a fault diagnosis classification model using a support vector machine model optimized by an arithmetic optimization algorithm include: Set the range of parameter penalty factor and kernel parameter, randomly select a value from the range of parameter penalty factor and kernel parameter to form a particle, and form a particle swarm from multiple particles; The parameters of the particle swarm, the penalty factor and the kernel parameter are updated iteratively, and the fitness value of the particle is calculated. If the fitness value of the updated particle is better than the fitness value before the update, then the updated particle is taken as the optimal particle. Determine if the number of update iterations is equal to the preset maximum number of iterations; If the number of iterations is equal to the preset maximum number of iterations, then the optimization is complete; obtain the optimal particle, and construct the optimal support vector machine model based on the parameter penalty factor and kernel parameters corresponding to the optimal particle. Otherwise, the parameter penalty factor and kernel parameters of the particle are repeatedly updated for iteration.

6. The bearing fault diagnosis method based on bubble entropy and AOA-optimized SVM model parameters according to claim 1, characterized in that, The fault types include one or more of the following: normal bearing, single-point fault of rolling element, multi-point fault of rolling element, single-point fault of inner ring, multi-point fault of inner ring, single-point fault of outer ring, multi-point fault of outer ring, combined fault of outer ring and roller, and combined fault of inner ring and roller.

7. A bearing fault diagnosis system based on bubble entropy and AOA-optimized SVM model parameters, characterized in that, include: The signal acquisition module is used to acquire vibration signals from bearing failures. The data processing module is used to add adaptive white noise to the vibration signal and then decompose it to obtain initial IMF components, and to filter all the initial IMF components to obtain the required IMF components. And the bubble entropy used to calculate the bubble entropy corresponding to each screened IMF component, and to reconstruct the feature vector based on all the obtained bubble entropies, wherein calculating the bubble entropy corresponding to each screened IMF component includes: Each IMF component obtained by screening is reconstructed in phase space to obtain a first matrix and a second matrix, wherein the dimension of the second matrix is ​​one dimension larger than that of the first matrix; The first Ruili entropy and the second Ruili entropy corresponding to each IMF component are obtained based on the first matrix and the second matrix; The bubble entropy corresponding to each IMF component is calculated based on the first Ruili entropy and the second Ruili entropy, and the bubble entropy is calculated according to the following formula: Where BE represents bubble entropy; This represents the second Ruili entropy; Represents the first Ruili entropy; The steps to obtain Ruili entropy include: Swap adjacent elements in each row of the matrix until the elements in each row are in ascending order from left to right, and count the number of swaps in each row. Obtain a histogram of the number of occurrences of each type of exchange and the number of occurrences of that exchange, normalize the histogram to obtain the probability of each type of exchange occurrence, and calculate its Rayleigh entropy; The model optimization module is used to optimize the support vector machine model using arithmetic optimization algorithms to obtain a fault diagnosis classification model. The fault diagnosis and classification module is used to input the feature vector into the fault diagnosis and classification model to obtain the fault type of the bearing.

Images