Keyless phase change rotating speed bearing fault diagnosis method and device

Abstract

The invention discloses a keyless phase-change rotating speed bearing fault diagnosis method and device, and belongs to the technical field of bearing fault diagnosis. The method comprises the following steps: acquiring a variable rotating speed bearing fault vibration signal; a pre-constructed TFRPET algorithm is applied to the vibration signals, and high-resolution time-frequency representation TFR is obtained; the TFRPET algorithm compresses the width of a TF ridge through a peak value extraction operator PEO, and the energy aggregation of the TF ridge is improved; detecting a TF ridge from a time-frequency representation (TFR) according to a ridge detection technique and selecting a reference instantaneous frequency (IF) from the TF ridge; and carrying out angular domain resampling and order spectrum diagnosis on the vibration signal by using the reference instantaneous frequency IF to obtain a fault diagnosis result of the keyless phase change rotating speed bearing. The method does not need to depend on a key phase device, can adapt to a variable rotating speed working condition, achieves high-resolution TFR through the TFRPET algorithm, and solves the problem that in the prior art, the TFR resolution is insufficient, and the bearing fault judgment accuracy is affected.

Description

Technical Field

[0001] This invention belongs to the field of bearing fault diagnosis technology, specifically relating to a method and apparatus for diagnosing faults in keyless phase change speed bearings. Background Technology

[0002] Bearings, as crucial transmission components in rotating machinery, directly impact the overall operation of the equipment. Bearing failure can lead to complete unit shutdown and incur significant repair costs. Due to variations in rotational speed and harsh working environments, measured bearing vibration signals exhibit strong non-stationarity and noise characteristics. In actual operation, the working environment of mechanical equipment is often harsh, hindering the installation of key phase hardware, making it difficult to obtain instantaneous frequencies and severely impacting the effectiveness of bearing fault diagnosis. Traditional time-domain and frequency-domain methods are only suitable for analyzing stationary signals.

[0003] Time-frequency analysis (TFA) methods, employing a joint analysis approach in the time and frequency domains, offer profound insights into the transient structure of non-stationary signals, thus attracting widespread attention. Traditional TFA methods, such as the Short-Time Fourier Transform (STFT) and Wavelet Transform (WT), suffer from low time-frequency resolution due to the Heisenberg uncertainty principle limiting their time-frequency representation (TFR) performance. While the Wigner-Vell distribution (WVD), unaffected by the Heisenberg uncertainty principle, offers high resolution, it introduces cross-term interference in multi-component signals, impacting fault feature extraction.

[0004] In recent years, some more advanced Time-Frequency Analysis (TFA) techniques have been used to analyze the time-varying structure of non-stationary signals. Examples include the Redistribution Method (RM) and the Synchronous Compression Transform (SST). The RM enhances the energy concentration of the TFR by redistributing time-frequency coefficients to the centroids of the signal along the time and frequency axes. However, RM does not support signal reconstruction, which significantly limits its application. The Synchronous Compression Transform (SST) is a special case of RM that redistributes time-frequency coefficients only along the frequency axis and allows signal reconstruction. While SST allows signal reconstruction, it is sensitive to noise, and its performance degrades significantly in noisy environments. Higher-order SSTs are computationally complex and rely on higher-order information of the signal, limiting their practical applications.

[0005] Some research has applied the SST method to fault diagnosis of rotating machinery, combining demodulation techniques with the SST method for analyzing mechanical signals, such as SST based on matched demodulation transform and SST based on generalized stepwise demodulation. Demodulation-based SST methods require knowledge of the instantaneous frequency (IF) or phase information, but accurate estimation of IF is difficult in actual operating conditions, especially under noise interference, where the IF estimation error is large. Demodulation techniques typically assume the signal is narrowband or has a specific modulation form, making them insufficiently adaptable to complex multi-component signals.

[0006] Furthermore, existing SST methods may amplify the effects of noise when compressing time-frequency coefficients. Under noise interference, the time-frequency representation is easily distorted, which can lead to weak fault features being submerged, making early fault detection difficult. Under low signal-to-noise ratio conditions, time-frequency energy diffusion is severe, reducing the reliability of fault feature extraction.

[0007] In terms of adaptability to actual working conditions, bearing signals exhibit strong non-stationarity when the speed or load changes, making it difficult for traditional methods to dynamically track instantaneous frequency (IF) changes. Summary of the Invention

[0008] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and device for diagnosing faults in keyless phase change speed bearings. The invention achieves high-resolution time-frequency representation (TFR) through the TFRPET algorithm, thus solving the problem of insufficient TFR resolution in the prior art, which affects the accuracy of bearing fault diagnosis.

[0009] To solve the above-mentioned technical problems, the present invention is implemented using the following technical solution:

[0010] In a first aspect, the present invention provides a fault diagnosis method for a keyless phase change speed bearing, comprising:

[0011] Vibration signals from a variable-speed bearing fault are collected; a pre-constructed TFRPET algorithm is applied to the vibration signals to obtain a high-resolution time-frequency representation (TFR); the TFRPET algorithm compresses the TF ridge width of the vibration signal through the peak extraction operator PEO, thereby improving the energy concentration of the TF ridge; the TF ridge is detected from the time-frequency representation (TFR) using ridge detection technology, and a reference instantaneous frequency (IF) is selected from the TF ridge; using the reference instantaneous frequency (IF), the vibration signal is resampled in the angular domain and subjected to order spectrum diagnosis to obtain the fault diagnosis results of the bondless phase-variable-speed bearing.

[0012] The aforementioned method for diagnosing faults in bondless phase change speed bearings, including the construction of the TFRPET algorithm, comprises:

[0013] The acquired vibration signals are processed using STFT to generate a 3D TFR and its 2D slice properties are analyzed. Based on the 2D slice properties, ridge crest TF points are extracted and set as the extracted TF ridge crest IF trajectory. A peak extraction operator PEO is defined based on the extracted TF ridge crest IF trajectory to extract the ridge crest TF points, and a TFRPET algorithm is generated. The calculation formula for the peak extraction operator PEO is:

[0014] ,

[0015] In the formula, Indicates the position of any TF point in the TFR; This represents the extracted IF trajectory at the top of the TF ridge;

[0016] Based on the peak extraction operator PEO, the TFRPET algorithm expression is obtained as follows:

[0017] ,

[0018] In the formula, This is the STFT expression for the vibration signal.

[0019] The aforementioned method for diagnosing faults in bondless phase change speed bearings, wherein the step of processing the acquired vibration signal using STFT to generate a three-dimensional TFR and analyzing its two-dimensional slice properties includes:

[0020] The collected vibration signals were processed by STFT to obtain a three-dimensional TFR including the TF ridge;

[0021] A two-dimensional time-definite TF ridge slice is obtained by selecting a time segment from the normal direction of the change of the three-dimensional TFR along the TF ridge.

[0022] Analyzing the curve properties of the TF ridge slices, we obtain:

[0023] Property 1: The first derivative at the maximum point of the slice curve is zero; Property 2: The second derivative at the maximum point of the slice curve is a minimum.

[0024] The aforementioned method for diagnosing faults in bondless phase change speed bearings, wherein the extraction of ridge TF points based on slice properties, and the aggregation of these points into the extracted TF ridge IF trajectory, includes:

[0025] Based on the properties of the two-dimensional slice, the TF point at the ridge apex is preliminarily determined;

[0026] The STFT expression of the vibration signal is subjected to Taylor expansion at each of the preliminarily determined TF points at the ridge tops;

[0027] Based on each Taylor expansion, the offset of each TF point from the initially determined TF point of each ridge is calculated.

[0028] TF points whose offsets meet the preset extraction range and whose Hessian matrix eigenvalues ​​meet the preset threshold requirements are retained as extracted ridge TF points;

[0029] The extracted set of TF points at the ridge crest is used as the extracted TF ridge crest IF trajectory.

[0030] The aforementioned method for diagnosing faults in bondless phase change speed bearings involves performing a Taylor expansion of the STFT expression of the vibration signal at each preliminarily determined TF point, including:

[0031] STFT expression for vibration signal Perform a second-order Taylor expansion at each TF point on the ridge crest. At a preliminarily determined TF point at the ridge apex. The second-order Taylor expansion is as follows:

[0032] ,

[0033] In the formula, The STFT expression of the vibration signal is in The value; and These are the STFT expressions for the vibration signal with respect to... and The first-order partial derivative; The STFT expression for the vibration signal is about The first-order partial derivative in The value; The STFT expression for the vibration signal is about The first-order partial derivative in The value; This represents the second-order partial derivative, which is the first-order partial derivative. about The partial derivatives; It is the first-order partial derivative about The partial derivatives in The value; This represents the second-order partial derivative, which is the first-order partial derivative. about The partial derivatives; It is the first-order partial derivative about The partial derivatives in The value; This represents the second-order partial derivative, which is the first-order partial derivative. about The partial derivatives; It is the first-order partial derivative about The partial derivatives in The value of .

[0034] In the aforementioned method for diagnosing faults in bondless phase change speed bearings, the first and second partial derivatives in the second-order Taylor expansion are obtained by convolving the TFR with a Gaussian kernel function; among them, the first partial derivative... and second-order partial derivatives and The calculation formula is as follows:

[0035] ,

[0036] ,

[0037] In the formula, It is a Gaussian kernel function. This represents the partial derivative of the Gaussian kernel function. It is the variance of the Gaussian function.

[0038] The aforementioned fault diagnosis method for keyless phase change speed bearings, wherein the step of calculating the offset from each TF point to each preliminarily determined ridge TF point according to each Taylor expansion includes:

[0039] Based on a preliminarily determined TF point at the ridge top The second-order Taylor expansion is used to solve for the distance from each TF point to the ridge TF point. Offset:

[0040] A preliminary TF point at the ridge apex was determined. The normal direction is defined as Introducing parameters The second-order Taylor expansion can be rewritten as:

[0041] ,

[0042] Based on the fact that the first-order partial derivative at the point where the maximum value of property 1 passes through a zero point, the rewritten Taylor expansion is applied to... Find the first derivative and set it equal to zero:

[0043] ,

[0044] Based on the fact that the first derivative is equal to zero, we can obtain... The parsing expression:

[0045] ,

[0046] From each TF point to a preliminarily determined TF point at the ridge top The time offset is The frequency offset is ;

[0047] A preliminary TF point at the ridge apex was identified. normal direction The calculations include:

[0048] Based on a preliminarily determined TF point at the ridge top The second derivative yields the corresponding Hessian matrix, and the ridge TF point. The expression for the Hessian matrix is:

[0049] ,

[0050] Based on the TF point at the ridge Calculating the TF point at the ridge crest using the Hessian matrix normal direction .

[0051] In the aforementioned fault diagnosis method for bondless phase change speed bearings, the eigenvector corresponding to the eigenvalue with the largest absolute value of the Hessian matrix is ​​the normal direction of the TF point at the ridge.

[0052] Retain the TF points whose Hessian matrix eigenvalues ​​meet the preset threshold requirements, including:

[0053] Each TF point whose corresponding Hessian matrix eigenvalue is less than a preset threshold Th is retained to filter out background noise in the flat area outside the TF ridge; wherein, the preset threshold Th is determined based on the Hessian matrix eigenvalue corresponding to each TF point under each frequency harmonic signal, and each frequency harmonic signal is a component of the bearing fault vibration signal.

[0054] The extracted ridge crest TF point set is the extracted TF ridge crest IF trajectory, including:

[0055] The extracted IF trajectory at the top of the TF ridge is:

[0056] ,

[0057] In the formula, This indicates the extracted ridge crest TF point. A set of.

[0058] The aforementioned fault diagnosis method for keyless phase change speed bearings, wherein the method utilizes the reference instantaneous frequency IF to perform angular domain resampling and order spectrum diagnosis on the vibration signal to obtain the fault diagnosis result of the keyless phase change speed bearing, includes:

[0059] Calculate the instantaneous angle based on the determined reference IF;

[0060] The collected vibration signals of the variable speed bearing fault are resampled at equal angular intervals according to the calculated instantaneous angle;

[0061] Perform an FFT on the resampling results to obtain the order spectrum;

[0062] Identify the peak value in the order spectrum that matches the theoretical fault order, perform fault diagnosis on the variable speed bearing, and output the fault diagnosis result of the variable speed bearing.

[0063] In a second aspect, the present invention provides a fault diagnosis device for a keyless phase change speed bearing, comprising: a signal acquisition module, an algorithm application module, an IF selection module, and a fault diagnosis module;

[0064] The signal acquisition module is used to: acquire vibration signals from variable speed bearing faults;

[0065] The algorithm application module is used to: apply a pre-constructed TFRPET algorithm to the vibration signal to obtain a high-resolution time-frequency representation (TFR); the TFRPET algorithm compresses the TF ridge width of the vibration signal through the peak extraction operator (PEO) to improve the energy concentration of the time-frequency TF ridge.

[0066] The IF selection module is used to: detect the TF ridge from the time-frequency representation TFR of the algorithm application module according to the ridge detection technology and select the reference instantaneous frequency IF from the TF ridge;

[0067] The fault diagnosis module is used to: use the reference instantaneous frequency IF selected by the IF selection module to perform angular domain resampling and order spectrum diagnosis on the vibration signal acquired by the signal acquisition module, and obtain the fault diagnosis result of the keyless phase change speed bearing.

[0068] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0069] The keyless variable speed bearing fault diagnosis method of the present invention does not rely on a key phaser and can adapt to variable speed conditions. By extracting the time-frequency TF ridge width of the compression vibration signal from the peak extraction operator PEO of the TFRPET algorithm, the energy concentration of the time-frequency TF ridge is improved, and a high-resolution time-frequency representation of TFR is achieved, solving the problem of insufficient TFR resolution in the prior art, which affects the accuracy of bearing fault diagnosis.

[0070] The fault diagnosis method for keyless variable speed bearings of the present invention solves the problems of high hardware cost and complex installation caused by reliance on key phasers in the prior art. It overcomes the defects of insufficient resolution and weak noise resistance of traditional time-frequency analysis methods under variable speed conditions. It eliminates the dependence on prior instantaneous frequency information based on demodulation technology, improves the accuracy and reliability of non-stationary signal analysis, and effectively overcomes the influence of speed fluctuations by using angular domain resampling and order spectrum analysis. While reducing hardware costs, it improves the accuracy, stability and engineering applicability of fault diagnosis, and can perform fault diagnosis of variable speed bearings under the condition of keyless speed estimation.

[0071] Specifically, it includes:

[0072] 1) The TFRPET method proposed in this invention retains only the TF points within a certain range at the top of the time-frequency ridge through the Peak Extracting Operator (PEO), which significantly improves the energy concentration of TFR and can reveal the instantaneous characteristics of the signal more clearly compared with traditional SST, STFT and other methods;

[0073] 2) Unlike SST and other methods, which are prone to introducing noise during compression, TFRPET effectively removes noise interference by discarding TF points that are not at the top of the ridge, and can still obtain a clear and stable time-frequency distribution, especially under low signal-to-noise ratio conditions.

[0074] 3) The present invention proposes a joint fault diagnosis scheme that combines TFRPET with order analysis to construct a speed measurement-free method. By selecting a reference IF, corner domain resampling is completed and fault order spectrum analysis is achieved, which simplifies the diagnosis process.

[0075] 4) This invention is applicable to non-stationary vibration signals with complex characteristics such as multiple components and rapidly changing instantaneous frequencies. It can effectively extract the time-varying characteristic frequencies of bearing faults and meet the fault diagnosis needs under variable speed conditions.

[0076] The keyless phase-change speed bearing fault diagnosis method of this invention can directly obtain the instantaneous frequency from the vibration signal, overcoming the dependence on keyed phase hardware equipment, and can efficiently and accurately extract bearing fault features. It overcomes the limitations of traditional time-frequency analysis methods in terms of time-frequency resolution and energy focusing, and can effectively remove background noise information in the TFR, improving the time-frequency representation resolution of non-stationary vibration signals. By combining TFRPET with order analysis, it can accurately distinguish and identify non-steady-state components caused by speed changes, achieving precise extraction of bearing fault features under complex backgrounds, especially suitable for operating conditions under variable speed conditions. The TFRPET proposed in this invention can perform stable time-frequency reconstruction when the non-stationary characteristics of the signal change strongly, solving the problem that traditional SST and STFT methods are difficult to adapt to actual variable speed signals. Compared with SST, the method proposed in this invention provides high-quality TFR while supporting signal reconstruction, facilitating data fusion in subsequent fault diagnosis systems. Attached Figure Description

[0077] Figure 1 This is a schematic flowchart of a keyless phase change speed bearing fault diagnosis method according to Embodiment 1 of the present invention;

[0078] Figure 2 The frequency of Embodiment 1 of the present invention is constant. A schematic diagram of the three-dimensional TFS of the harmonic signal and a two-dimensional slice thereof;

[0079] Figure 3 This is Embodiment 1 of the present invention. Figure 2 A schematic diagram illustrating the properties of a two-dimensional slice;

[0080] Figure 4 This is a schematic diagram comparing the feature extraction effects of the TFRPET algorithm, STFT method, and SST method in Embodiment 1 of the present invention;

[0081] Figure 5This is a schematic diagram illustrating the effect of the TFRPET algorithm outputting high-resolution TFR in Embodiment 1 of the present invention;

[0082] Figure 6 This is based on Embodiment 1 of the present invention. Figure 5 IF extracted from vibration signal TFR;

[0083] Figure 7 This is the selection of Embodiment 1 of the present invention. Figure 6 A schematic diagram of the order spectrum generated using IF1 as the reference IF;

[0084] Figure 8 The frequency of Embodiment 1 of the present invention is constant. A schematic diagram showing the distribution of Hessian matrix eigenvalues ​​at each time-frequency point in the time-frequency representation of a harmonic signal. Detailed Implementation

[0085] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments and specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations thereof. In the absence of conflict, the embodiments and technical features in the embodiments can be combined with each other.

[0086] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0087] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. These drawings are simplified schematic diagrams, which are only used to illustrate the basic structure of the present invention and therefore only show the components relevant to the present invention.

[0088] Example 1:

[0089] Figure 1 This is a flowchart illustrating a fault diagnosis method for a bondless phase change speed bearing according to Embodiment 1 of the present invention. This flowchart only shows the logical sequence of the method described in this embodiment. Provided there are no conflicts, different methods may be used in other possible embodiments of the present invention. Figure 1 Complete the steps shown or described in the order indicated.

[0090] like Figure 1 As shown, the fault diagnosis method for keyless phase change speed bearings in this embodiment includes:

[0091] S1: Collect vibration signals from variable speed bearing faults;

[0092] S2: Apply the pre-constructed TFRPET algorithm to the vibration signal from step S1 to obtain a high-resolution time-frequency representation (TFR); the TFRPET algorithm compresses the time-frequency TF ridge width of the vibration signal through the peak extraction operator (PEO) to improve the energy concentration of the ridge line.

[0093] S3: Detect the TF ridge from the time-frequency representation TFR of step S2 according to the ridge detection technology and select a reference instantaneous frequency IF from the TF ridge;

[0094] S4: Using the reference instantaneous frequency IF from step S3, perform angular domain resampling and order spectrum diagnosis on the vibration signal from step S1 to obtain the fault diagnosis results of the keyless phase change speed bearing.

[0095] The specific implementation of each step is described below:

[0096] Step S1 includes:

[0097] Accelerometers are used to collect vibration signals of bearings under variable speed conditions. Local bearing faults can cause periodic impact vibrations, the frequency of which is related to the rotational speed. Therefore, collecting vibration signals of variable speed bearing faults can be used for subsequent fault diagnosis.

[0098] In step S2, the construction of the TFRPET algorithm includes:

[0099] S21: Perform a short-time Fourier transform (STFT) on the signal acquired in step S1 to generate a three-dimensional TFR and analyze the slice properties;

[0100] The vibration signal acquired in step S1 is subjected to a short-time Fourier transform (STFT) to obtain a signal as shown below. Figure 2 (a) shows the three-dimensional time-frequency representation (TFR) including the TF ridge, as shown in Figure 1. Figure 2 (a) The TF point at the top of the TF ridge, indicated by the red dashed line, has the maximum energy. Then, a slice is taken at a selected time along the normal direction of the TF ridge's time variation to obtain a two-dimensional, time-dependent TF ridge slice. Figure 2 (b) The curve of the TF ridge slice is defined as the edge line, resulting in the following: Figure 3 The two properties shown:

[0101] Property 1: The first derivative of a slice curve is zero at the maximum point, which means it passes through a zero point;

[0102] Property 2: The second derivative at the maximum point of the slice curve is a minimum, that is, a concave point.

[0103] Using these two properties, the TF point located at the top of the TF ridge can be searched along the normal direction of the TF ridge variation. This allows for the acquisition of a more concentrated TFR containing only the TF point at the top of the TF ridge.

[0104] In this embodiment, step S21 is implemented as follows:

[0105] The vibration signal of the bearing fault collected in step S1 is expressed as follows:

[0106] ,

[0107] In the formula, and These represent the instantaneous amplitude and instantaneous phase of the bearing fault vibration signal, respectively; i represents an imaginary number;

[0108] Equation (1) is expressed using STFT as follows:

[0109] (2),

[0110] In the formula: Represents the window function. This represents the signal to be analyzed.

[0111] To calculate the subsequent instantaneous frequency, phase compensation is performed, and a phase operator is introduced here. Formula (2) can be rewritten as:

[0112] (3),

[0113] According to Passevar's theorem, formula (3) can be further expressed as:

[0114] (4),

[0115] In the formula, and These represent the Fourier transforms of the original window function and the bearing fault vibration signal, respectively.

[0116] like Figure 2 As shown in (a), the present invention uses an amplitude The frequency remains constant. The harmonic signal yields a three-dimensional TFR including the TF ridge;

[0117] According to formula (4), the STFT expression of the vibration signal is:

[0118] (5),

[0119] Equation (5) shows that the energy distribution of TFR is concentrated in Nearby, and meets the requirements This means that in The maximum amplitude is reached at this point. To better describe the distribution structure of TFR, TFR is extended to, for example... Figure 2 (a) shows the three-dimensional time-frequency space (TFS). The three-dimensional TFS of the harmonic signal exhibits a ridge-like structure. Subsequently, as shown in... Figure 2 (a) shows the TF ridge direction It is frequency A timeline showing changes over time; select a time frame to capture a slice, such as... Figure 2 As shown in (b), the slice is in It has the largest energy amplitude. .

[0120] like Figure 3 As shown, from Figure 2 The structure of the TF ridge slice shown in (b) yields two properties: the first derivative at the point of maximum value is a zero-crossing point, and the second derivative at the point of maximum value is a minimum point. The principle is that the energy of the harmonic signal... The slices converge at specific points, forming a time-frequency ridge. At the ridge apex, the slices satisfy the extremum condition, meaning the gradient is zero and the curvature is maximum.

[0121] Step S22: Based on the properties of the two-dimensional slice, extract the set of TF points at the ridge crest as the extracted TF ridge crest IF trajectory:

[0122] Based on the properties of the two-dimensional slice, the TF point at the ridge apex is preliminarily determined;

[0123] The STFT expression of the vibration signal is subjected to Taylor expansion at each of the preliminarily determined TF points at the ridge tops;

[0124] Based on each Taylor expansion, the offset of each TF point from the initially determined TF point of each ridge is calculated.

[0125] TF points whose offsets meet the preset extraction range and whose Hessian matrix eigenvalues ​​meet the preset threshold requirements are retained as extracted ridge TF points;

[0126] The extracted set of TF points at the ridge crest is used as the extracted TF ridge crest IF trajectory.

[0127] In this embodiment, step S22 is implemented as follows:

[0128] First, based on the slice property that the first derivative is zero at the maximum point and the second derivative is a minimum at the maximum point, the TF point at the ridge apex is initially determined;

[0129] Then, the STFT expression for the vibration signal. Perform a second-order Taylor expansion at each TF point on the ridge crest. At a preliminarily determined TF point at the ridge apex. The second-order Taylor expansion is as follows:

[0130] (6),

[0131] In the formula, It is the STFT of the vibration signal in The value; and These are the STFTs of the vibration signal with respect to time. and frequency The first-order partial derivative; It is the STFT of the vibration signal about The first-order partial derivative in The value; It is the STFT of the vibration signal about The first-order partial derivative in The value; This represents the second-order partial derivative, which is the first-order partial derivative. about The partial derivatives; It is the first-order partial derivative about The partial derivatives in The value; This represents the second-order partial derivative, which is the first-order partial derivative. about The partial derivatives; It is the first-order partial derivative about The partial derivatives in The value; This represents the second-order partial derivative, which is the first-order partial derivative. about The partial derivatives; It is the first-order partial derivative about The partial derivatives in The value;

[0132] The first-order partial derivative in the Taylor expansion of formula (6) is obtained by convolving TFR with the Gaussian kernel function. and ) and second-order partial derivatives ( , and );

[0133] Wherein, the first-order partial derivative and second-order partial derivatives and The calculation formula is as follows:

[0134] (7),

[0135] (8),

[0136] In the formula, It is a Gaussian kernel function. This represents the partial derivative of the Gaussian kernel function. It is the variance of the Gaussian function;

[0137] According to formula (7), the following can be obtained: and In the formula, and Formula (8) can be used to... , In Replace with get.

[0138] Based on a preliminarily determined TF point at the ridge top The second-order Taylor expansion is used to solve for the distance from each TF point to the ridge TF point. Offset:

[0139] The TF point at the top of the TF ridge is located where the first derivative of the TF ridge passes through zero and the second derivative reaches its minimum value along the normal direction of the TF ridge; along the TF ridge, the TF point at the top of the TF ridge... Based on the normal direction, a preliminary vertices point (TF) is determined at the ridge crest. The second-order Taylor expansion, namely the expansion formula (6), will initially determine a ridge apex TF point. The normal direction is defined as Introducing parameters Formula (6) can be rewritten as:

[0140] (9),

[0141] Then, using the first-order partial derivative of the point where the maximum value of property 1 is located passing through the zero point, equation (9) is applied to... Find the first derivative and set it equal to zero, which means setting the function... For parameters The first derivative is set to zero:

[0142] (10)

[0143] Then, by transforming formula (10), we obtain... The parsing expression:

[0144] (11),

[0145] From each TF point to a preliminarily determined TF point at the ridge top The time offset is The frequency offset is ;

[0146] A preliminary TF point at the ridge apex was identified. normal direction The calculations include:

[0147] Based on a preliminarily determined TF point at the ridge top The second derivative yields the corresponding Hessian matrix, and the ridge TF point. The expression for the Hessian matrix is:

[0148] (12),

[0149] Based on the TF point at the ridge Calculating the TF point at the ridge crest using the Hessian matrix normal direction In this embodiment, the eigenvector corresponding to the eigenvalue with the largest absolute value of the Hessian matrix is ​​the normal direction of the TF point at the ridge crest.

[0150] TF points whose offsets meet the preset extraction range and whose Hessian matrix eigenvalues ​​meet the preset threshold requirements are retained as extracted ridge apex TF points, as follows:

[0151] The preset extraction range for the offset is: By retaining only the time-frequency points centered on the initially determined time-frequency ridge and within a certain frequency offset range, the diffusion width of the time-frequency ridge can be compressed, thereby improving the energy concentration of the time-frequency ridge. In other words, by extracting the time-frequency points in the neighborhood of the ridge line and discarding the energy diffused in the periphery, the purpose of sharpening the ridge line and improving the energy concentration can be achieved.

[0152] Retain the TF points whose Hessian matrix eigenvalues ​​meet the preset threshold requirements, including:

[0153] Retaining the TF points whose corresponding Hessian matrix eigenvalues ​​are less than a preset threshold Th can filter out background noise in the flat region outside the time-frequency ridge. The preset threshold Th is determined based on the Hessian matrix eigenvalues ​​corresponding to each TF point under each frequency harmonic signal. Each frequency harmonic signal is a component of the bearing fault vibration signal, such as... Figure 8 The frequency shown is constant. The diagram shows the distribution of Hessian matrix eigenvalues ​​corresponding to each time-frequency point in the time-frequency representation of the harmonic signal. The horizontal axis represents the order of the time-frequency points, and the vertical axis represents the magnitude of the eigenvalues. The area below -0.45 (red dashed line) in the diagram represents the time-frequency ridge TF point and its vicinity. In a specific embodiment, a preset threshold Th is set to -0.45, which can filter out background noise in the flat area outside the time-frequency ridge. Based on the distribution diagram of Hessian matrix eigenvalues ​​corresponding to each time-frequency point in the time-frequency representation of the harmonic signal, the final preset threshold Th can be obtained.

[0154] TF points whose offsets meet the preset extraction range and whose Hessian matrix eigenvalues ​​meet the preset threshold requirements are retained. As the extracted ridge apex TF point;

[0155] The extracted set of TF points at the ridge crest is used as the extracted TF ridge crest IF trajectory;

[0156] According to formula (9), the extracted TF ridge apex IF trajectory can be defined as In the formula, This indicates the extracted ridge crest TF point. A set;

[0157] Step S23: Define the peak extraction operator PEO for extracting the TF points at the ridge crest based on the extracted TF ridge crest IF trajectory, and generate the TFRPET algorithm:

[0158] Based on the extracted TF ridge crest IF trajectory Peak Extraction Operator (PEO) is used to extract peak values. The formula for calculating PEO is as follows:

[0159] (14)

[0160] In the formula, This indicates the position of any TF point in the TFR.

[0161] Based on the peak extraction operator PEO obtained from formula (14), the expression for the time-frequency ridge peak extraction transform (TFRPET) is obtained as follows:

[0162] (15)

[0163] PEO is used to directly extract the time-frequency point of the time-frequency ridge, map the original TFR to the sharpened new TFR, and eliminate energy diffusion.

[0164] Step S3 includes:

[0165] The TFRPET signal is applied to the fault vibration signal acquired in step S1 to output a high-resolution time spectrum.

[0166] Because TFRPET compresses the time-frequency ridge width through PEO, it improves energy concentration compared to traditional STFT.

[0167] Figure 4 This diagram illustrates a comparison of the feature extraction performance of the STFT method, SST method, and TFRPET algorithm in this embodiment. Figure 4 (a) and Figure 4 (b) shows the feature extraction performance of the STFT method; Figure 4 (c) and Figure 4 (d) indicates the feature extraction performance of the SST method; Figure 4 (e) and Figure 4 (f) represents the feature extraction performance of the TFRPET algorithm. From Figure 4 (a) to Figure 4 As can be seen from the comparison in (f), the dark blue region outside the TF ridge in the time-frequency features represented by STFT and SST is doped with a lot of background noise, and the feature energy is severely diverged. However, the TFRPET algorithm proposed in this invention is better than STFT and SST in reducing background noise and enhancing energy concentration, and has higher resolution.

[0168] Figure 5 This is a schematic diagram illustrating the effect of the TFRPET algorithm outputting high-resolution TFR in Embodiment 1 of the present invention. Figure 5 (a) is the vibration signal of the variable speed bearing fault measured in step S1. Figure 5 (b) is the method of this embodiment for Figure 5 (a) TFR results of the signal. From Figure 5 As can be seen from (b), the fault characteristic energy of features 1, 2, 3 and 4 is relatively concentrated, and the background noise is well removed.

[0169] Step S4 includes:

[0170] A reliable IF is estimated and selected as a reference IF using ridge line detection technology;

[0171] This embodiment uses a bearing outer ring failure as an example. A failure in the outer ring of a rolling bearing causes transient impacts when the rolling elements pass over the defect, resulting in periodic vibration signals. These vibration signals appear as periodic impact ridges on a time-frequency plane (such as the STFT), and the ridge frequency is the instantaneous frequency. The instantaneous frequency of the rolling bearing outer ring failure... The formula can be expressed as:

[0172] (16)

[0173] In the formula, Indicates the number of rolling elements; Indicates the diameter of the rolling element; Indicates the bearing pitch circle diameter (the diameter of the circle in which the rolling elements are distributed). This represents the contact angle, which is the contact angle between the rolling element and the inner and outer rings; This indicates a proportional correction term for the rolling element dimensions and bearing geometry. This indicates that half of the rolling elements will come into contact with the outer ring defect on each revolution. This indicates the rotational frequency of the shaft.

[0174] In this embodiment, a path tracing algorithm based on amplitude thresholds is used to detect ridge lines from the TFRPET results, such as... Figure 6 based on Figure 5 As shown in the IF extracted from the vibration signal TFR, the TFR obtained through the TFRPET method shows that all IF trajectories (IF1 to IF4) are relatively accurate. Therefore, as... Figure 6 In this context, selecting any reliable IF (IF1 to IF4) as the reference IF can effectively show the order relationship between different IFs.

[0175] Bearing failure impacts appear as clear ridges on TFRPET. In this embodiment, the instantaneous frequency of the ridge line for a rolling bearing outer ring failure is... The high resolution of TFRPET ensures the accuracy of all IF trajectories, and any IF trajectory in TFR obtained by the TFRPET algorithm reliably reflects the rotation frequency change.

[0176] Step S5: Angular domain resampling and order spectrum diagnosis

[0177] The fault vibration signal is resampled in the angular domain and the order spectrum is plotted. Fault diagnosis of the variable speed bearing is performed by identifying the fault-related order.

[0178] The TFR obtained from TFRPET is directly used to estimate the IF and obtain the order spectrum, with the fundamental frequency IF selected as the reference IF. Using this reference IF, the original vibration data is resampled in the angular domain. The order spectrum results can effectively identify the fault order and its multiple harmonic orders. Fault diagnosis of the transmission bearing is completed by identifying the fault characteristic orders.

[0179] Specifically, it includes:

[0180] Calculate the instantaneous angle based on the reference IF determined in step S4;

[0181] The vibration signal of the variable speed bearing fault acquired in step S1 is resampled at equal angular intervals according to the calculated instantaneous angle; angular domain resampling converts the non-stationary signal of the bearing fault vibration signal into an angular periodic stationary signal.

[0182] Perform an FFT on the resampling results to obtain the order spectrum; the order spectrum transforms the fault characteristics into a constant order independent of the rotational speed.

[0183] Identify the peak values ​​in the order spectrum that match the theoretical fault order to perform variable speed bearing fault diagnosis; the fault characteristic order is clearly marked in the order spectrum, and the bearing fault type can be determined by combining the bearing parameters. Fault types include damage to the outer ring, inner ring, and rolling elements.

[0184] like Figure 7 The selection shown is as follows Figure 6 Using IF1 as a reference IF, the order analysis spectrum generated in step S5 is used. The order spectrum diagram shows the first, second, and third order rotational frequencies generated by rotor dynamics faults, as well as the bearing fault characteristic order of 3.628 (approximately 3.6) caused by local component damage faults. In this embodiment, the theoretical fault characteristic order of the rolling bearing outer ring fault is 3.572 (approximately 3.6), where the theoretical fault characteristic order M of the rolling bearing outer ring fault is calculated according to the formula: Based on the near-consistency between the measured order spectrum and the theoretical fault characteristic order, the fault diagnosis for the keyless phase change speed bearing can be obtained as follows: Figure 7 The bearing has an outer ring fault.

[0185] Example 2

[0186] Based on the same inventive concept as Embodiment 1, this embodiment introduces a fault diagnosis device for a keyless phase change speed bearing, characterized in that it includes: a signal acquisition module, an algorithm application module, an IF selection module, and a fault diagnosis module;

[0187] The signal acquisition module is used to: acquire vibration signals from variable speed bearing faults;

[0188] The algorithm application module is used to: apply a pre-constructed TFRPET algorithm to the vibration signal to obtain a high-resolution time-frequency representation (TFR); the TFRPET algorithm compresses the TF ridge width of the vibration signal through the peak extraction operator (PEO) to improve the energy concentration of the time-frequency TF ridge.

[0189] The IF selection module is used to: detect the TF ridge from the time-frequency representation TFR of the algorithm application module according to the ridge detection technology and select the reference instantaneous frequency IF from the TF ridge;

[0190] The fault diagnosis module is used to: use the reference instantaneous frequency IF selected by the IF selection module to perform angular domain resampling and order spectrum diagnosis on the vibration signal acquired by the signal acquisition module, and obtain the fault diagnosis result of the keyless phase change speed bearing.

[0191] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0192] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0193] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0194] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0195] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A fault diagnosis method for a keyless phase change speed bearing, characterized in that, include: Collect vibration signals from variable speed bearing faults; A pre-constructed TFRPET algorithm is applied to the vibration signal to obtain a high-resolution time-frequency representation (TFR). The TFRPET algorithm compresses the TF ridge width of the vibration signal through the peak extraction operator (PEO) to improve the energy concentration of the TF ridge. The TF ridge is detected from the time-frequency representation (TFR) using ridge detection technology, and a reference instantaneous frequency (IF) is selected from the TF ridge. The vibration signal is resampled in the angular domain and subjected to order spectrum diagnosis using the reference instantaneous frequency IF to obtain the fault diagnosis results of the bondless phase change speed bearing.

2. The fault diagnosis method for keyless phase change speed bearings according to claim 1, characterized in that, The construction of the TFRPET algorithm includes: The collected vibration signals were processed by STFT to generate a three-dimensional TFR and its two-dimensional slice properties were analyzed. Based on the properties of the two-dimensional slice, the TF points at the ridge crest are extracted and set as the extracted TF ridge crest IF trajectory; Based on the extracted TF ridge crest IF trajectory, a peak extraction operator (PEO) is defined to extract the TF points at the ridge crest, and the TFRPET algorithm is generated; the calculation formula for the peak extraction operator (PEO) is as follows: ， In the formula, Indicates the position of any TF point in the TFR; This represents the extracted IF trajectory at the top of the TF ridge; Based on the peak extraction operator PEO, the TFRPET algorithm expression is obtained as follows: ， In the formula, This is the STFT expression for the vibration signal.

3. The fault diagnosis method for keyless phase change speed bearings according to claim 2, characterized in that, The process of processing the acquired vibration signals using STFT to generate a three-dimensional TFR and analyzing its two-dimensional slice properties includes: The collected vibration signals were processed by STFT to obtain a three-dimensional TFR including the TF ridge; A two-dimensional time-definite TF ridge slice is obtained by selecting a time segment from the normal direction of the change of the three-dimensional TFR along the TF ridge. Analyzing the curve properties of the TF ridge slices, we obtain: Property 1: The first derivative of a slice curve is zero at the maximum point of the slice curve; Property 2: The second derivative at the maximum point of the slice curve is a minimum.

4. The fault diagnosis method for keyless phase change speed bearings according to claim 3, characterized in that, The step of extracting ridge crest TF points based on the two-dimensional slice properties, and setting them as the extracted TF ridge crest IF trajectory, includes: Based on the properties of the two-dimensional slice, the TF point at the ridge apex is preliminarily determined; The STFT expression of the vibration signal is subjected to Taylor expansion at each of the preliminarily determined TF points at the ridge tops; Based on each Taylor expansion, the offset of each TF point from the initially determined TF point of each ridge is calculated. TF points whose offsets meet the preset extraction range and whose Hessian matrix eigenvalues ​​meet the preset threshold requirements are retained as extracted ridge TF points; The extracted set of TF points at the ridge crest is used as the extracted TF ridge crest IF trajectory.

5. The fault diagnosis method for keyless phase change speed bearings according to claim 4, characterized in that, The STFT expression for the vibration signal is subjected to Taylor expansion at each preliminarily determined TF point on the ridge, including: STFT expression for vibration signal Perform a second-order Taylor expansion at each TF point on the ridge crest. At a preliminarily determined TF point at the ridge apex The second-order Taylor expansion is as follows: ， In the formula, The STFT expression of the vibration signal is in The value; and These are the STFT expressions of the vibration signal with respect to time. and frequency The first-order partial derivative; The STFT expression for the vibration signal is about The first-order partial derivative in The value; The STFT expression for the vibration signal is about The first-order partial derivative in The value; This represents the second-order partial derivative, which is the first-order partial derivative. about The partial derivatives; It is the first-order partial derivative about The partial derivatives in The value; This represents the second-order partial derivative, which is the first-order partial derivative. about The partial derivatives; It is the first-order partial derivative about The partial derivatives in The value; This represents the second-order partial derivative, which is the first-order partial derivative. about The partial derivatives; It is the first-order partial derivative about The partial derivatives in The value of .

6. The fault diagnosis method for keyless phase change speed bearings according to claim 5, characterized in that, The first and second partial derivatives in the second-order Taylor expansion are obtained by convolving the TFR with the Gaussian kernel function; Wherein, the first-order partial derivative and second-order partial derivatives and The calculation formula is as follows: ， ， In the formula, It is a Gaussian kernel function. This represents the partial derivative of the Gaussian kernel function. It is the variance of the Gaussian function.

7. The fault diagnosis method for keyless phase change speed bearings according to claim 5, characterized in that, The step of calculating the offset from each TF point to each preliminarily determined ridge apex TF point based on each Taylor expansion includes: Based on a preliminarily determined TF point at the ridge top The second-order Taylor expansion is used to solve for the distance from each TF point to the ridge TF point. Offset: A preliminary TF point at the ridge apex was determined. The normal direction is defined as Introducing parameters The second-order Taylor expansion can be rewritten as: ， Based on the fact that the first-order partial derivative at the point where the maximum value of property 1 passes through a zero point, the rewritten Taylor expansion is applied to... Find the first derivative and set it equal to zero: ， Based on the fact that the first derivative is equal to zero, we can obtain... The parsing expression: ， From each TF point to a preliminarily determined TF point at the ridge top The time offset is The frequency offset is ; A preliminary TF point at the ridge apex was identified. normal direction The calculations include: Based on a preliminarily determined TF point at the ridge top The second derivative yields the corresponding Hessian matrix, and the ridge TF point. The expression for the Hessian matrix is: ， Based on the TF point at the ridge Calculating the TF point at the ridge crest using the Hessian matrix normal direction .

8. The method for diagnosing faults in a keyless phase change speed bearing according to claim 7, characterized in that, The eigenvector corresponding to the eigenvalue with the largest absolute value of the Hessian matrix is ​​the normal direction of the TF point at the ridge crest. Retain the TF points whose Hessian matrix eigenvalues ​​meet the preset threshold requirements, including: Each TF point whose corresponding Hessian matrix eigenvalue is less than a preset threshold Th is retained to filter out background noise in the flat area outside the TF ridge; wherein, the preset threshold Th is determined based on the Hessian matrix eigenvalue corresponding to each TF point under each frequency harmonic signal, and each frequency harmonic signal is a component of the bearing fault vibration signal. The extracted ridge crest TF point set is the extracted TF ridge crest IF trajectory, including: The extracted IF trajectory at the top of the TF ridge is: ， In the formula, This indicates the extracted ridge crest TF point. A set of.

9. The fault diagnosis method for keyless phase change speed bearings according to claim 1, characterized in that, The method of using the reference instantaneous frequency IF to perform angular domain resampling and order spectrum diagnosis on the vibration signal to obtain the fault diagnosis results of the keyless phase change speed bearing includes: Calculate the instantaneous angle based on the determined reference IF; The collected vibration signals of the variable speed bearing fault are resampled at equal angular intervals according to the calculated instantaneous angle; Perform an FFT on the resampling results to obtain the order spectrum; Identify the peak value in the order spectrum that matches the theoretical fault order, perform fault diagnosis on the variable speed bearing, and output the fault diagnosis result of the variable speed bearing.

10. A fault diagnosis device for a keyless phase change speed bearing, characterized in that, include: Signal acquisition module, algorithm application module, IF selection module, and fault diagnosis module; The signal acquisition module is used to: acquire vibration signals from variable speed bearing faults; The algorithm application module is used to: apply a pre-constructed TFRPET algorithm to the vibration signal to obtain a high-resolution time-frequency representation (TFR); the TFRPET algorithm compresses the TF ridge width of the vibration signal through the peak extraction operator (PEO) to improve the energy concentration of the time-frequency TF ridge. The IF selection module is used to: detect the TF ridge from the time-frequency representation TFR of the algorithm application module according to the ridge detection technology and select the reference instantaneous frequency IF from the TF ridge; The fault diagnosis module is used to: use the reference instantaneous frequency IF selected by the IF selection module to perform angular domain resampling and order spectrum diagnosis on the vibration signal acquired by the signal acquisition module, and obtain the fault diagnosis result of the keyless phase change speed bearing.

Images