Adaptive mnad-based harmonic reducer early fault diagnosis method, system, device and storage medium

Abstract

The application discloses a harmonic reducer early fault diagnosis method, system and device based on adaptive MNAD, and a storage medium. The method comprises the following steps: collecting original vibration signals of a harmonic reducer to be diagnosed; determining an optimal key parameter combination of a minimum noise amplitude deconvolution algorithm model (MNAD) based on the original vibration signals by using a multiverse optimization algorithm (MVO); performing signal enhancement processing on the original vibration signals by using the MNAD based on the optimal key parameter combination; performing modal decomposition on the enhanced signals by using an ensemble empirical mode decomposition algorithm (EEMD) to obtain K modal components; constructing a multi-scale fuzzy entropy feature set based on the K modal components; inputting the multi-scale fuzzy entropy feature set into a fault classifier to perform fault recognition and classification, and determining whether the harmonic reducer to be diagnosed has a fault and a fault type. The application can effectively improve the diagnosis result accuracy of the early fault diagnosis of the harmonic reducer.

Description

Technical Field

[0001] This application relates to the field of harmonic reducer technology, and in particular to a method, system, device and storage medium for early fault diagnosis of harmonic reducers based on adaptive MNAD. Background Technology

[0002] A harmonic reducer is a speed reduction device composed of three basic components: a rigid gear (rigid wheel) with an internal gear ring, a flexible gear (flexible wheel) with an external gear ring, and a wave generator. During operation, the rigid wheel is fixed, and the wave generator is driven by a motor. The flexible wheel, acting as the driven wheel, outputs rotation, driving the load. The wave generator is a rod-shaped component with rolling bearings at both ends forming rollers that press against the inner wall of the flexible wheel. The flexible wheel is a thin-walled gear capable of large elastic deformation, with its inner diameter slightly smaller than the total length of the wave generator. The wave generator is the component that causes the flexible wheel to produce controllable elastic deformation. When the wave generator is inserted into the flexible wheel, it forces the cross-section of the flexible wheel to change from a circle to an ellipse. The teeth near the ends of its major axis fully mesh with the teeth of the rigid wheel, while the teeth near the ends of its minor axis completely disengage from the rigid wheel. The teeth in other sections of the circumference are in a transitional state between engagement and disengagement. When the wave generator rotates continuously in a preset direction, the deformation of the flexible wheel changes continuously, causing the meshing state between the flexible wheel and the rigid wheel to change continuously, from meshing, meshing, meshing out, disengaging, and meshing again, repeating in a cycle, thereby realizing the slow rotation of the flexible wheel relative to the rigid wheel in the opposite direction of the wave generator.

[0003] Harmonic reducers are key components widely used in industrial robots, and their health directly affects the operational accuracy and reliability of the entire industrial robot. Currently, in the early fault diagnosis process of harmonic reducers in industrial robots, environmental noise weakens fault feature information, and the values ​​of key parameter combinations (filter length L and periodic noise amplitude ratio ρ) in the Minimum Noise Amplitude Deconvolution Algorithm (MNAD) model can affect the filtering results. In existing technologies, the values ​​of filter length L and periodic noise amplitude ratio ρ in the key parameter combination of MNAD rely on manual experience for selection, and the selected values ​​of filter length L and periodic noise amplitude ratio ρ are often not optimal. Therefore, the performance of early fault diagnosis feature extraction for harmonic reducers is poor, resulting in low accuracy of fault diagnosis results.

[0004] Therefore, improving the accuracy of early fault diagnosis for harmonic reducers is an urgent problem to be solved. Summary of the Invention

[0005] To address the aforementioned technical problems, this application provides a method, system, device, and storage medium for early fault diagnosis of harmonic reducers based on adaptive MNAD, thereby improving the accuracy of early fault diagnosis results for harmonic reducers. This application also provides a system, device, and storage medium for early fault diagnosis of harmonic reducers based on adaptive MNAD, achieving the same technical effect.

[0006] The first objective of this application is to provide an early fault diagnosis method for harmonic reducers based on adaptive MNAD.

[0007] The aforementioned objective of this application is achieved through the following technical solution:

[0008] An early fault diagnosis method for harmonic reducers based on adaptive MNAD, the method comprising the following steps:

[0009] S1, Acquire the original vibration signal of the harmonic reducer to be diagnosed;

[0010] S2, based on the acquired original vibration signal, the Multiverse Optimization (MVO) algorithm is used to adaptively determine the optimal combination of key parameters for the Minimum Noise Amplitude Deconvolution (MNAD) algorithm model, wherein the combination of key parameters includes the filter length L and the periodic noise amplitude ratio ρ.

[0011] S3, based on the optimal combination of key parameters, the minimum noise amplitude deconvolution algorithm model MNAD is used to perform signal enhancement processing on the original vibration signal to obtain the enhanced signal corresponding to the original vibration signal;

[0012] S4, The enhanced signal is modally decomposed using the Integrated Empirical Mode Decomposition (EEMD) algorithm to obtain K modal components, where K is a positive integer;

[0013] S5, construct a multi-scale fuzzy entropy feature set based on the K modal components obtained from mode decomposition;

[0014] S6, input the multi-scale fuzzy entropy feature set into a preset fault classifier for fault identification and classification, and determine whether the harmonic reducer to be diagnosed has a fault and the type of fault.

[0015] Preferably, in step S2, the optimal combination of key parameters for determining the minimum noise amplitude deconvolution algorithm model MNAD based on the acquired original vibration signal using the Multiverse Optimization (MVO) algorithm includes:

[0016] S21, set the initial parameters of the Multiverse Optimization Algorithm (MVO) and the range of values ​​for the filter length L and the periodic noise amplitude ratio ρ in the key parameter combination of the Minimum Noise Amplitude Deconvolution Algorithm (MNAD) model;

[0017] S22, using the envelope spectrum peak factor of the original vibration signal as the fitness function of the multiverse optimization algorithm (MVO), and using the minimum value of the reciprocal of the result of the envelope spectrum peak factor as the objective function value, the multiverse optimization algorithm (MVO) is used to search for the optimal combination of key parameters based on the initial parameters and the fitness function, and the optimal combination of key parameters corresponding to the initial minimum reciprocal value of the envelope spectrum peak factor is obtained.

[0018] S23, update the wormhole existence probability and travel distance rate of the Multiverse Optimization Algorithm (MVO), and then perform differential mutation operation to update the cosmic position;

[0019] S24. Execute steps S22 to S23 iteratively. When the preset maximum number of iterations is reached or the minimum reciprocal value of the envelope spectrum peak factor is obtained, the termination condition for iterative iteration is met, and the filter length L and periodic noise amplitude ratio ρ in the optimal combination of key parameters are obtained.

[0020] Preferably, in the combination of key parameters, the filter length L ranges from [100, 1000], and the periodic noise amplitude ratio ρ ranges from [0.1, 0.9].

[0021] Preferably, in step S3, the step of using the Minimum Noise Amplitude Deconvolution Algorithm (MNAD) to enhance the original vibration signal based on the optimal combination of key parameters to obtain the enhanced signal corresponding to the original vibration signal includes:

[0022] S31, Substitute the filter length L and periodic noise amplitude ratio ρ from the optimal combination of key parameters into the minimum noise amplitude deconvolution algorithm model MNAD;

[0023] S32, the repetitive characteristics of the impact are enhanced by reducing the ratio of periodic noise amplitude, and the periodic pulse component representing the fault in the original vibration signal is extracted to obtain the enhanced signal.

[0024] Preferably, in step S5, constructing a multi-scale fuzzy entropy feature set based on the K modal components obtained from modal decomposition includes:

[0025] S51, calculate the multi-scale fuzzy entropy of the other K-1 modal components besides the residual component among the K modal components;

[0026] S52, construct the multi-scale fuzzy entropy feature set based on the calculated multi-scale fuzzy entropy of the K-1 modal components.

[0027] The second objective of this application is to provide an early fault diagnosis system for harmonic reducers based on adaptive MNAD.

[0028] The second objective of this application is achieved through the following technical solution:

[0029] An early fault diagnosis system for harmonic reducers based on adaptive MNAD, the system comprising:

[0030] The vibration signal acquisition module is used to acquire the raw vibration signal of the harmonic reducer to be diagnosed.

[0031] The optimal parameter combination adaptive determination module is used to adaptively determine the optimal key parameter combination of the minimum noise amplitude deconvolution algorithm model MNAD based on the acquired original vibration signal using the multiverse optimization algorithm (MVO). The key parameter combination includes the filter length L and the periodic noise amplitude ratio ρ.

[0032] The signal enhancement processing module is used to perform signal enhancement processing on the original vibration signal based on the optimal combination of key parameters and the minimum noise amplitude deconvolution algorithm model MNAD to obtain the enhanced signal corresponding to the original vibration signal;

[0033] The signal decomposition module is used to perform mode decomposition on the enhanced signal using the Integrated Empirical Mode Decomposition (EEMD) algorithm to obtain K mode components, where K is a positive integer;

[0034] The feature set construction module is used to construct a multi-scale fuzzy entropy feature set based on the K modal components obtained from modal decomposition.

[0035] The fault identification module is used to input the multi-scale fuzzy entropy feature set into a preset fault classifier for fault identification and classification, and to determine whether the harmonic reducer to be diagnosed has a fault and the type of fault.

[0036] Preferably, the optimal parameter combination adaptive determination module includes:

[0037] The parameter setting unit is used to set the initial parameters of the Multiverse Optimization Algorithm (MVO) and the range of values ​​for the filter length L and the periodic noise amplitude ratio ρ in the key parameter combination of the Minimum Noise Amplitude Deconvolution Algorithm (MNAD) model.

[0038] An initialization processing unit is used to use the envelope spectrum peak factor of the original vibration signal as the fitness function of the multiverse optimization algorithm (MVO), and the minimum value of the reciprocal of the result of the envelope spectrum peak factor as the objective function value. Based on the initial parameters and the fitness function, the multiverse optimization algorithm (MVO) is used to search for the optimal combination of key parameters to obtain the optimal combination of key parameters corresponding to the initial minimum reciprocal value of the envelope spectrum peak factor.

[0039] The parameter update unit is used to update the wormhole existence probability and travel distance rate of the Multiverse Optimization (MVO) algorithm, and then perform differential mutation operation to update the cosmic position;

[0040] The iterative calculation unit is used to perform iterative calculations based on the preset maximum number of iterations and the reciprocal value of the minimum envelope spectrum peak factor to obtain the filter length L and the periodic noise amplitude ratio ρ in the optimal combination of key parameters.

[0041] Preferably, in the combination of key parameters, the filter length L ranges from [100, 1000], and the periodic noise amplitude ratio ρ ranges from [0.1, 0.9].

[0042] Preferably, the signal enhancement processing module includes:

[0043] The parameter input unit is used to substitute the filter length L and the periodic noise amplitude ratio ρ in the optimal combination of key parameters into the minimum noise amplitude deconvolution algorithm model MNAD.

[0044] The signal enhancement unit enhances the repetitive characteristics of the impact by reducing the ratio of periodic noise amplitude, and extracts the periodic pulse component representing the fault from the original vibration signal to obtain the enhanced signal.

[0045] Preferably, the feature set construction module includes:

[0046] The fuzzy entropy calculation unit is used to calculate the multi-scale fuzzy entropy of the other K-1 modal components, excluding the residual component, among the K modal components;

[0047] The fuzzy entropy feature set construction unit is used to construct the multi-scale fuzzy entropy feature set based on the calculated multi-scale fuzzy entropy of the K-1 modal components.

[0048] The third objective of this application is to provide an early fault diagnosis device for harmonic reducers based on adaptive MNAD.

[0049] The aforementioned objective three of this application is achieved through the following technical solution:

[0050] An early fault diagnosis device for harmonic reducers based on adaptive MNAD, comprising:

[0051] The present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the early fault diagnosis method for harmonic reducers based on adaptive MNAD as described in any of the first objectives of this application.

[0052] The fourth objective of this application is to provide a computer-readable storage medium.

[0053] The fourth objective of this application is achieved through the following technical solution:

[0054] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the early fault diagnosis method for harmonic reducers based on adaptive MNAD as described in any one of the first objectives of this application.

[0055] In summary, this application discloses a method, system, device, and storage medium for early fault diagnosis of harmonic reducers based on adaptive MNAD. The method involves: acquiring the original vibration signal of the harmonic reducer to be diagnosed; using the acquired original vibration signal, employing the Multiverse Optimization (MVO) algorithm to adaptively determine the optimal combination of key parameters for the Minimum Noise Amplitude Deconvolution Algorithm (MNAD); using the MNAD algorithm to enhance the original vibration signal based on the optimal combination of key parameters to obtain an enhanced signal; using the Integrated Empirical Mode Decomposition (EEMD) algorithm to perform mode decomposition on the enhanced signal to obtain K mode components; constructing a multi-scale fuzzy entropy feature set based on the K mode components obtained from the mode decomposition; and inputting the multi-scale fuzzy entropy feature set into a preset fault classifier for fault identification and classification to determine whether the harmonic reducer to be diagnosed has a fault and the type of fault.

[0056] This application utilizes the Multiverse Optimization (MVO) algorithm to adaptively determine the optimal combination of key parameters for the Minimum Noise Amplitude Deconvolution (MNAD) algorithm model. The optimal combination of key parameters is then used to enhance the original vibration signal with weak fault characteristic signals via the MNAD algorithm model. Furthermore, the integrated Empirical Mode Decomposition (EEMD) algorithm leverages its advantages in fault feature extraction to perform mode decomposition on the enhanced signal before fault identification and classification. This effectively improves the accuracy of early fault diagnosis results for harmonic reducers. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 This is a flowchart illustrating the early fault diagnosis method for harmonic reducers based on adaptive MNAD in the embodiments of this application.

[0059] Figure 2 This is a flowchart illustrating the process of adaptively determining the optimal combination of key parameters for the Minimum Noise Amplitude Deconvolution Algorithm (MNAD) using the Multiverse Optimization (MVO) algorithm in an embodiment of this application.

[0060] Figure 3 This is a waveform diagram of the original vibration signal (fatigue fracture signal of flexible thin-walled bearing) of the harmonic reducer in the embodiment of this application;

[0061] Figure 4 The embodiment of this application uses the minimum noise amplitude deconvolution algorithm model MNAD to... Figure 3 The waveform of the enhanced signal obtained after signal enhancement processing of the fatigue fracture signal of the flexible thin-walled bearing is shown.

[0062] Figure 5 This is a schematic diagram of the structure of an early fault diagnosis system for a harmonic reducer based on adaptive MNAD in an embodiment of this application;

[0063] Figure 6 This is a schematic diagram of the adaptive determination module for the optimal parameter combination in the embodiments of this application;

[0064] Figure 7 This is a schematic diagram of the structure of an early fault diagnosis device for a harmonic reducer based on adaptive MNAD in an embodiment of this application. Detailed Implementation

[0065] To enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0066] In the embodiments provided in this application, it should be understood that the disclosed methods and systems can be implemented in other ways. The system embodiments described below are merely illustrative. For example, the division of units and modules is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or modules can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, and can be electrical, mechanical, or other forms.

[0067] In addition, each functional unit in the various embodiments of this application can be integrated into a single processor, or each unit can be a separate device, or two or more units can be integrated into a single device; each functional unit in the various embodiments of this application can be implemented in hardware or in the form of hardware plus software functional units.

[0068] Those skilled in the art will understand that all or part of the steps of the following method embodiments can be implemented by program instructions and related hardware. The aforementioned program instructions can be stored in a computer-readable storage medium. When the program instructions are executed, they perform the steps of the following method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), magnetic disks, or optical disks.

[0069] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "a plurality of" or "several" means two or more, unless otherwise explicitly specified.

[0070] This application provides an early fault diagnosis method for harmonic reducers based on adaptive MNAD, such as... Figure 1 As shown, the method may include the following steps:

[0071] S1, Acquire the original vibration signal of the harmonic reducer to be diagnosed;

[0072] Early failures of harmonic reducers mainly include three common types: fatigue fracture of flexible thin-walled bearings, cracking of the flexible wheel, and wear of the rigid wheel tooth surface. When an early failure occurs in a harmonic reducer, the stiffness of the faulty part will change, which will cause the vibration signal of the harmonic reducer to change accordingly during operation. The vibration signal of a faulty harmonic reducer will contain fault characteristic vibrations. Therefore, the fault status and type of the harmonic reducer can be determined by analyzing its vibration signal.

[0073] When early fault diagnosis of a harmonic reducer is required, the first step is to acquire the original vibration signal of the reducer to be diagnosed. Specifically, the original vibration signal of the reducer to be diagnosed can be acquired using an accelerometer installed on the reducer.

[0074] S2. Based on the acquired original vibration signal, the Multiverse Optimization (MVO) algorithm is used to adaptively determine the optimal combination of key parameters for the Minimum Noise Amplitude Deconvolution (MNAD) algorithm model. The key parameter combination includes the filter length L and the periodic noise amplitude ratio ρ.

[0075] To avoid the problem of poor feature extraction performance and low accuracy of fault diagnosis results in the Minimum Noise Amplitude Deconvolution (MNAD) algorithm model, which is caused by manually selecting the key parameter combination (filter length L and periodic noise amplitude ratio ρ) based on experience, this embodiment adopts the Multi-verse Optimization Algorithm (MVO) based on the acquired original vibration signal to adaptively determine the optimal key parameter combination of the MNAD algorithm model, thereby achieving adaptive parameter tuning of the MNAD algorithm model.

[0076] Specifically, in this embodiment, the process of adaptively determining the optimal combination of key parameters for the Minimum Noise Amplitude Deconvolution Algorithm (MNAD) model using the Multiverse Optimization (MVO) algorithm is as follows: Figure 2 As shown, it may include the following steps:

[0077] S21, set the initial parameters of the Multiverse Optimization Algorithm (MVO) and the range of values ​​for filter length L and periodic noise amplitude ratio ρ in the key parameter combination of the Minimum Noise Amplitude Deconvolution Algorithm (MNAD) model.

[0078] Specifically, the initial parameters of the Multiverse Optimization (MVO) algorithm that need to be set include parameters such as the probability of wormhole existence W, the travel distance rate R, and the multiverse population U; the range of values ​​for filter length L and periodic noise amplitude ratio ρ in the key parameter combination is set according to the characteristics of the fault signal of the harmonic reducer. In this embodiment, the range of values ​​for filter length L is set to [100, 1000], and the range of values ​​for periodic noise amplitude ratio ρ is set to [0.1, 0.9].

[0079] S22, using the envelope spectrum peak factor of the original vibration signal as the fitness function of the multiverse optimization algorithm (MVO), and using the minimum value of the reciprocal of the result of the envelope spectrum peak factor as the objective function value, the multiverse optimization algorithm (MVO) is used to search for the optimal combination of key parameters based on the initial parameters and the fitness function, and the optimal combination of key parameters corresponding to the initial minimum value of the reciprocal of the envelope spectrum peak factor is obtained.

[0080] The envelope peak factor is a dimensionless index that considers both periodicity and intensity. The higher the value of the envelope peak factor, the stronger the periodic impact. Therefore, in this embodiment, the minimum value of the reciprocal of the envelope peak factor (i.e., the reciprocal of the maximum value of the envelope peak factor) is selected to determine the optimal combination of key parameters (filter length L and periodic noise amplitude ratio ρ) in the minimum noise amplitude deconvolution algorithm model MNAD.

[0081] S23, update the wormhole existence probability and travel distance rate of the Multiverse Optimization Algorithm (MVO), and then perform differential mutation operation to update the cosmic position;

[0082] S24. Execute steps S22 to S23 iteratively. When the preset maximum number of iterations is reached or the minimum reciprocal value of the envelope spectrum peak factor is obtained, the termination condition for iterative iteration is met, and the filter length L and periodic noise amplitude ratio ρ in the optimal combination of key parameters are obtained.

[0083] In this embodiment, the Multiverse Optimization (MVO) algorithm is used to determine the optimal combination of key parameters for the Minimum Noise Amplitude Deconvolution Algorithm (MNAD). This not only inherits the advantages of the MNAD algorithm in fault feature extraction but also has the function of adaptive parameter tuning, avoiding the problem of poor feature extraction performance caused by manual parameter selection in the MNAD algorithm.

[0084] S3, based on the optimal combination of key parameters, the minimum noise amplitude deconvolution algorithm model MNAD is used to perform signal enhancement processing on the original vibration signal to obtain the enhanced signal corresponding to the original vibration signal;

[0085] After the optimal combination of key parameters is adaptively determined by the Multiverse Optimization (MVO) algorithm, the weak original vibration signal is enhanced by the Minimum Noise Amplitude Deconvolution (MNAD) algorithm model based on the determined optimal combination of key parameters, and the corresponding enhanced signal is obtained.

[0086] In this embodiment, the signal enhancement processing of the original vibration signal based on the minimum noise amplitude deconvolution algorithm model MNAD and the optimal combination of key parameters may specifically include the following steps:

[0087] Substitute the filter length L and the periodic noise amplitude ratio ρ from the optimal combination of key parameters into the minimum noise amplitude deconvolution algorithm model MNAD.

[0088] By reducing the amplitude ratio of periodic noise, the repetitive characteristics of the impact are enhanced, and the periodic pulse component representing the fault in the original vibration signal is extracted to obtain the enhanced signal corresponding to the original vibration signal.

[0089] S4. The integrated empirical mode decomposition algorithm EEMD is used to perform mode decomposition on the enhanced signal to obtain K mode components, where K is a positive integer;

[0090] After the original vibration signal is enhanced to obtain the corresponding enhanced signal, the enhanced signal is further subjected to modal decomposition. In this embodiment, the Ensemble Empirical Mode Decomposition (EEMD) algorithm is specifically used to perform modal decomposition on the enhanced signal, thereby randomly obtaining K modal components.

[0091] S5, construct a multi-scale fuzzy entropy feature set based on the K modal components obtained from mode decomposition;

[0092] After performing mode decomposition on the enhanced signal, a multi-scale fuzzy entropy feature set is constructed based on the K mode components obtained from the mode decomposition.

[0093] In this embodiment, constructing a multi-scale fuzzy entropy feature set may specifically include the following steps:

[0094] Calculate the multiscale fuzzy entropy of the K-1 modal components other than the residual component among the K modal components;

[0095] A multi-scale fuzzy entropy feature set is constructed based on the calculated K-1 modal components.

[0096] It should be noted that when using the Integrated Empirical Mode Decomposition (EEMD) algorithm to decompose a signal, the last residual component is usually excluded, and K-1 multi-scale fuzzy entropies are calculated to assess the signal's complexity or irregularity. This is because the residual components mainly contain low-frequency trend information and contribute little to the signal's dynamic characteristics, potentially masking the influence of other periodic and random components. K-1 intrinsic mode functions (IMFs) can more accurately reflect the signal's inherent dynamic behavior, improving the efficiency and accuracy of complexity analysis.

[0097] S6. Input the multi-scale fuzzy entropy feature set into the preset fault classifier for fault identification and classification, and determine whether the harmonic reducer to be diagnosed has a fault and the type of fault.

[0098] Finally, the multi-scale fuzzy entropy feature set is input into a preset fault classifier for fault identification and classification. The fault classifier determines whether the harmonic reducer to be diagnosed has a fault and the type of fault based on the multi-scale fuzzy entropy features.

[0099] Specifically, in this embodiment, the fault classifier can be an SVM classifier or other existing fault classifiers.

[0100] In summary, the early fault diagnosis method for harmonic reducers based on adaptive MNAD in the above embodiments involves: acquiring the original vibration signal of the harmonic reducer to be diagnosed; using the acquired original vibration signal, employing the Multiverse Optimization (MVO) algorithm to adaptively determine the optimal combination of key parameters for the Minimum Noise Amplitude Deconvolution (MNAD) algorithm model; using the MNAD algorithm model to enhance the original vibration signal based on the optimal combination of key parameters to obtain the enhanced signal corresponding to the original vibration signal; using the Integrated Empirical Mode Decomposition (EEMD) algorithm to perform mode decomposition on the enhanced signal to obtain K mode components; constructing a multi-scale fuzzy entropy feature set based on the K mode components obtained from the mode decomposition; and inputting the multi-scale fuzzy entropy feature set into a preset fault classifier for fault identification and classification to determine whether the harmonic reducer to be diagnosed has a fault and the type of fault, thereby achieving early fault diagnosis of the harmonic reducer.

[0101] This application utilizes the Multiverse Optimization (MVO) algorithm to adaptively determine the optimal combination of key parameters for the Minimum Noise Amplitude Deconvolution Algorithm (MNAD). This adaptively determined optimal parameter combination is then used with the MNAD to enhance the original vibration signal, which has weak fault characteristic signals. Combined with the advantages of the Integrated Empirical Mode Decomposition (EEMD) algorithm in fault feature extraction, the enhanced signal is subjected to mode decomposition before fault identification and classification. This effectively improves the accuracy of early fault diagnosis for harmonic reducers. Case studies demonstrate that the classification accuracy of the early fault diagnosis method for harmonic reducers based on adaptive MNAD in this embodiment can reach 91.5%.

[0102] To verify the effectiveness of the early fault diagnosis method for harmonic reducers based on adaptive MNAD proposed in this application, the inventors built an experimental platform using the BHS-25-XX-U-IV type harmonic reducer as the research object. The experimental platform needed to meet the installation requirements of the BHS-25-XX-U-IV type harmonic reducer. The reducer's flex wheel is a standard length with a hollow top hat shape, and the reducer structure is a complete "U" shape. The high-speed input end of the reducer has its own input shaft. During the experiment, the rigid wheel was fixed, and the output connection of the flex wheel was used.

[0103] The drive system of the experimental platform provides power to the harmonic reducer test piece. Through the motor, frequency converter, and load system, it can simulate various complex working scenarios of the harmonic reducer. The motor is the main part of the reducer drive, and the frequency converter is used to control the motor output speed. The load system connects a magnetic powder brake and a tension controller to the output end of the reducer bearing. The input current of the tension controller is controlled to control the addition of load to the magnetic powder brake.

[0104] In the harmonic reducer fault test bench, this application simulated the normal state, as well as three common harmonic reducer fault types: fatigue fracture of flexible thin-walled bearings, cracking of flexible gears, and wear of rigid gear teeth. Vibration signals were acquired using NI's USB-9234 acquisition card and a data acquisition system built using LabVIEW's DAQ data acquisition module. Two piezoelectric accelerometers were used to test the vibration signals at the X and Y axis positions of the harmonic reducer input, respectively. The sampling frequency was set to 12000Hz, and the number of sampling points was 636000.

[0105] First, the optimal combination of key parameters (filter length L and periodic noise amplitude ratio ρ) of the minimum noise amplitude deconvolution algorithm model MNAD is found by introducing the multiverse optimization algorithm (MVO). The number of search universes is set to 30, the maximum number of iterations is 20, and the dimension is 2. Taking the fatigue fracture signal of the flexible thin-walled bearing of the harmonic reducer as an example, the above method is used to process the fatigue fracture signal of the flexible thin-walled bearing using MNAD. The filter length L is [100, 1000], and the periodic noise amplitude ratio ρ is [0.1, 0.9]. Figure 3 This is a waveform diagram of the original vibration signal (fatigue fracture signal of flexible thin-walled bearing) of the harmonic reducer in the embodiments of this application, that is, the waveform diagram of the vibration signal before enhancement. Figure 4 The waveform of the fatigue fracture signal of the flexible thin-walled bearing after MNAD enhancement is shown. Comparing the signals before and after enhancement, it is found that the impact component of the signal is significantly increased after MNAD processing.

[0106] After performing EEMD decomposition on the four enhanced signals (vibration signals corresponding to four states: normal state, fatigue fracture of flexible thin-walled bearing, crack of flexible wheel and wear of rigid wheel tooth surface), 100 sets of data samples were extracted from each signal, for a total of 400 sets of data, with a data sample length of 1000 points.

[0107] Then, a multi-scale fuzzy entropy is established, and the first 9 modal components are calculated respectively. In the process of finding the fuzzy entropy, the parameters of the multi-scale fuzzy entropy algorithm are first set: embedding dimension m=2, similarity tolerance r=0.15Std (Std represents the standard deviation of the vibration signal), fuzzy power n=2, and time delay tau=1, so as to obtain the multi-scale fuzzy entropy feature set of each mode.

[0108] Seventy sets of data from four types of signal samples were used as training samples, and 30 sets of data were used as test samples. The ratio of training set to test set samples was 7:3, and different sample labels were set.

[0109] The constructed feature vector set is then input into the SVM for classification and recognition, and the classification and recognition rate under different signal states is 91.5%.

[0110] The early fault diagnosis method for harmonic reducers based on adaptive MNAD in this application embodiment can be used for early fault diagnosis of harmonic reducers in strong noise environments.

[0111] like Figure 5 As shown in the figure, this application provides an early fault diagnosis system for harmonic reducers based on adaptive MNAD. The system may include:

[0112] Vibration signal acquisition module 21 is used to acquire the original vibration signal of the harmonic reducer to be diagnosed;

[0113] The optimal parameter combination adaptive determination module 22 is used to adaptively determine the optimal key parameter combination of the minimum noise amplitude deconvolution algorithm model MNAD based on the acquired original vibration signal using the multiverse optimization algorithm MVO. The key parameter combination includes the filter length L and the periodic noise amplitude ratio ρ.

[0114] Signal enhancement processing module 23 is used to perform signal enhancement processing on the original vibration signal based on the optimal combination of key parameters and the minimum noise amplitude deconvolution algorithm model MNAD to obtain the enhanced signal corresponding to the original vibration signal;

[0115] Signal decomposition module 24 is used to perform mode decomposition on the enhanced signal using the Integrated Empirical Mode Decomposition (EEMD) algorithm to obtain K mode components, where K is a positive integer;

[0116] Feature set construction module 25 is used to construct a multi-scale fuzzy entropy feature set based on the K modal components obtained from mode decomposition;

[0117] The fault identification module 26 is used to input the multi-scale fuzzy entropy feature set into the preset fault classifier for fault identification and classification, and to determine whether the harmonic reducer to be diagnosed has a fault and the type of fault.

[0118] like Figure 6 As shown, in one embodiment, the optimal parameter combination adaptive determination module 22 includes:

[0119] The parameter setting unit 221 is used to set the initial parameters of the multiverse optimization algorithm MVO, as well as the range of values ​​for the filter length L and the periodic noise amplitude ratio ρ in the key parameter combination of the minimum noise amplitude deconvolution algorithm model MNAD.

[0120] The initialization processing unit 222 is used to use the envelope spectrum peak factor of the original vibration signal as the fitness function of the multiverse optimization algorithm (MVO), and the minimum value of the reciprocal of the result of the envelope spectrum peak factor as the objective function value. Based on the initial parameters and the fitness function, the multiverse optimization algorithm (MVO) is used to search for the optimal combination of key parameters to obtain the optimal combination of key parameters corresponding to the initial minimum reciprocal value of the envelope spectrum peak factor.

[0121] The parameter update unit 223 is used to update the wormhole existence probability and travel distance rate of the Multiverse Optimization (MVO) algorithm, and then perform differential mutation operation to update the cosmic position;

[0122] The iterative calculation unit 224 is used to perform iterative calculations based on the preset maximum number of iterations and the reciprocal value of the minimum envelope spectrum peak factor to obtain the filter length L and the periodic noise amplitude ratio ρ in the optimal combination of key parameters.

[0123] In one embodiment, in the combination of key parameters, the filter length L ranges from [100, 1000], and the periodic noise amplitude ratio ρ ranges from [0.1, 0.9].

[0124] In one embodiment, the signal enhancement processing module 23 includes:

[0125] The parameter input unit is used to substitute the filter length L and the periodic noise amplitude ratio ρ from the optimal combination of key parameters into the minimum noise amplitude deconvolution algorithm model MNAD.

[0126] The signal enhancement unit enhances the repetitive characteristics of the impact by reducing the amplitude ratio of the periodic noise, and extracts the periodic pulse component representing the fault from the original vibration signal to obtain the enhanced signal corresponding to the original vibration signal.

[0127] In one embodiment, the feature set construction module 25 includes:

[0128] The fuzzy entropy calculation unit is used to calculate the multi-scale fuzzy entropy of the K-1 modal components other than the residual component among the K modal components;

[0129] The fuzzy entropy feature set construction unit is used to construct a multi-scale fuzzy entropy feature set based on the calculated K-1 modal components of the multi-scale fuzzy entropy.

[0130] It should be noted that the early fault diagnosis system for harmonic reducers based on adaptive MNAD in the above embodiments has the same working principle and technical effect as the early fault diagnosis method for harmonic reducers based on adaptive MNAD in the above embodiments, and will not be repeated here.

[0131] like Figure 7As shown, this application provides an early fault diagnosis device 3 for harmonic reducers based on adaptive MNAD. The early fault diagnosis device 3 for harmonic reducers based on adaptive MNAD includes a memory 301, a processor 302, and a computer program 303 stored in the memory 301 and executable on the processor 302. The processor 301 and the memory 302 communicate with each other through a bus 304. When the processor 302 executes the computer program 303, it implements the steps of the early fault diagnosis method for harmonic reducers based on adaptive MNAD as described in the above method embodiment of this application.

[0132] Specifically, the early fault diagnosis device 3 for harmonic reducers based on adaptive MNAD can be an intelligent device with memory and processor, such as an industrial control computer, PC, or smart mobile terminal, or a computer component with memory and processor, such as a CPU or GPU.

[0133] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the early fault diagnosis method for harmonic reducers based on adaptive MNAD as described in the above-described method embodiments of this application.

[0134] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0135] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0136] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly using hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0137] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for early fault diagnosis of harmonic reducers based on adaptive MNAD, characterized in that, The method includes the following steps: S1, Acquire the original vibration signal of the harmonic reducer to be diagnosed; S2, based on the acquired original vibration signal, the optimal combination of key parameters for the Minimum Noise Amplitude Deconvolution Algorithm (MNAD) model is adaptively determined using the Multiverse Optimization (MVO) algorithm, specifically including: S21, set the initial parameters of the Multiverse Optimization Algorithm (MVO), and the range of values ​​for the filter length L and the periodic noise amplitude ratio ρ in the key parameter combination of the Minimum Noise Amplitude Deconvolution Algorithm (MNAD) model; S22, using the envelope spectrum peak factor of the original vibration signal as the fitness function of the Multiverse Optimization (MVO) algorithm, and using the minimum value of the reciprocal of the envelope spectrum peak factor as the objective function value, the MVO algorithm is used to search for the optimal combination of key parameters based on the initial parameters and the fitness function, thereby obtaining the optimal combination of key parameters corresponding to the initial minimum reciprocal value of the envelope spectrum peak factor; S23, update the wormhole existence probability and travel distance rate of the Multiverse Optimization (MVO) algorithm, and then perform a differential mutation operation to update the cosmic position; S24, iteratively execute steps S22~S23. When the preset maximum number of iterations is reached or the minimum reciprocal value of the envelope spectrum peak factor is obtained, the termination condition of the iterative cycle is met, and the filter length L and periodic noise amplitude ratio ρ in the optimal combination of key parameters are obtained. S3, based on the optimal combination of key parameters, the minimum noise amplitude deconvolution algorithm model MNAD is used to perform signal enhancement processing on the original vibration signal to obtain the enhanced signal corresponding to the original vibration signal, specifically including: S31, Substitute the filter length L and periodic noise amplitude ratio ρ from the optimal combination of key parameters into the minimum noise amplitude deconvolution algorithm model MNAD; S32, the repetitive characteristics of the impact are enhanced by reducing the ratio of periodic noise amplitude, and the periodic pulse component representing the fault in the original vibration signal is extracted to obtain the enhanced signal; S4, The enhanced signal is modally decomposed using the Integrated Empirical Mode Decomposition (EEMD) algorithm to obtain K modal components, where K is a positive integer; S5, constructing a multi-scale fuzzy entropy feature set based on the K modal components obtained from mode decomposition, specifically including: S51, calculate the multi-scale fuzzy entropy of the other K-1 modal components besides the residual component among the K modal components; S52, construct the multi-scale fuzzy entropy feature set based on the calculated K-1 modal components; S6, input the multi-scale fuzzy entropy feature set into a preset fault classifier for fault identification and classification, and determine whether the harmonic reducer to be diagnosed has a fault and the type of fault.

2. The method for early fault diagnosis of harmonic reducers based on adaptive MNAD according to claim 1, characterized in that, In the combination of key parameters, the filter length L ranges from [100, 1000], and the periodic noise amplitude ratio ρ ranges from [0.1, 0.9].

3. A harmonic reducer early fault diagnosis system based on adaptive MNAD, characterized in that, The system includes: The vibration signal acquisition module is used to acquire the raw vibration signal of the harmonic reducer to be diagnosed. An adaptive optimal parameter combination determination module is used to adaptively determine the optimal key parameter combination of the minimum noise amplitude deconvolution algorithm model MNAD based on the acquired original vibration signal using the multiverse optimization algorithm (MVO). The adaptive optimal parameter combination determination module includes: The parameter setting unit is used to set the initial parameters of the Multiverse Optimization Algorithm (MVO), as well as the range of values ​​for the filter length L and the periodic noise amplitude ratio ρ in the key parameter combination of the Minimum Noise Amplitude Deconvolution Algorithm (MNAD) model; An initialization processing unit is used to use the envelope spectrum peak factor of the original vibration signal as the fitness function of the Multiverse Optimization (MVO) algorithm, and the minimum value of the reciprocal of the result of the envelope spectrum peak factor as the objective function value. Based on the initial parameters and the fitness function, the MVO algorithm is used to search for the optimal combination of key parameters to obtain the optimal combination of key parameters corresponding to the initial minimum reciprocal value of the envelope spectrum peak factor. The parameter update unit is used to update the wormhole existence probability and travel distance rate of the Multiverse Optimization (MVO) algorithm, and then perform a differential mutation operation to update the cosmic position; The iterative calculation unit is used to perform iterative calculations based on the preset maximum number of iterations and the reciprocal value of the minimum envelope spectrum peak factor to obtain the filter length L and the periodic noise amplitude ratio ρ in the optimal combination of key parameters. A signal enhancement processing module is used to perform signal enhancement processing on the original vibration signal based on the optimal combination of key parameters and the minimum noise amplitude deconvolution algorithm model MNAD to obtain the enhanced signal corresponding to the original vibration signal. The signal enhancement processing module includes: The parameter input unit is used to substitute the filter length L and the periodic noise amplitude ratio ρ from the optimal combination of key parameters into the minimum noise amplitude deconvolution algorithm model MNAD. The signal enhancement unit enhances the repetitive characteristics of the impact by reducing the amplitude ratio of the periodic noise, and extracts the periodic pulse component representing the fault from the original vibration signal to obtain the enhanced signal. The signal decomposition module is used to perform mode decomposition on the enhanced signal using the Integrated Empirical Mode Decomposition (EEMD) algorithm to obtain K mode components, where K is a positive integer; A feature set construction module is used to construct a multi-scale fuzzy entropy feature set based on the K modal components obtained from mode decomposition. The feature set construction module includes: The fuzzy entropy calculation unit is used to calculate the multi-scale fuzzy entropy of the other K-1 modal components, excluding the residual component, among the K modal components; The fuzzy entropy feature set construction unit is used to construct the multi-scale fuzzy entropy feature set based on the calculated multi-scale fuzzy entropy of the K-1 modal components. The fault identification module is used to input the multi-scale fuzzy entropy feature set into a preset fault classifier for fault identification and classification, and to determine whether the harmonic reducer to be diagnosed has a fault and the type of fault.

4. A harmonic reducer early fault diagnosis device based on adaptive MNAD, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the early fault diagnosis method for harmonic reducers based on any one of claims 1-2.

5. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the early fault diagnosis method for harmonic reducers based on adaptive MNAD as described in any one of claims 1-2.

Images