Machine tool bearing fault feature extraction method and system, storage medium and equipment

Abstract

The invention belongs to the technical field of rotating machinery fault diagnosis, and provides a machine tool bearing fault feature extraction method and system, a storage medium and equipment, and the method comprises the steps: respectively building dictionaries for a fault impact component, a harmonic component and a modal component of a vibration signal of a machine tool bearing; establishing a physical information multi-dictionary sparse decomposition framework based on the original vibration signals and the constructed dictionaries, and solving the physical information multi-dictionary sparse decomposition framework until a global optimal solution is converged; and according to a solving result, removing a modal component and a harmonic component in the vibration signal, and extracting a fault impact component for fault diagnosis. According to the method, the problems of low feature extraction accuracy and more interference in the existing method are solved, the fault impact component, the separation mode and the harmonic component can be effectively extracted, and the noise is suppressed, so that the accuracy and the reliability of fault diagnosis are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of rotating machinery fault diagnosis, and in particular relates to a method, system, storage medium and equipment for extracting fault features of machine tool bearings. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] In the fault diagnosis of machine tool bearings, the vibration signal usually contains multiple components, such as fault impact components, harmonic components, modal components and noise. The traditional sparse decomposition method has the following problems when processing complex vibration signals: (1) The dictionary design does not match the actual fault impulse waveform well, resulting in inaccurate fault feature extraction; (2) The modal components and harmonic components have a great impact on the extraction of fault features, making it difficult to effectively separate them; (3) The L1 norm of traditional sparse decomposition methods may over-penalize small amplitude coefficients, thereby losing important fault information. Summary of the invention

[0004] In order to solve the above problems, the present invention proposes a machine tool bearing fault feature extraction method, system, storage medium and equipment. The present invention is based on physical information multi-dictionary sparse decomposition, which can effectively extract fault impulse components, separate modal and harmonic components, and suppress noise, thereby improving the accuracy and reliability of fault diagnosis.

[0005] According to some embodiments, the present invention adopts the following technical solutions: A method for extracting machine tool bearing fault features comprises the following steps: Obtain vibration signals of machine tool bearings; By using the physical characteristics of fault impulse, a generalized double-pulse response wavelet dictionary is established to represent the fault impulse component in the vibration signal. A harmonic dictionary is constructed using discrete Fourier transform to represent the harmonic component in the vibration signal. The random subspace method based on the cepstrum is used to perform working modal analysis to obtain modal parameters, and then a modal dictionary is constructed to represent the modal components in the vibration signal. Based on the original vibration signal and the constructed dictionaries, and introducing the weighted L1 norm, a physical information multi-dictionary sparse decomposition framework is established, and the physical information multi-dictionary sparse decomposition framework is solved until it converges to a global optimal solution; According to the solution results of the physical information multi-dictionary sparse decomposition framework, the modal components and harmonic components in the vibration signal are removed, and the fault impulse components are extracted for fault diagnosis.

[0006] As an optional implementation method, the process of establishing a generalized double-pulse response wavelet dictionary using the physical characteristics of fault impulses includes: establishing a time domain expression of the generalized double-pulse response wavelet, parameterizing the size of the spalling zone therein, determining the size of the spalling zone by the pulse duration, and introducing it into the construction of the double-pulse response wavelet dictionary.

[0007] As a further limited implementation, the generalized double impulse response wavelet dictionary is constructed as: ; Among them, the pulse appearing in the spalling zone is , represents the convolution operation, To express the initial time of the pulse, the time domain expression of the generalized double pulse response wavelet is: ; in, A i is the amplitude parameter, i=1, 2, 3, 4, , and are the damping ratio, undamped natural frequency and damped natural frequency respectively, is the time interval between two pulses, and , is the inner diameter of the bearing, is the rotation frequency, is the diameter of the spalling zone.

[0008] As an optional implementation, based on the original vibration signal and the constructed dictionaries, and introducing the weighted L1 norm, the process of establishing the objective function of the physical information multi-dictionary sparse decomposition framework includes: integrating the weighted L1 norm into the sparse regularization decomposition, and establishing the physical information multi-dictionary sparse decomposition framework as follows: ; in, Di It is with i The framework is a unified form of sparse decomposition, which decomposes the vibration signal into multiple components and corresponding dictionaries. It is i trade-off parameters, It is i The sparse coefficients of the excitation sources, is the weight distribution coefficient, given by p The matrix is ​​composed of , Represents a diagonal matrix , weight coefficient allocation model Defined as: ; in is the stability parameter.

[0009] As a further limited implementation, the physical information multi-dictionary sparse decomposition framework can be used to perform single-dictionary sparse decomposition, dual-dictionary sparse decomposition or multi-dictionary decomposition, which can be used for fault diagnosis of single excitation source and / or multiple excitation sources.

[0010] As an optional implementation method, the process of solving the physical information multi-dictionary sparse decomposition framework until converging to the global optimal solution includes: dividing the physical information multi-dictionary sparse decomposition framework into the sum of two functions, performing variable splitting, introducing auxiliary variables, converting the unconstrained optimization problem into a constrained problem, using the alternating minimization method to handle the constraint problem, using the enhanced Lagrangian method to rewrite the constraint problem, and performing a minimization process until convergence to the global minimum.

[0011] As an optional implementation, during the minimization process, the matrix inverse lemma is used to simplify the solution process.

[0012] A machine tool bearing fault feature extraction system, comprising: A signal acquisition module is configured to acquire a vibration signal of a machine tool bearing; The multi-dictionary construction module is configured to use the physical characteristics of the fault impulse to establish a generalized double-pulse response wavelet dictionary for representing the fault impulse component in the vibration signal, and to use discrete Fourier transform to construct a harmonic dictionary for representing the harmonic component in the vibration signal; use the random subspace method based on the cepstrum to perform working modal analysis to obtain modal parameters, and then construct a modal dictionary for representing the modal components in the vibration signal; A multi-dictionary sparse decomposition module is configured to establish a physical information multi-dictionary sparse decomposition framework based on the original vibration signal and each constructed dictionary, and introduce a weighted L1 norm, and solve the physical information multi-dictionary sparse decomposition framework until it converges to a global optimal solution; The fault feature extraction module is configured to remove the modal components and harmonic components in the vibration signal according to the solution results of the physical information multi-dictionary sparse decomposition framework, and extract the fault impact component for fault diagnosis.

[0013] A computer-readable storage medium is used to store computer instructions, and when the computer instructions are executed by a processor, the steps in the above method are completed.

[0014] An electronic device comprises a memory and a processor, and computer instructions stored in the memory and executed on the processor. When the computer instructions are executed by the processor, the steps in the above method are completed.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The generalized dual-pulse response wavelet function constructed by the present invention has better morphological similarity with the fault impulse component, and the proposed generalized dual-pulse response wavelet dictionary has better sparse representation performance. The present invention identifies modal parameters and constructs a modal dictionary through a random subspace working modal analysis method based on the inverse spectrum, and can take into account the influence of the mode during sparse representation. (3) The present invention constructs a physical information multi-dictionary sparse decomposition framework, which can remove the interference of modal components and harmonic components while extracting fault components. The weighted strategy is used to overcome the shortcomings of traditional L1 norm regularization, and can better retain the fault impulse component while suppressing noise.

[0016] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0018] Figure 1 The present invention is a flowchart of a method for extracting machine tool bearing fault features by multi-dictionary sparse decomposition of physical information according to an embodiment. DETAILED DESCRIPTION

[0019] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0020] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0021] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0022] In the absence of conflict, the embodiments in this application and the features in the embodiments may be combined with each other.

[0023] Embodiment 1 A machine tool bearing fault feature extraction method based on multi-dictionary sparse decomposition of physical information, such as Figure 1As shown, the following steps are included: Step 1: Obtain the vibration signal of the machine tool bearing.

[0024] In this step, the vibration signal of the machine tool bearing can be collected by using a sensor, or the vibration signal of the machine tool bearing uploaded by the collection terminal can be obtained.

[0025] Step 2: Build dictionaries for different components in the vibration signal.

[0026] Specifically, a generalized double-pulse response wavelet is designed according to the physical characteristics of the fault impact, and a set of generalized double-pulse response wavelet functions with different forms are generated by adjusting the parameters, and they are combined into a generalized double-pulse response wavelet dictionary to represent the fault components; A harmonic dictionary is constructed using discrete Fourier transform to represent the harmonic components in the vibration signal; The operational modal analysis is performed using a cepstrum-based random subspace method to identify the modal parameters, and a modal dictionary is constructed based on these parameters to represent the modal components in the vibration signal.

[0027] Step 3: Use the original signal and the dictionary constructed in step 2 to establish a multi-dictionary sparse decomposition framework of physical information, introduce the weighted L1 norm, and then use the split augmented Lagrangian shrinkage algorithm to solve the optimization problem to ensure that it can converge to the global optimal solution.

[0028] Step 4: Based on the solution results, remove the modal components and harmonic components in the signal and extract the fault impact components for subsequent fault diagnosis.

[0029] Each step is described in detail below.

[0030] First, in step 2, according to morphological component analysis (MCA), the vibration signal can be regarded as a linear combination of multiple components with different morphologies, and the performance of signal sparse decomposition depends on the similarity between the constructed dictionary and the decomposed components. Therefore, in order to accurately represent the waveform of the impact component, this embodiment designs a generalized double impulse response wavelet dictionary.

[0031] The time domain expression of the generalized double impulse response wavelet is: ; in, , , and is the amplitude parameter, , and are the damping ratio, undamped natural frequency and damped natural frequency respectively, is the time interval between two pulses.

[0032] Through appropriate parameter settings, the constructed dictionary can generate different waveforms, including single impulse response wavelet, single-sided decay wavelet, bilateral symmetric decay wavelet and bilateral asymmetric decay wavelet.

[0033] The generalized dual pulse response wavelet of this embodiment can be used to represent multiple components due to its flexible form, such as the fault transient bearing and gear components, the unbalance component and the friction component. The performance of fault feature extraction is related to the structure of the constructed dual pulse response wavelet basis function. Therefore, the dual pulse response wavelet basis function with higher morphological similarity can better express the process of two collisions.

[0034] In addition, in order to further improve the accuracy of the sparse signal decomposition model, this embodiment parameterizes the size of the spalling zone and introduces it into the construction of the double pulse response wavelet dictionary. The size of the rolling bearing spalling zone can be determined by the pulse duration To estimate, the specific relationship is: ; in is the inner diameter of the bearing. is the rotation frequency. is the diameter of the spalling zone. The pulses appearing in the spalling zone can be expressed as: ; Therefore, the final generalized double impulse response wavelet dictionary is constructed as: ; in represents the convolution operation, is the initial time when the pulse appears.

[0035] By introducing parameters related to the size of the spalling zone, a series of different An overcomplete dictionary of parameterized dual impulse response wavelet basis is designed to ensure that the generated waveform has better morphological similarity with the impulse response. In previous studies, the size of the peeling zone was not considered in the dictionary design process, resulting in a mismatch between the dictionary and the vibration data, thereby reducing the signal-to-noise ratio of the extracted fault features. In this embodiment, the size of the peeling zone is considered in the dictionary design process to extract accurate fault features, so the designed generalized dual impulse response wavelet dictionary has better performance.

[0036] In order to improve the signal-to-noise ratio of the extracted fault features, the modal components can be separated from the vibration signal when performing fault feature extraction. This embodiment uses a random subspace operating modal analysis method based on cepstrum (SSI-OMA) and a stability diagram to identify modal parameters with modal assurance criteria. The identified parameters are used to construct a modal dictionary so that the modal components are sparsely represented and separated from the original signal.

[0037] In step three, a physical information multi-dictionary sparse decomposition framework was constructed. The designed dictionary accurately represented the various components of the vibration signal, thereby extracting the fault components and removing the interference of harmonic components and modal components, achieving high-precision fault feature extraction.

[0038] The generalized dual impulse response wavelet dictionary proposed in this embodiment can be used as a transient impact dictionary because it has better morphological similarity. The proposed modal dictionary can represent modal components. The harmonic components are represented by a harmonic dictionary obtained by discrete Fourier transform (DFT).

[0039] According to the weighting mechanism, a series of weighting coefficients can be used to emphasize the important information of the target and suppress some irrelevant details. In order to improve the performance of sparse decomposition and overcome the shortcomings of the L1 norm, the weighted L1 norm is integrated into the sparse regularization decomposition, thereby constructing a physical information multi-dictionary sparse decomposition framework as shown below: ; in It is with i The framework is a unified form of sparse decomposition, which can decompose the signal into several components and their corresponding dictionaries. It is i A trade-off parameter. It is i The sparse coefficients of the excitation sources. is the weight distribution coefficient, given by p The matrix is ​​composed of , Represents a diagonal matrix , weight coefficient allocation model Defined as: ; in is a stability parameter, which is set to 0.3 in this embodiment.

[0040] The weighted coefficient allocation realizes the selection and retention of fault impact components, thereby improving the performance of fault impact extraction. The proposed physical information multi-dictionary sparse decomposition framework can perform single dictionary sparse decomposition, dual dictionary sparse decomposition and multi-dictionary (i.e. more than two dictionaries) decomposition, so it can be used for single excitation source and multi-excitation source fault diagnosis.

[0041] This embodiment uses the split augmented Lagrangian shrinkage algorithm to obtain the optimal solution of the framework. To facilitate the solution, the physical information multi-dictionary sparse decomposition framework can be rewritten as the sum of two sub-functions, namely: ; ; ; Since the two sub-functions are strictly convex functions, the physical information multi-dictionary sparse decomposition framework is also a strictly convex function. To simplify the derivation process, the following symbols are introduced: , , , ; After variable splitting, the unconstrained optimization problem can be transformed into the following constrained problem, namely: ; Introducing auxiliary variables z After that, the alternating minimization method can be used to deal with the constraint problem. Using the enhanced Lagrangian method, the problem can be rewritten as: ; ; in, μ is the penalty parameter, For the k Auxiliary variables in iterations z , For the k The sparse coefficients in the iteration x , For the k The Lagrange multiplier at iteration .

[0042] The minimization of the above problem is performed alternately between x and z, and the algorithm can converge to the global minimum. x and z When minimizing alternately between , we can obtain the following steps: ; ; ; The alternating minimization problem is converted into subproblems. It is an L1 norm regularized basis pursuit denoising problem, which can be solved by soft thresholding, namely: ; For the k +1 auxiliary variable in iteration z ; Sub-questions is a strictly convex quadratic function, which belongs to the constrained least squares problem. Its minimization process can be expressed as: ; When the dictionary When is a tight frame, the matrix inverse lemma can be used to simplify the solution process, so the minimization process is: ; In order to further simplify the iteration process, the iteration process is slightly adjusted, namely: ; Through the above steps, the variable x and z When applying the algorithm, the amplitude parameter of the wavelet dictionary is set according to the amplitude of the analysis signal, because providing an approximate value helps accelerate convergence. The value of the trade-off parameter is set in (0,1]. A too small value will cause the processed result to retain some redundant components, while a too large value will lose the details of the waveform.

[0043] The computational complexity of the proposed multi-dictionary sparse decomposition framework for physical information is as follows: The algorithm involves some simple computational steps, namely the operator , , , , , and soft threshold operator.

[0044] Among them, the harmonic dictionary D1 is a discrete Fourier transform dictionary, and its computational complexity is O(NlogN) . Generalized Double Pulse Response Wavelet Dictionary D2 and the modal dictionary D3 It involves wavelet transform, and its computational complexity is O(NlogN) The computational complexity of the soft threshold operator is O(N) Therefore, the total computational complexity of the proposed physical information multi-dictionary sparse decomposition framework is O (kNlogN+N) ,in k is the number of cycles. The computational complexity is not high, and less computing resources are used, which can improve efficiency.

[0045] As a typical application example, the modular test bench consists of a motor, a bearing test module, a flywheel module, and a load motor. The damage to the outer raceway of the bearing is caused by electrospark machining (EDM). The sampling frequency of the vibration signal is 64000 Hz, the speed is 1500 r / min, and the theoretical fault characteristic frequency of the outer ring of the bearing is 79.6 Hz. When the outer ring of the bearing fails, a spectrum peak will appear at the fault characteristic frequency in the frequency domain. If the spectrum peak can be extracted in the frequency domain, the outer ring fault of the bearing can be diagnosed.

[0046] In this example, the proposed physical information multi-dictionary sparse decomposition framework is applied to the fault feature extraction of the vibration signal. The parameters of the double impulse response wavelet are set as follows: , , , , , , the trade-off parameters are set as follows: , , (Parameters are optimized by exhaustive grid search of peak signal-to-noise ratio.) A random subspace operating modal analysis method based on cepstrum is used to identify modal parameters.

[0047] In this example, the first four modes of the system and their parameters are identified, and then these parameters are used to construct four elements and form a modal dictionary. Considering that the size of the spalling zone is unknown, the generalized double impulse response wavelet dictionary and the modal dictionary are The value range is 0 to 0.1 (the interval is 0.01), and then they are combined into a redundant dictionary. Through the sparse representation learning process, the most matching sub-dictionary is selected to extract the fault features. The experimental results show that the proposed physical information multi-dictionary sparse decomposition framework can effectively extract the fault impact component. The amplitude of the extracted fault impact component in the time domain and frequency domain is retained, and the spectrum peak is obvious in the frequency domain. The method provided in this embodiment provides an effective solution for rolling bearing fault diagnosis. Of course, in other embodiments, the number of modes identified in the system can be changed according to specific circumstances and requirements, and is not limited to the above values.

[0048] Embodiment 2 A machine tool bearing fault feature extraction system, comprising: A signal acquisition module is configured to acquire a vibration signal of a machine tool bearing; The multi-dictionary construction module is configured to use the physical characteristics of the fault impulse to establish a generalized double-pulse response wavelet dictionary for representing the fault impulse component in the vibration signal, and to use discrete Fourier transform to construct a harmonic dictionary for representing the harmonic component in the vibration signal; use the random subspace method based on the cepstrum to perform working modal analysis to obtain modal parameters, and then construct a modal dictionary for representing the modal components in the vibration signal; A multi-dictionary sparse decomposition module is configured to establish a physical information multi-dictionary sparse decomposition framework based on the original vibration signal and the constructed dictionaries, and introduce a weighted L1 norm, and solve the physical information multi-dictionary sparse decomposition framework until it converges to a global optimal solution; The fault feature extraction module is configured to remove the modal components and harmonic components in the vibration signal according to the solution results of the physical information multi-dictionary sparse decomposition framework, and extract the fault impact component for fault diagnosis.

[0049] Embodiment 3 A computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the steps in the method provided in embodiment 1 are completed.

[0050] Embodiment 4 An electronic device includes a memory and a processor, and computer instructions stored in the memory and executed on the processor. When the computer instructions are executed by the processor, the steps in the method provided in the first embodiment are completed.

[0051] It should be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of an all-hardware embodiment, an all-software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of one or more computer-usable storage media (including but not limited to disk storage, CD - ROM , optical storage, etc.).

[0052] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0053] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0054] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0055] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made by those skilled in the art within the spirit and principle of the present invention without creative labor shall be included in the protection scope of the present invention.

Claims

1. A method for extracting machine tool bearing fault features, characterized in that: The following steps are involved: Obtain vibration signals of machine tool bearings; By using the physical characteristics of fault impulse, a generalized double-pulse response wavelet dictionary is established to represent the fault impulse component in the vibration signal. A harmonic dictionary is constructed using discrete Fourier transform to represent the harmonic component in the vibration signal. The random subspace method based on the cepstrum is used to perform working modal analysis to obtain modal parameters, and then a modal dictionary is constructed to represent the modal components in the vibration signal. Based on the original vibration signal and the constructed dictionaries, and introducing the weighted L1 norm, a physical information multi-dictionary sparse decomposition framework is established, and the physical information multi-dictionary sparse decomposition framework is solved until it converges to a global optimal solution; According to the solution results of the physical information multi-dictionary sparse decomposition framework, the modal components and harmonic components in the vibration signal are removed, and the fault impulse components are extracted for fault diagnosis.

2. A method for extracting machine tool bearing fault features as claimed in claim 1, characterized in that: Using the physical characteristics of fault impulse, the process of establishing a generalized double-pulse response wavelet dictionary includes: establishing the time domain expression of the generalized double-pulse response wavelet, parameterizing the size of the spalling zone, determining the size of the spalling zone by the pulse duration, and introducing it into the construction of the double-pulse response wavelet dictionary.

3. A method for extracting machine tool bearing fault features as claimed in claim 2, characterized in that: The generalized double impulse response wavelet dictionary is constructed as: ； Among them, the pulse appearing in the spalling zone is , represents the convolution operation, To express the initial time of the pulse, the time domain expression of the generalized double pulse response wavelet is: ； in, A i is the amplitude parameter, i=1, 2, 3, 4, , and are the damping ratio, undamped natural frequency and damped natural frequency respectively, is the time interval between two pulses, and , is the inner diameter of the bearing, is the rotation frequency, is the diameter of the spalling zone.

4. A method for extracting machine tool bearing fault features as claimed in claim 1, characterized in that: Based on the original vibration signal and the constructed dictionaries, and introducing the weighted L1 norm, the process of establishing the objective function of the physical information multi-dictionary sparse decomposition framework includes: integrating the weighted L1 norm into the sparse regularization decomposition, and establishing the physical information multi-dictionary sparse decomposition framework as follows: ； in, Di It is with i The framework is a unified form of sparse decomposition, which decomposes the vibration signal into multiple components and corresponding dictionaries. It is i trade-off parameters, It is i The sparse coefficients of the excitation sources, is the weight distribution coefficient, given by p The matrix is ​​composed of , Represents a diagonal matrix , weight coefficient allocation model Defined as: ； in is the stability parameter.

5. A method for extracting machine tool bearing fault features as claimed in claim 4, characterized in that: The physical information multi-dictionary sparse decomposition framework performs single-dictionary sparse decomposition, dual-dictionary sparse decomposition or multi-dictionary decomposition, which can be used for fault diagnosis of single excitation source and / or multiple excitation sources.

6. A method for extracting machine tool bearing fault features as claimed in claim 1, characterized in that: The process of solving the physical information multi-dictionary sparse decomposition framework until converging to the global optimal solution includes: dividing the physical information multi-dictionary sparse decomposition framework into the sum of two functions, performing variable splitting, introducing auxiliary variables, converting the unconstrained optimization problem into a constrained problem, using the alternating minimization method to handle the constraint problem, using the enhanced Lagrangian method to rewrite the constraint problem, and performing a minimization process until converging to the global minimum.

7. A method for extracting machine tool bearing fault features as claimed in claim 6, characterized in that: During the minimization process, the matrix inverse lemma is used to simplify the solution process.

8. A machine tool bearing fault feature extraction system, characterized in that: include: A signal acquisition module is configured to acquire a vibration signal of a machine tool bearing; The multi-dictionary construction module is configured to use the physical characteristics of the fault impulse to establish a generalized double-pulse response wavelet dictionary for representing the fault impulse component in the vibration signal, and to use discrete Fourier transform to construct a harmonic dictionary for representing the harmonic component in the vibration signal; use the random subspace method based on the cepstrum to perform working modal analysis to obtain modal parameters, and then construct a modal dictionary for representing the modal components in the vibration signal; A multi-dictionary sparse decomposition module is configured to establish a physical information multi-dictionary sparse decomposition framework based on the original vibration signal and each constructed dictionary, and introduce a weighted L1 norm, and solve the physical information multi-dictionary sparse decomposition framework until it converges to a global optimal solution; The fault feature extraction module is configured to remove the modal components and harmonic components in the vibration signal according to the solution results of the physical information multi-dictionary sparse decomposition framework, and extract the fault impact component for fault diagnosis.

9. A computer-readable storage medium, characterized in that: Used to store computer instructions, which, when executed by a processor, complete the steps of the method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: The method comprises a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the steps in the method according to any one of claims 1 to 7 are completed.

Images