Non-contact acoustic signal detection device and method for bearing fault

Abstract

The present invention discloses a non-contact acoustic signal detection device and method for bearing faults. The method is as follows: a rotating bearing to be tested, wherein a defect thereof interacts with other components of the bearing to generate a time-series phase difference wave packet; the time-series phase difference wave packet is amplified by an acoustic superstructure provided on the bearing to form a vortex pulse acoustic soliton signal; the vortex pulse acoustic soliton signal is output; an acoustic detection mechanism detects the degree of acoustic field focusing of the vortex pulse acoustic soliton signal, determines a final measurement plane, and collects the transmitted acoustic pulse signal; amplifies the collected signal; decomposes and analyzes the detection signal; collects the acoustic vortex signal to obtain a corresponding data set, trains and learns the data set, uses vibration signals at different time scales as input, fuses multiple LSTM neural network models, extracts features and designs a classifier, designs a bearing fault information sequence in the time domain, and constructs a bearing digital twin model for virtualized analysis of fault detection.

Description

Technical Field

[0001] The present invention belongs to the technical field of bearing fault detection, and in particular relates to a non-contact acoustic signal detection device and method for bearing fault. Background Art

[0002] Bearings are crucial components of rotating machinery and are also vulnerable to wear. In various engineering fields, such as transportation, aviation, and precision machine tools, the dynamic performance, service life, and stability of these critical high-speed rolling bearings are highly demanding. Dynamic instabilities such as roller slippage, inner and outer ring wear, bearing runout, and cage twisting and fracture are the root causes of high-speed roller bearing failure and reduced service reliability. The operating condition of a bearing directly impacts the performance of the entire machine and can even have catastrophic consequences. Therefore, it is necessary to identify and analyze the internal bearing failure factors that may affect mechanical operation. Existing methods for detecting bearing fault information include embedded sensors or contact ultrasonic nondestructive testing, or analyzing the composition of the operating oil in the bearing equipment system to determine the possible stage of the bearing failure. However, these contact signal detection methods often have to overcome the difficulties of power supply, weak signals, and complex detection. In addition, the fault information identified is already in the middle and late stages of the bearing under test. In addition, in low-robustness systems such as steam turbines, aircraft gearboxes, and liquid rocket engines, early weak bearing faults can have catastrophic consequences. However, the vibration signals of early bearing faults in such environments are quite weak and easily drowned out by the relatively large low-frequency ambient noise, making it impossible to effectively detect the existence of the fault. In view of this, the present invention proposes a non-contact technical solution for externally detecting the internal structure of the bearing. This solution overcomes the power supply, installation, and detection problems of existing traditional bearing detection technologies. The solution captures, amplifies, and detects the early weak bearing fault signals, thereby detecting rich internal machine operating status information in the first place. This solution can greatly improve the quality and sensitivity of early bearing fault warnings, realize the ability to perceive weak bearing fault signals, and achieve the goal of "early detection and early prevention" for the bearing under test. Summary of the Invention

[0003] In order to solve the above problems existing in the prior art, the present invention provides a non-contact acoustic signal detection device and method for bearing faults.

[0004] To achieve the above object, the technical solution adopted by the present invention is:

[0005] The non-contact acoustic signal detection method for bearing faults is carried out in the following steps:

[0006] S1: The rotating bearing to be tested has its defect interacting with other parts of the bearing to produce a time-series phase difference wave packet;

[0007] S2: The time-sequential phase difference wave packet in step S1 is amplified by the acoustic superstructure provided on the bearing to form a vortex pulse acoustic soliton signal;

[0008] S3: output the vortex pulse acoustic soliton signal generated in step S2;

[0009] S4: The acoustic detection mechanism detects the degree of acoustic field focusing of the vortex pulse acoustic soliton signal, determines the final measurement plane, and collects the transmitted acoustic pulse signal;

[0010] S5: amplify the signal collected in step S4;

[0011] S6: Decomposing and analyzing the detection signal;

[0012] S7: Collect acoustic vortex signals and obtain the corresponding data set through preprocessing and feature extraction. Build an LSTM neural network model to train and learn the data set. Further, adopt a multi-scale learning model, take vibration signals at different time scales as input, fuse multiple LSTM neural network models, extract features and design classifiers, design bearing fault information sequences in the time domain, and build a bearing digital twin model for virtual analysis of fault detection.

[0013] Preferably, in step S1, for the fixed-point defect of the outer ring of the bearing, the rolling element of the bearing interacts with the defect to generate a fixed-point excitation pulse, which is a fixed-point excitation wave; for the indeterminate-point defect of the inner ring of the bearing, the inner ring of the bearing and the rolling element exhibit differential rotation, and the excited defect signal presents a rolling-type timing pulse; for the indeterminate-point defect on the rolling element of the bearing, the rolling element and the inner and outer rings will all generate pulse wave packets, and the excited defect signal presents a dense chasing-type timing pulse.

[0014] Preferably, step S6 is specifically as follows: through FFT transformation or wavelet transformation, the change of energy of each frequency band with the increase of fault cycle is obtained, the frequency band that characterizes the change of damage degree of early bearing fault in the wavelet packet energy spectrum is extracted, the change of bearing damage degree is provided from the energy spectrum of each frequency band, and the possible location and degree of bearing fault are analyzed.

[0015] The present invention also discloses a non-contact acoustic signal detection device for bearing faults, which comprises a motor (41), a coupling (42), a first bearing seat (43), a shaft (44) and a second bearing seat (45), wherein one end of the coupling (42) is connected to the motor shaft of the motor (41), and the other end is connected to the first end of the shaft (44); the first end of the shaft (44) is supported by the first bearing seat (43), the second end of the shaft (44) is inserted into the inner ring of the bearing to be tested (1), and the outer ring of the bearing to be tested (1) is supported by the second bearing seat (45). Support; an acoustic superstructure (3) is provided on the bearing to be tested (1) by the acoustic detection mechanism, and an acoustic detection mechanism (2) is provided in the direction of the central axis of the bearing to be tested (1); a defective portion of the rotating bearing to be tested interacts with other components of the bearing to generate a time-series phase difference wave packet, which is amplified by the acoustic superstructure (3) provided on the bearing to form a vortex pulse acoustic soliton signal; the acoustic detection mechanism (2) detects the degree of acoustic field focusing of the vortex pulse acoustic soliton signal, determines a final measurement plane, and collects the transmitted acoustic pulse signal.

[0016] Preferably, the acoustic detection mechanism (2) comprises an acoustic array module (21) and an acoustic intensity detection module (22), and the acoustic array module (21) and the acoustic intensity detection module (22) are both located in the direction of the central axis of the bearing (1) to be tested.

[0017] Preferably, the acoustic superstructure further comprises a film and a frame, wherein the film is placed on both sides of the frame, and both the film and the frame are made of high molecular polymer materials and have high toughness, plasticity and chemical corrosion resistance.

[0018] Preferably, the acoustic superstructure is a Helmholtz resonant cavity and forms an array arrangement structure, which has a resonance amplification effect on incident sound waves of a certain characteristic frequency.

[0019] Preferably, the acoustic superstructure forms an opening, and a resonance membrane is provided at the opening.

[0020] Preferably, the bearing is a single-row tapered roller bearing with a cage.

[0021] Preferably, the acoustic superstructure (3) is located on the end face of the retainer of the bearing to be tested (1); or, the acoustic superstructure (3) is located on the end face of the bearing outer ring in the direction away from the shaft (44); or, the acoustic superstructure (3) is located on the end face of the bearing inner ring in the direction away from the shaft (44).

[0022] Preferably, the superacoustic structure is located on the end face of the retaining frame, with the opening direction pointing in the direction of the anti-shaft of the bearing.

[0023] Preferably, the number of the acoustic superstructure resonant cavities is not less than the number of the bearing rollers, and corresponds one-to-one to the rollers on the projection of the bearing end surface.

[0024] Preferably, the bearing testing device is further provided with a noise reduction and vibration reduction platform to reduce disturbances from the external environment.

[0025] The present invention provides a non-contact acoustic signal detection device for bearing faults, which is placed on a test bench for noise reduction and vibration reduction. The drive motor is started, and the bearing to be tested is driven to rotate by the shaft. The defects in the bearing interact with the rollers to excite a time-series pulse wave packet, which is further acquired and amplified by the acoustic super surface structure of the retaining frame to construct a far-field acoustic vortex transmission. The acoustic detection module also includes an acoustic array module and an acoustic intensity detection module. The acoustic array module and the acoustic intensity detection module are both located in the output axis direction of the detection bearing. The acoustic intensity detection module detects the focusing degree of the sound field at the far end to determine the final measurement plane. The acoustic array module is used to collect the sound field information at the measurement plane and to generate a signal for the signal. The signal is decomposed and analyzed, that is, through Fourier transform or wavelet transform, the change of energy of each frequency band with the increase of fault cycle is obtained, and the frequency band that characterizes the change of damage degree of early bearing fault in the wavelet packet energy spectrum is extracted. The change of bearing damage degree is provided from the energy spectrum of each frequency band. Finally, a multi-scale learning model is constructed, that is, a bearing fault detection model based on feature extraction and multi-layer perceptron. The FFT spectrum is used to extract the characteristic frequencies of faults in different positions of the bearing. The LSTM neural network model and identification method are applied to conduct experimental research on the transmission signal to improve the confidence of model detection and the high-sensitivity perception ability of early bearing faults.

[0026] Based on the roller defect contact principle, the present invention constructs a bearing roller timing phase difference model, and uses the acoustic superstructure set on the bearing end face to construct an undistorted acoustic vortex transmission channel for bearing internal fault information, providing an enhanced signal for external detection and processing for far-field non-contact detection of bearing faults, which helps to achieve the engineering goal of "early detection and early prevention" of bearing faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 This is a schematic structural diagram of a detection device according to a preferred embodiment of the present invention;

[0028] Figure 2 Schematic diagram of the phase difference of the self-excitation timing of the bearing to be tested according to the present invention;

[0029] Figure 3 A schematic diagram of the acoustic superstructure of the bearing retainer structure according to the present invention;

[0030] Figure 4 This is a schematic diagram of an acoustic superstructure cavity according to a preferred embodiment of the present invention.

[0031] Figure 5 This is a schematic diagram of an acoustic superstructure cavity according to a preferred embodiment of the present invention.

[0032] Figure 6This is a schematic diagram of an acoustic superstructure cavity according to a preferred embodiment of the present invention.

[0033] Figure 7 This is a schematic diagram of an acoustic superstructure cavity according to a preferred embodiment of the present invention.

[0034] Figure 8 Schematic diagram of the phase difference focusing of the acoustic signal detected by the present invention Figure 1 ;

[0035] Figure 9 Schematic diagram of the phase difference focusing of the acoustic signal detected by the present invention Figure 2 ;

[0036] Figure 10 This is a simulation diagram of the roller excitation timing phase difference model of the present invention;

[0037] Figure 11 This is a flow chart of a detection method according to a preferred embodiment of the present invention;

[0038] Figure 12 This is the exploded view of the bearing to be tested;

[0039] Figure 13 This is a partial structural diagram of a detection device according to a preferred embodiment of the present invention.

[0040] In the figure: 1-bearing to be tested; 2-acoustic detection mechanism; 21-acoustic array module; 22-acoustic intensity detection module; 3-acoustic superstructure; 31-resonant cavity one; 32-resonant cavity two; 33-resonant cavity three; 34-resonant cavity four; 341-opening; 342-film; 4-bearing detection device; 41-motor; 42-coupling; 43-bearing seat one; 44-shaft; 45-bearing seat two; 5-roller; 6-defect; 7-signal to be tested; 8-shaft; 9-chassis; F-signal transmission direction. DETAILED DESCRIPTION

[0041] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numbers in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Instead, they are intended to serve as preferred examples consistent with certain aspects of the present application, as detailed in the appended claims.

[0042] like Figure 1As shown, the non-contact acoustic signal detection device for bearing faults in this embodiment includes a bearing drive mechanism 4 and an acoustic detection mechanism 2. The bearing drive mechanism 4 includes a motor 41, a coupling 42, a bearing seat 1 43, a shaft 44, and a bearing seat 2 45. One end of the coupling 42 is connected to the motor shaft of the motor 41, and the other end is connected to the shaft 44. The end of the shaft 44 close to the motor 41 is also connected to the bearing seat 1 43, and the bearing seat 1 43 supports this end of the shaft 44; the other end of the shaft 44 is installed with the bearing 1 to be tested, the shaft 44 penetrates into the inner ring of the bearing 1 to be tested, and the outer ring of the bearing 1 to be tested is connected to the bearing seat 2 45.

[0043] An acoustic detection mechanism 2 is provided in the direction of the central axis of the bearing 1 to be tested. In this embodiment, the acoustic detection mechanism 2 includes an acoustic array module 21 and an acoustic intensity detection module 22. Both the acoustic array module 21 and the acoustic intensity detection module 22 are located in the direction of the central axis of the bearing 1 to be tested. Both the acoustic array module 21 and the acoustic intensity detection module 22 can adopt existing technologies. The acoustic intensity detection module is used to detect the intensity of the sound field, and the acoustic array module further detects and analyzes the output acoustic signal. An acoustic superstructure 3 is also provided on the bearing 1 to be tested. Further, the bearing 1 to be tested in this embodiment includes a retaining frame, and an acoustic superstructure 3 is formed on the retaining frame. The acoustic superstructure is similar to a Helmholtz resonant cavity, and its function is to sense and amplify the acoustic emission signal at the excitation point of the acoustic fault. In more detail, the bearing 1 to be tested in this embodiment is a single-row tapered roller bearing, which includes an outer ring, an inner ring, and a group of tapered rollers, and an inner ring assembly wrapped by a basket-shaped retaining frame. The rollers are guided by the large inner ring ribs, and the inner ring pipe surface, the outer ring pipe surface, and the extension of the roller rolling surface are substantially the same. The long lines intersect at one point on the bearing axis; its unique centerline far-field extension line focuses, and during the fault collision process between the bearing roller and the inner and outer rings, its far-field focusing characteristics help to form a focused sound field; in addition, the main function of the cage is to guide and isolate the roller and rotate synchronously with the roller assembly and remain relatively stationary. The acoustic Doppler effect can be effectively avoided for the roller-excited acoustic signal passing through the cage, and an acoustic sensing superstructure transmission channel is constructed on the end face of the cage to sense and transmit the fault phonon information inside the bearing, thereby improving the sensitivity and confidence of the far-field detection of the bearing fault signal.

[0044] From slight surface damage in the early stage of bearings to plastic deformation and even rupture and failure of bearing components, when bearings encounter these defects during operation, transient elastic stress waves will be generated and energy will be released, such as Figure 2As shown in the figure, when the roller hits the local fault of the outer ring or inner ring, or when the fault on the roller passes through the outer ring or inner ring, a high-frequency resonant pulse sequence will be generated; for the outer ring defect, each roller and the outer ring defect interact in sequence to generate a phonon pulse wave packet sequence, and the generated wave packet sequence has a certain wave path difference, resulting in the sound wave from each roller and the defect having a discrete phase difference, thereby generating an acoustic vortex; for the defect on the roller, the roller rotation will randomly interact with the inner ring and outer ring respectively to generate a high-frequency pulse sound wave sequence. At this time, the resonance amplitude of the defect and the inner and outer rings is different, which will generate a vortex pulse sequence with alternating amplitudes; for the inner ring defect, the inner ring will maintain a relative differential rotation with the roller and cage. The contact between the roller and the inner ring defect is a catch-up problem, which will generate a catch-up pulse wave packet, and then generate a frequency comb vortex pulse sequence; then, using envelope demodulation technology, the fault characteristic frequency of the bearing can be analyzed, and the characteristics of the bearing under multiple working conditions, multiple directions, multiple loads, multiple fault levels, and multiple fault points can be known.

[0045] Furthermore, the acoustic superstructure can be embedded in the bearing to be tested, such as Figure 3-7 As shown, the acoustic superstructure 3 on the cage is an acoustic supersurface structure, which is composed of tiny structural units. By controlling and modulating the incident sound wave, the amplitude, phase and propagation direction of the sound wave are controlled. Specifically, the acoustic cage supersurface structure is constructed to highly sensitively sense the incident sound wave and resonate the phonon signal. The acoustic superstructure 3 of this embodiment is located on the end face of the bearing cage away from the shaft 44. The acoustic superstructures on the cage are arranged into tiny resonant cavities, such as Figure 3 As shown, the resonance cavity is arranged corresponding to each roller to sense the elastic wave excited by each collision between the roller and the defect and resonate with it to achieve undistorted transmission of the fault signal. Figure 7 It is a preferred resonant cavity, including a film 342 and a frame. The film is placed on both sides of the frame. The film and the frame are both made of high molecular polymer materials. The resonant cavity will achieve acoustic amplification of the vibration signal at the same resonant frequency.

[0046] The idea of ​​the present invention is to set up a certain microcavity structure, which is shaped like a resonant cavity. Its purpose is to sense the acoustic signal generated by the vibration of the bearing fault. Because the existing bearing fault detection methods are all contact or sensor type detection, the present invention uses the acoustic signal vibration to sense and resonate and amplify the acoustic superstructure for transmission. Therefore, for different types of bearings in different service environments, their fault types are also different. During the operation of the bearing, the outer ring is embedded in the chassis and fixed, and the inner ring rotates differentially with the rollers. Therefore, the technical effects produced by the acoustic superstructure set on the inner and outer rings or the retaining frame are also very different.

[0047] Specifically speaking, early bearing faults are relatively weak, mixed and not concentrated enough, and are full of nonlinear characteristics and many uncertainties. The multi-degree-of-freedom, strong nonlinearity, chaos, time-varying, noise, multi-scale in space and time, and uncertainty in initial conditions in the rotor system will cause unpredictable disorder to the effective output fault information; early bearing faults mostly appear in the form of small pulses, mixed with a large number of invalid signals, so Figure 4-7 As shown, by constructing an acoustic surface structure array shaped like a Helmholtz resonant cavity, the acoustic perception frequency of the system can be controlled within a specific range, which will effectively eliminate some interference frequencies in the mixed fault frequency and can more efficiently perceive and transmit the characteristic frequency of the bearing fault.

[0048] In detail, according to the rotation characteristics of the rotating bearing, for the fixed-point defects on the outer ring of the bearing, the roller will interact with the defect to generate a fixed-point excitation pulse, which is a fixed-point excitation wave; for the indeterminate-point defects on the inner ring, the inner ring and the roller rotate at a differential speed, and the excited defect signal presents a rolling timing pulse; for the indeterminate-point defects at the roller, the roller and the inner and outer rings will generate pulse wave packets, and the excited defect signal will present a dense chasing timing pulse. For the latter two types, the following can be constructed: Figure 8 and Figure 9 The ring pulse signal timing phase difference model array is shown.

[0049] In this embodiment, the rollers in the bearing ring are abstracted as a circular β-FPU fault particle chain, and a Cartesian coordinate system is established for analysis and calculation. The coordinate origin is taken from the intersection of the plane where the rollers are located and the center line of the bearing. Among them, the rectangular coordinates of the nth roller are in are their respective initial phases, and a is the radius of the circle where the roller is located, Figure 8 The right side of the middle represents the signal detection generation plane, R0 is the distance from the roller plane to the detection plane, R n It represents the distance from the nth roller to the focal point of the detection plane, that is, the single-point excitation sound pressure is:

[0050]

[0051] Since there is a phase difference between different excitations, the sound pressure of each transducer at the detection point can be superimposed to obtain the detection point The sound pressure (r is the radius of the circle where the detection point is located) can be expressed as:

[0052]

[0053] Where A0 is the maximum value of the sound source amplitude, which is related to the roller speed and the size of the fault, k represents the wave number, that is, k = ω / c, c represents the speed of sound,

[0054] The acoustic wave sequence at the center of the annular bearing roller has a discretely varying path difference, which results in a discretely varying phase difference in the acoustic wave radiated from each roller. The soliton signal excited by the defect on each roller is converted into an acoustic soliton signal after the phase difference is rotated and then transmitted through the cage acoustic channel. This can be controlled to form an acoustic vortex with a phase spiral distribution, and a leaky wave antenna structure similar to a ring waveguide is constructed to generate an acoustic vortex. A nonlinear model of acoustic focusing scalar and vector optimization is established. Phase difference excitation is performed from the annular array to the center line of the detection plane according to a certain focusing law. The synthetic wavefront of the acoustic beam at each excitation point has a curvature center point P, which realizes the far-field focusing of the fault phonon. The excitation delay relationship of each unit roller is: where t n 、T n and τ n are the fault excitation delays on the bearing inner ring, outer ring and roller, respectively, v s represents the speed ratio of the roller to the inner ring, s is the roller rotation speed, and n represents the number of rollers). The hysteresis phase difference is determined by the acoustic path difference between the rotating bodies. Finally, the signal components are analyzed in the host computer software. Through time-frequency conversion, the change of energy of each frequency band with the increase of fault cycle is obtained. The frequency band that represents the change of damage degree of early bearing fault in the wavelet packet energy spectrum is extracted. The change of damage degree is provided from the energy spectrum of each frequency band. The analysis further obtains the specific fault location information of the bearing. Figure 10 This is the time-series phase-difference vortex model in the COMSOL acoustic simulation software under this method. The signal shape spirals forward like a spaghetti roll.

[0055] Combining the above content, such as Figure 1 and Figure 11 As shown, during the test, the device of the present invention is placed on a noise reduction and vibration reduction test bench, the drive motor is started, and the bearing to be tested is driven to rotate by the shaft. The defects in the bearing interact with the roller to excite a time-series pulse wave packet, which is further acquired and amplified by the acoustic super surface structure of the retaining frame to construct a far-field acoustic vortex transmission. The detection method also includes an acoustic array module and an acoustic intensity detection module. The acoustic array module and the acoustic intensity detection module are both located in the output axis direction of the detection bearing. The acoustic intensity module detects the focusing degree of the sound field at the far end to determine the final measurement plane. The acoustic array module is used to collect the sound field information at the measurement plane and decompose and analyze the signal in the upper computer software. Finally, a multi-scale learning model is constructed to improve the confidence of model detection and the high-sensitivity perception capability of early bearing faults.

[0056] like Figure 11 In the embodiment of this method, the following steps are mainly used to complete the far-field acoustic signal detection of the bearing to be tested:

[0057] S1: According to the rotation characteristics of the rotating bearing, for fixed-point defects on the outer ring of the bearing, the roller will interact with the defect to generate fixed-point excitation pulses, which are fixed-point excitation waves; for irregular-point defects on the inner ring, the inner ring and the roller rotate differentially, and the excited defect signal presents a rolling timing pulse; for irregular-point defects at the roller, the roller and the inner and outer rings will all generate pulse wave packets, and the excited defect signal will present a dense chasing timing pulse (in detail, these three excitation modes are important bases for subsequent identification of defect locations and quantities).

[0058] S2: The various excitation modes in step S1 are amplified by the (acoustic superstructure) provided on the bearing end surface;

[0059] S3: Output step S2 generates three soliton transmission signals: a linear fixed-point excitation pulse signal, a rotary pulse vortex signal, and a rotary vortex double pulse signal.

[0060] S4: Sound intensity signal detection Sound intensity focusing plane, the acoustic array module collects the transmitted acoustic pulse signal. Specifically, the roller in the bearing ring is abstracted as a circular β-FPU fault particle chain, and a Cartesian coordinate system is established for analysis and calculation. The coordinate origin is taken from the intersection of the roller plane and the bearing centerline. Among them, the rectangular coordinate of the nth roller is in are their respective initial phases, and a is the radius of the circle where the roller is located, Figure 8 The right side of the middle represents the signal detection generation plane, R0 is the distance from the roller plane to the detection plane, R n It represents the distance from the nth roller to the focal point of the detection plane, that is, the single-point excitation sound pressure is:

[0061]

[0062] Since there is a phase difference between different excitations, the sound pressure of each transducer at the detection point can be superimposed to obtain the detection point The sound pressure (r is the radius of the circle where the detection point is located) can be expressed as:

[0063]

[0064] Where A0 is the maximum value of the sound source amplitude, which is related to the roller speed and the size of the fault, k represents the wave number, that is, k = ω / c, c represents the speed of sound,

[0065] The acoustic wave sequence at the center of the annular bearing roller has a discretely varying path difference, which results in a discretely varying phase difference in the acoustic wave radiated from each roller. The soliton signal excited by the defect on each roller is converted into an acoustic soliton signal after the phase difference is rotated and then transmitted through the cage acoustic channel. This can be controlled to form an acoustic vortex with a phase spiral distribution, and a leaky wave antenna structure similar to a ring waveguide is constructed to generate an acoustic vortex. A nonlinear model of acoustic focusing scalar and vector optimization is established. Phase difference excitation is performed from the annular array to the center line of the detection plane according to a certain focusing law. The synthetic wavefront of the acoustic beam at each excitation point has a curvature center point P, which realizes the far-field focusing of the fault phonon. The excitation delay relationship of each unit roller is: where t n 、T n and τ n are the fault excitation delays on the bearing inner ring, outer ring and roller, respectively, v s represents the speed ratio of the roller to the inner ring, s is the roller rotation speed, and n represents the number of rollers). The hysteresis phase difference is determined by the acoustic path difference between the rotating bodies. Finally, the signal components are analyzed in the host computer software. Through time-frequency conversion, the change of energy of each frequency band with the increase of fault cycle is obtained. The frequency band that represents the change of damage degree of early bearing fault in the wavelet packet energy spectrum is extracted. The change of damage degree is provided from the energy spectrum of each frequency band. The analysis further obtains the specific fault location information of the bearing. Figure 10 It is a time-series phase difference vortex model in COMSOL acoustic simulation software under the method of the present invention, and the signal shape spirals forward like a spaghetti roll.

[0066] S5: The waveform signal collected in step S4 is further amplified by a digital power amplifier circuit.

[0067] S6: Decompose and analyze the detected waveform to extract the frequency bands in the wavelet packet energy spectrum that characterize the changes in the damage degree of early bearing faults. Through FFT or wavelet transform, the energy changes of each frequency band as the fault cycle increases are obtained. That is, the energy value, mean square error value, and kurtosis value obtained by decomposing the detected acoustic signal in the time domain and frequency domain are used as signal features. The energy spectrum of each frequency band is used to preliminarily analyze the changes in the bearing damage degree.

[0068] More specifically, the signals induced by local abnormalities in mechanical equipment often have singularity, which manifests as irregular transient structures such as mutations and sharp points. When different types of bearing faults occur, their frequency bands increase to varying degrees, and the center of energy accumulation will also change accordingly. Time domain analysis can determine the trend of bearing fault evolution, and frequency domain analysis can determine the location and extent of bearing faults. Furthermore, through the wavelet transform method, while improving the signal-to-noise ratio, a fairly high time resolution is maintained. The singularity of the fault characteristic signal represents the existence of the fault. The maximum value of the mutation point of the fault characteristic signal increases with the increase of scale under the wavelet transform, while the modulus maximum of the background noise decays rapidly with the increase of scale. That is, after removing the environmental noise to the greatest extent, the bearing fault behavior is analyzed to obtain a preliminary understanding of the degree of bearing damage.

[0069] In order to clarify the bearing failure analysis of various types in various stages and environments, it is necessary to build a data set training method to further collect the possible locations and degrees of bearings serving in different environments.

[0070] S7: Continue to collect the acoustic vortex signals transmitted by the above process, and obtain the corresponding data sets through preprocessing and feature extraction. Then, build an LSTM neural network model to train and learn these data sets. Further, adopt a multi-scale learning model, use vibration signals at different time scales as input, fuse multiple LSTM neural network models, extract features and design classifiers. In this way, the bearing fault information sequence can be designed in the time domain, in order to build a digital twin model for bearing fault detection, so as to achieve advanced and highly sensitive perception of early bearing faults at a higher level.

[0071] More specifically, this step aims to improve the detection robustness of the model process method and provide a design strategy for bearing fault data models under different service environments. This involves repeatedly testing and modifying variables (for different bearing fault types and service environments) to detect and analyze bearing faults in different service environments. This allows for the construction of a time-domain fault information sequence and digital twin model for the tested bearing under different service environments. More specifically, by integrating traditional mechanical systems with computer simulation, a digital system consisting of physical and mathematical models is established. This system utilizes a dual-tower architecture to compare and calibrate test and simulation data to improve the accuracy and robustness of bearing fault detection. The digital twin model enables virtual analysis of existing mechanical systems in a time- and cost-effective manner, thereby reducing system operating costs and improving efficiency. Based on acoustic signal data collected by sensors, this digital twin model analyzes the bearing fault information sequence using algorithms such as LSTM neural networks and multi-scale learning models. This information is then compared with digital simulation results, enabling proactive fault detection of early-stage bearing failures.

[0072] In general, the present invention provides a new non-contact acoustic signal detection method for bearing faults, such as Figure 11 and Figure 12 As shown in the figure: Based on the roller defect contact principle, a bearing roller time-series phase difference model is constructed. With the help of the acoustic resonance cavity set on the bearing end face, an undistorted acoustic vortex transmission channel for bearing internal fault information is constructed, which provides an enhanced signal for external detection and processing for far-field non-contact detection of bearing faults, and realizes high-sensitivity detection of bearing fault signals in the far field without contacting the bearing. Further, fault information sequences and digital twin models of different bearing types in the time domain can be constructed to realize realistic bearing fault simulation in a virtual environment, thereby realizing advanced and highly sensitive perception of early bearing faults at a higher level.

[0073] Optionally, this detection device is a specific detection example to embody this detection method, and is not limited to the device parts mentioned in the present invention.

[0074] Optionally, the resonance cavity 1 31, the resonance cavity 2 32 and the resonance cavity 3 33 mentioned in the present invention are all preferred embodiments of the acoustic superstructure 1 3A. Similar to the resonance cavity 4 34, the resonance cavity 4 34 includes an opening 341 and a film 342. The film 342 is placed at both sides of the opening 342 to form a resonance chamber.

[0075] Optionally, the cavity mentioned in the present method may be constructed not only on the retaining frame but also on the end face of the outer ring or the inner ring, which can also achieve the purpose of fault sound resonance perception and transmission.

[0076] Alternatively, as Figure 13 As shown, the bearing to be tested in the present invention can be not only a tapered roller bearing with a cage, but also a common bearing, and can be used for common industrial bearing fault detection.

[0077] Optionally, the cavity mentioned in the present invention is not limited to its circular or square shape, and other cavities that can be used for acoustic resonance are also suitable for the construction concept of this method.

[0078] Optionally, the arrangement of the retaining frame superstructure may also be a series of microcavity grooves shaped like water vortices, and the incident acoustic wave passes through the microcavity to achieve amplified transmission of the acoustic resonance phonon signal.

[0079] The above content describes the operating principles, features and beneficial effects of the present invention. This document uses more reference numerals in the figures, but does not exclude the possibility of using other terms. Relevant personnel in this field can understand from the above content that the above content does not limit the present invention. The above embodiments and descriptions describe the basic principles and features of the present invention. Under the premise of being consistent with the concept of the present invention, the present invention can also be subjected to various changes and improvements, and these improvements should fall within the scope of protection claimed by the present invention.

Claims

1. Non-contact acoustic signal detection method for bearing faults, characterized by Proceed as follows: S1. The rotating bearing to be tested has its defect interacting with other parts of the bearing to produce a time-series phase difference wave packet; S2, the time sequence phase difference wave packet in step S1 is amplified by the acoustic superstructure provided on the bearing to form a vortex pulse acoustic soliton signal; S3, outputting the vortex pulse acoustic soliton signal generated in step S2; S4, the acoustic detection mechanism detects the degree of acoustic field focusing of the vortex pulse acoustic soliton signal, determines the final measurement plane, and collects the transmitted acoustic pulse waveform signal; S5, amplifying the waveform signal collected in step S4; S6. Decomposing and analyzing the detection waveform signal; S7. Continuously collect vortex pulse acoustic soliton signals and obtain the corresponding data set through preprocessing and feature extraction. Build an LSTM neural network model to train and learn the data set. Furthermore, adopt a multi-scale learning model, use vibration signals at different time scales as input, fuse multiple LSTM neural network models, extract features, and design a classifier. Design a bearing fault information sequence in the time domain, and build a bearing digital twin model for virtualized analysis of fault detection. The acoustic superstructure is a Helmholtz resonant cavity, and a plurality of acoustic superstructures are arranged in an array on the bearing to be tested; The bearing to be tested is a single-row tapered roller bearing with a cage; the acoustic superstructure is located on the end face of the cage of the bearing to be tested; or, the acoustic superstructure is located on the end face of the bearing outer ring in the direction away from the shaft; or, the acoustic superstructure is located on the end face of the bearing inner ring in the direction away from the shaft; The acoustic superstructure corresponds one-to-one to the projection of the roller of the bearing to be tested on the bearing end surface.

2. The non-contact acoustic signal detection method for bearing faults according to claim 1, characterized in that: In step S1, for the fixed-point defects on the outer ring of the bearing, the rolling elements of the bearing interact with the defect to generate a fixed-point excitation pulse, which is a fixed-point excitation wave; for the non-fixed-point defects on the inner ring of the bearing, the inner ring of the bearing and the rolling elements exhibit differential rotation, and the excited defect signal presents a rolling-type timing pulse; for the non-fixed-point defects on the rolling elements of the bearing, the rolling elements and the inner and outer rings will all generate pulse wave packets, and the excited defect signal presents a dense chasing-type timing pulse.

3. The non-contact acoustic signal detection method for bearing faults according to claim 1, characterized in that: Step S6 is specifically as follows: Through FFT transformation or wavelet transformation, the change of energy of each frequency band as the fault cycle increases is obtained, and the frequency band that characterizes the change of the damage degree of early bearing fault in the wavelet packet energy spectrum is extracted. The change of the bearing damage degree is provided from the energy spectrum of each frequency band, and the possible location and degree of the bearing fault are analyzed.

4. Bearing fault non-contact acoustic signal detection device, characterized by include: A motor (41), a coupling (42), a bearing seat 1 (43), a shaft (44) and a bearing seat 2 (45), one end of the coupling (42) is connected to the motor shaft of the motor (41), and the other end is connected to the first end of the shaft (44); the first end of the shaft (44) is supported by the bearing seat 1 (43), the second end of the shaft (44) is inserted into the inner ring of the bearing to be tested (1), and the outer ring of the bearing to be tested (1) is supported by the bearing seat 2 (45); an acoustic superstructure (3) is provided on the bearing to be tested (1), and an acoustic detection mechanism (2) is provided in the direction of the central axis of the bearing to be tested (1); the rotating bearing to be tested, its defect interacts with other parts of the bearing to generate a time-series phase difference wave packet, and the time-series phase difference wave packet is amplified by the acoustic superstructure (3) provided on the bearing to form a vortex pulse acoustic soliton signal; the acoustic detection mechanism (2) detects the acoustic field focusing degree of the vortex pulse acoustic soliton signal, determines the final measurement plane, and collects the transmitted acoustic pulse waveform signal; The acoustic superstructure (3) is a Helmholtz resonant cavity, and a plurality of acoustic superstructures (3) are arranged in an array on the bearing to be tested (1); The bearing to be tested (1) is a single-row tapered roller bearing with a cage; the acoustic superstructure (3) is located on the end face of the cage of the bearing to be tested (1); or, the acoustic superstructure (3) is located on the end face of the bearing outer ring in a direction away from the shaft (44); or, the acoustic superstructure (3) is located on the end face of the bearing inner ring in a direction away from the shaft (44); The acoustic superstructure (3) corresponds one-to-one to the projection of the roller of the bearing to be tested (1) on the bearing end surface.

5. The non-contact acoustic signal detection device for bearing fault according to claim 4, characterized in that: The acoustic detection mechanism (2) comprises an acoustic array module (21) and an acoustic intensity detection module (22), and both the acoustic array module (21) and the acoustic intensity detection module (22) are located in the direction of the central axis of the bearing to be tested (1).

Images