Digital twinborn guided rolling bearing early fault signal denoising method

Abstract

The invention relates to the technical field of signal denoising, and discloses a digital twinborn guided rolling bearing early fault signal denoising method, which comprises the following steps: S1, inputting the operation condition of an experiment table and bearing parameters into a bearing kinetic equation, and constructing a rolling bearing digital twinborn model in combination with real equipment bearing historical data, a simulation vibration signal in a healthy state is generated; s2, acquiring a small number of actually measured signals to update the digital twin model; wherein a small amount of real-time monitoring data at the initial stage of operation is combined with the generated simulation vibration signal in the healthy state. According to the digital twinning guided rolling bearing early fault signal denoising method, high-quality simulation data is generated through digital twinning, and the digital twinning technology is applied to rolling bearing fault signal denoising for the first time; and the feasibility of performing signal processing by utilizing simulation data which are highly consistent with actual operation and are generated by digital twinning in a complex industrial environment is verified.

Description

Technical Field

[0001] The present invention relates to the technical field of signal denoising, and in particular to a method for denoising early fault signals of rolling bearings guided by digital twins. Background Art

[0002] In modern manufacturing, bearings, as key components of rotating machinery, are widely used in various equipment and systems, such as automobile engines, aerospace equipment, industrial robots, and energy production devices. The stable operation of bearings directly affects the performance, efficiency, and safety of the entire mechanical system. Therefore, early fault diagnosis of bearings, timely detection of potential problems, and ensuring their safe and reliable service have important practical significance and economic value.

[0003] However, the factory environment is complex and changeable. The vibration signals generated during the operation of the bearings are significantly affected by a variety of environmental factors. First, various vibration sources will introduce interference signals. Second, factors such as friction will generate noise. These influences often cause the amplitude and frequency of the vibration signal to deviate from the ideal state, posing significant challenges to fault detection.

[0004] Based on this, it is necessary to propose a digital twin-guided rolling bearing early fault signal denoising method. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a digital twin-guided rolling bearing early fault signal denoising method, which has the advantages of being able to effectively eliminate the noise of early fault signals and highlight fault characteristics, thereby solving the problems raised in the background technology.

[0006] The present invention provides the following technical solution: a method for denoising early fault signals of rolling bearings guided by digital twins, comprising the following steps:

[0007] Step S1, inputting the operating conditions and bearing parameters of the test bench into the bearing dynamics equation and combining them with the historical data of the real equipment bearing, constructing a digital twin model of the rolling bearing, and generating a simulated vibration signal under a healthy state;

[0008] Step S2, obtaining a small amount of measured signals to update the digital twin model; wherein a small amount of real-time monitoring data at the initial stage of operation is combined with the simulated vibration signal generated in a healthy state, the digital twin model is updated by using cosine similarity, and early weak fault features are injected into the updated model;

[0009] Step S3, constructing a denoising algorithm based on the adversarial strategy, inputting the real-time collected measured signal into the generator to generate a preliminary denoised signal, which is then input into the discriminator, and at the same time using the generated noise-free simulation signal containing early weak fault characteristics as a discrimination signal to correct the denoising result;

[0010] In step S4, the value of the loss function gradually decreases and tends to be stable during training, and finally the final denoising result is output.

[0011] Preferably, in step S1, the bearing dynamics equation is constructed and solved in combination with mechanical knowledge to generate a bearing simulation signal; wherein the bearing dynamics equation is:

[0012]

[0013] Preferably, M, C and K represent the total mass, damping coefficient and total contact stiffness of the system respectively (the system includes rollers, bearings and bearing base); Z is the number of rolling elements, i is 1 to Z, and are the outer ring vibration accelerations in the x and y directions at time t; F x and F y are the load components respectively; (where 1.5 represents the load deformation of the ball bearing) is the contact deformation between the i-th rolling element and the raceway, which can be specifically expressed as:

[0014]

[0015] Preferably, x and y are the vibration deformation of the outer ring in the x direction and y direction at time t; c is the radial clearance of the rolling bearing; θ i is the angular position of the i-th rolling element at time t, H0 is the displacement excitation that changes with the fault position. When the rolling bearing components are fault-free, H0 = 0; λ is the effective contact coefficient of the rolling element, which can be expressed as:

[0016]

[0017] Preferably, in step S2, the cosine similarity is expressed as:

[0018] The P SN Represents the measured signal collected from the device, V SN Indicates that P SN Simulation signals under the same working conditions.

[0019] Preferably, the fault feature is injected into the healthy bearing digital twin model by changing H0, which can be specifically expressed as:

[0020] Inner ring fault (H0=H out ),

[0021]

[0022] Outer ring fault (H0=H out ),

[0023]

[0024] Inner and outer ring compound fault,

[0025]

[0026] Preferably, the θ out and θ in are the fault angle positions of the outer ring and the inner ring respectively, L is the fault width of each component of the rolling bearing, r b 、r o and r i They represent the rolling element radius, outer ring radius and inner ring radius respectively, and mod(.) represents the remainder function.

[0027] Preferably, in step S3, the optimization objective function of the denoising algorithm of the adversarial strategy can be expressed as:

[0028]

[0029] Preferably, the P data , P Z and Represent the measured signal, simulated signal and The probability distribution of It can be expressed as ε is a random number, ε∈[0,1]; D(X), D(G(z)) and Represent the measured signal, simulated signal and The probability value of ; E represents expectation; λ is a constant, the default value is 10; ||·||2 represents the second norm.

[0030] Compared with the prior art, the present invention has the following beneficial effects:

[0031] 1. This digital twin-guided rolling bearing early fault signal denoising method applies digital twin technology to rolling bearing fault signal denoising, verifies the feasibility of using digital twins to generate simulation data that is highly consistent with actual operation for signal processing in a complex industrial environment, and improves denoising accuracy and robustness. The proposed method uses digital twins to accurately simulate the operating state of the bearing, effectively reducing the impact of environmental noise and interference on the fault signal, making the denoised signal closer to the actual fault characteristics, and enhancing the reliability of fault detection.

[0032] 2. This digital twin-guided rolling bearing early fault signal denoising method realizes real-time fault diagnosis based on digital twins. By closely integrating the digital twin model with the fault detection system, the operating status of the physical bearing and the digital model are synchronized in real time to ensure timely capture and denoising of fault signals, so that the fault detection system can respond quickly and meet the strict requirements of industrial sites for real-time monitoring and early warning, and enhance the adaptability and robustness of the system. The method combined with digital twin technology can dynamically adapt to different working environments and operating conditions. By continuously updating the digital twin model, real-time monitoring of the bearing status and signal optimization can be achieved, ensuring that efficient and stable fault detection capabilities are maintained in a changing industrial environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0034] Figure 1 It is a flow chart of the denoising method of the present invention;

[0035] Figure 2 Update the architecture diagram for the digital twin model of the present invention;

[0036] Figure 3 This is a structural diagram of the adversarial strategy algorithm of the present invention;

[0037] Figure 4 This is a test result diagram of three indicators of the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] See also Figure 1 , Figure 2 , Figure 3 and Figure 4 , a digital twin-guided rolling bearing early fault signal denoising method, comprising the following steps:

[0040] Step S1, inputting the operating conditions and bearing parameters of the test bench into the bearing dynamics equation and combining them with the historical data of the real equipment bearing, constructing a digital twin model of the rolling bearing, and generating a simulated vibration signal under a healthy state;

[0041] Step S2, obtaining a small amount of measured signals to update the digital twin model; wherein a small amount of real-time monitoring data at the initial stage of operation is combined with the simulated vibration signal generated in a healthy state, the digital twin model is updated by using cosine similarity, and early weak fault features are injected into the updated model;

[0042] Step S3, constructing a denoising algorithm based on the adversarial strategy, inputting the real-time collected measured signal into the generator to generate a preliminary denoised signal, which is then input into the discriminator, and at the same time using the generated noise-free simulation signal containing early weak fault characteristics as a discrimination signal to correct the denoising result;

[0043] In step S4, the value of the loss function gradually decreases and tends to be stable during training, and finally the final denoising result is output.

[0044] As a preferred technical solution of the present invention, step S1, combining mechanical knowledge to construct and solve the bearing dynamics equation to generate a bearing simulation signal; wherein the bearing dynamics equation is:

[0045] M, C and K represent the total mass, damping coefficient and total contact stiffness of the system respectively (the system includes rollers, bearings and bearing base); Z is the number of rolling elements, i ranges from 1 to Z, and are the outer ring vibration accelerations in the x and y directions at time t; F x and F y are the load components respectively; (where 1.5 represents the load deformation of the ball bearing) is the contact deformation between the i-th rolling element and the raceway, which can be specifically expressed as:

[0046] x and y are the vibration deformation of the outer ring in the x and y directions at time t; c is the radial clearance of the rolling bearing; θ i is the angular position of the i-th rolling element at time t, H0 is the displacement excitation that changes with the fault position. When the rolling bearing components are fault-free, H0 = 0; λ is the effective contact coefficient of the rolling element, which can be expressed as:

[0047]

[0048] M=2kg,C=200Ns / m,Z=8,F x =0,F y =12,000N,c=3×10 -3 .

[0049] As a preferred technical solution of the present invention, in step S2, the cosine similarity is expressed as:

[0050] P SN Represents the measured signal collected from the device, V SN Indicates that P SN For the simulation signals under the same working conditions, it is believed that E(P, V)>0.6 represents a reasonable model.

[0051] As a preferred technical solution of the present invention, the fault feature is injected into the healthy bearing digital twin model by changing H0, which can be specifically expressed as:

[0052] Inner ring fault (H0=H out ),

[0053]

[0054] Outer ring fault (H0=H out ),

[0055]

[0056] Inner and outer ring compound fault,

[0057]

[0058] θ out and θ in are the fault angle positions of the outer ring and the inner ring respectively, L is the fault width of each component of the rolling bearing, r b , r0 and r i Represent the rolling element radius, outer ring radius and inner ring radius respectively, mod(.) represents the remainder function, where r b =3.96, r o =19.9, r i =14.65, L=0.1778mm.

[0059] As a preferred technical solution of the present invention, in step S3, the training of the denoising algorithm of the adversarial strategy involves three stages; in the first stage, the real-time collected measured signal is input into the generator to generate a preliminary denoised signal, and the noise-free simulation signal is used to train the discriminator; in the second stage, the preliminary denoised signal from the generator is input into the discriminator, and the discriminator is allowed to determine which are the signals from the generator and which are the noise-free simulation signals, and the distinguished signals from the generator are re-input into the generator for re-denoising and feature extraction; in the fourth stage, the work of the previous stage is repeated until the discriminator cannot distinguish which are the signals from the generator and which are the noise-free simulation signals, wherein the optimization objective function can be expressed as:

[0060]

[0061] As a preferred technical solution of the present invention, P data , P Z and Represent the measured signal, simulated signal and The probability distribution of It can be expressed as ε is a random number, ε∈[0,1]; D(X), D(G(z)) and Represent the measured signal, simulated signal and The probability value of ; E represents expectation; λ is a constant, the default value is 10; ||·||2 represents the second norm.

[0062] As a preferred technical solution of the present invention, step S4, as the value of the loss function tends to be stable, the training gradually converges, and the mean square error between the output denoised signal and the signal with obvious fault characteristics gradually decreases and tends to be stable, and finally the final denoising result is output. The accuracy of the denoising result and the superiority of the method are verified by performing time domain-envelope analysis on the denoising result and comparing the signal characteristic indicators.

[0063] This application uses a bearing test bench to verify the feasibility of the method proposed in this application:

[0064] According to the equipment parameters of a test bench, the digital twin dynamic model of the bearing is constructed and solved, and the PCB352C33 acceleration sensor is used to collect the vibration data of the rotating mechanical bearing. The rolling bearing is 6203-(SKF) and is divided into 3 categories, including 3 bearing states under 2 working conditions, specifically 3 working conditions with a speed of 35Hz and a load of 12kN, a speed of 37.5Hz and a load of 11kN, and a speed of 30Hz and a load of 0.5kN. The bearing has three types of faults: outer ring, inner ring, and mixed inner and outer rings; the sampling frequency is 25.6kHz, the sampling time is 1.28s, and the data is measured once a minute as a sample. The type of vibration signal measured is acceleration. There are 32,678 sampling points in each sampling direction;

[0065] A small amount of measured data under the healthy state obtained at the beginning is used to update the digital twin model. Fault characteristics are injected into the updated digital twin model to generate noise-free, high-fidelity simulation signals under three fault states: inner circle, outer circle, and inner and outer circle, and input them into the adversarial strategy algorithm.

[0066] After the denoising model is trained, the final denoising result is as follows Figure 4As shown in the figure, for the outer ring fault, the denoised signal contains obvious repetitive impacts stimulated by the early outer ring defect of the bearing, and the outer ring fault frequency and its harmonic spectrum peak are obvious. In addition, due to the difference in centrifugal force in the actual system, we can also observe some sidebands near the outer ring fault frequency and its harmonic spectrum peak. For the inner ring fault, the denoised signal contains obvious repetitive impacts stimulated by the early inner ring defect of the bearing, and the inner ring fault frequency and its harmonic spectrum peak are obvious. In fact, due to the rotational modulation, there are more obvious sidebands near the inner ring fault frequency and its harmonic spectrum peak. For inner and outer ring faults, the denoised signal retains the outer ring fault characteristics on the basis of the original signal while highlighting the inner ring fault characteristics.

[0067] In terms of index analysis, for outer ring faults, the kurtosis of the denoised signal increased by 479.42%, the peak factor increased by 170.62%, and the Gini index increased by 165.00% compared with the original signal. For inner ring faults, the kurtosis of the denoised signal increased by 242.80%, the peak factor increased by 138.25%, and the Gini index increased by 114.93% compared with the original signal. For inner and outer ring faults, the kurtosis of the denoised signal increased by 407.25%, the peak factor increased by 154.41%, and the Gini index increased by 168.97% compared with the original signal.

[0068] In the scenario of early weak faults, the digital twin method can solve the problem that the signals collected in actual engineering contain strong noise. The digital twin combined with the adversarial strategy is applied to the field of signal denoising for the first time. In order to verify the effectiveness of the method proposed in this application, the time domain-envelope analysis is combined with three commonly used statistical indicators, namely kurtosis, peak factor and Gini index, to comprehensively evaluate the experimental results provided in the embodiment, which is more robust than the evaluation of a single method.

[0069] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for denoising early fault signals of rolling bearings guided by digital twins, characterized in that: The following steps are involved: Step S1, inputting the operating conditions and bearing parameters of the test bench into the bearing dynamics equation and combining them with the historical data of the real equipment bearing, constructing a digital twin model of the rolling bearing, and generating a simulated vibration signal under a healthy state; Step S2, obtaining a small amount of measured signals to update the digital twin model; wherein a small amount of real-time monitoring data at the initial stage of operation is combined with the simulated vibration signal generated in a healthy state, the digital twin model is updated by using cosine similarity, and early weak fault features are injected into the updated model; Step S3, constructing a denoising algorithm based on the adversarial strategy, inputting the real-time collected measured signal into the generator to generate a preliminary denoised signal, which is then input into the discriminator, and at the same time using the generated noise-free simulation signal containing early weak fault characteristics as a discrimination signal to correct the denoising result; In step S4, the value of the loss function gradually decreases and tends to be stable during training, and finally the final denoising result is output.

2. According to claim 1, a method for denoising early-stage rolling bearing fault signals guided by digital twins is characterized in that: In step S1, the bearing dynamics equation is constructed and solved in combination with mechanical knowledge to generate a bearing simulation signal; wherein the bearing dynamics equation is:

3. The method for denoising early-stage rolling bearing fault signals guided by digital twin according to claim 2 is characterized in that: The M, C and K represent the total mass, damping coefficient and total contact stiffness of the system respectively (the system includes rollers, bearings and bearing base); Z is the number of rolling elements, i is 1 to Z, and are the outer ring vibration accelerations in the x and y directions at time t; F x and F y are the load components respectively; (where 1.5 represents the load deformation of the ball bearing) is the contact deformation between the i-th rolling element and the raceway, which can be specifically expressed as:

4. The method for denoising early-stage rolling bearing fault signals guided by digital twin according to claim 3 is characterized in that: x and y are the vibration deformation of the outer ring in the x direction and y direction at time t; c is the radial clearance of the rolling bearing; θ i is the angular position of the i-th rolling element at time t, H0 is the displacement excitation that changes with the fault position. When the rolling bearing components are fault-free, H0 = 0; λ is the effective contact coefficient of the rolling element, which can be expressed as:

5. The method for denoising early-stage fault signals of rolling bearings guided by digital twins according to claim 1 is characterized in that: In step S2, the cosine similarity is expressed as: The P SN Represents the measured signal collected from the device, V SN Indicates that P SN Simulation signals under the same working conditions.

6. The method for denoising early-stage rolling bearing fault signals guided by digital twin according to claim 1 is characterized in that: The method of injecting fault features into the healthy bearing digital twin model is to change H0, which can be specifically expressed as: Inner ring fault (H0=H out ), Outer ring fault (H0=H out ), Combined fault of inner and outer rings, 7. The method for denoising early-stage rolling bearing fault signals guided by digital twin according to claim 6 is characterized in that: The θ out and θ in are the fault angle positions of the outer ring and the inner ring respectively, L is the fault width of each component of the rolling bearing, r b 、r o and r i They represent the rolling element radius, outer ring radius and inner ring radius respectively, and mod(.) represents the remainder function.

8. The method for denoising early-stage rolling bearing fault signals guided by digital twin according to claim 1, characterized in that: In step S3, the optimization objective function of the denoising algorithm of the adversarial strategy can be expressed as:

9. The method for denoising early-stage rolling bearing fault signals guided by digital twin according to claim 1, characterized in that: The P data , P z and Represent the measured signal, simulated signal and The probability distribution of It can be expressed as ε is a random number, ε∈[0,1]; D(X), D(G(z)) and Represent the measured signal, simulated signal and The probability value of ; E represents expectation; λ is a constant, the default value is 10; ||·||2 represents the second norm.

Images