Method and device for assessing the degradation of a long-life electromagnetic bearing protection bearing

Abstract

The application provides a degradation evaluation method and device suitable for long-life electromagnetic bearing protection bearings, the method comprising: collecting n times of rotor displacement data in the falling process of the electromagnetic bearing protection bearing, n>=2 and is an integer; extracting the rotor speed information changing with time from the rotor displacement data; sequentially performing time domain analysis and wavelet packet decomposition on the rotor speed information to obtain time domain features and wavelet packet decomposition entropy, and determining a high-dimensional feature set according to the time domain features and the wavelet packet decomposition entropy; performing dimension reduction on the high-dimensional feature set by using a local preserving projection algorithm to obtain a low-dimensional feature set; extracting the first low-dimensional feature of the low-dimensional feature set, calculating the JS divergence of the i-th rotor displacement data and the (i-1)-th rotor displacement data in the n times of rotor displacement data for any i-th rotor displacement data, i=2, 3,..., n; calculating the degradation index under different falling times according to the JS divergence, and determining the failure threshold of the protection bearing according to the degradation index.

Description

Technical Field

[0001] This invention relates to the field of bearing technology, and in particular to a degradation assessment method and apparatus suitable for protecting long-life electromagnetic bearings. Background Technology

[0002] Electromagnetic bearing systems require no lubrication and are frictionless, and are currently used in numerous fields. Electromagnetic bearings contain many precision components; if the rotor were to fall directly, it would cause catastrophic consequences for the equipment. Therefore, protective bearings (also called auxiliary bearings, spare bearings, etc.) are often installed internally, with rolling bearings being the most common type. The protective bearing has the smallest clearance from the rotor. In the event of an overload or drop accident, the protective bearing makes initial contact with the high-speed rotating rotor, protecting other components of the unit at the cost of bearing damage. Deciding whether to replace the protective bearing after a drop accident is a difficult decision. Due to the special nature of protective bearings, it is difficult to determine whether a removed protective bearing can continue to function. Furthermore, electromagnetic bearing units are often complex and do not allow for frequent disassembly. Therefore, there is an urgent need for a non-disassembly method for assessing the degradation of protective bearings to answer these questions.

[0003] The lifespan of a protective bearing is typically characterized by "drop counts," which is the number of times a protective bearing can safely stop the rotor at its rated speed without damaging the unit under certain braking conditions after the electromagnetic bearing fails. Under favorable operating conditions, such as rapid braking, a protective bearing may withstand ten or even dozens of drops at its rated speed without failure. However, due to potential manufacturing deviations in protective bearings, even after calculations and prototype verification, concerns remain about its ability to perform the next protective task after a dozen or so drops. Therefore, a method is needed to assess the degradation status of a protective bearing after multiple drops to determine whether replacement is necessary. Summary of the Invention

[0004] In view of the above problems, the present invention provides a degradation assessment method and apparatus suitable for long-life electromagnetic bearings to protect bearings.

[0005] This invention provides a degradation assessment method for electromagnetic bearing protection bearings with long service life, comprising: Step S1, collecting n rotor displacement data during the drop process of the electromagnetic bearing protection bearing, where n ≥ 2 and is an integer; Step S2, extracting time-varying rotational speed information from the rotor displacement data; Step S3, sequentially performing time-domain analysis and wavelet packet decomposition on the rotational speed information to obtain time-domain features and wavelet packet decomposition entropy, and determining a high-dimensional feature set based on the time-domain features and wavelet packet decomposition entropy; Step S4, using a local preserving projection algorithm to reduce the dimensionality of the high-dimensional feature set to obtain a low-dimensional feature set; Step S5, extracting the first low-dimensional feature of the low-dimensional feature set, and for any i-th rotor displacement data in the n rotor displacement data, calculating the JS divergence between the i-th rotor displacement data and the (i-1)-th rotor displacement data, where i = 2, 3, ..., n; Step S6, calculating the degradation index under different drop counts based on the JS divergence, and determining the failure threshold of the protection bearing based on the degradation index.

[0006] According to an embodiment of the present invention, each rotor displacement data includes low-speed segment data of high-speed rotor drop data or low-speed drop data of rotor tested before each drop accident; step S2 specifically includes: performing short-time Fourier transform on the rotor displacement data of a single drop to obtain a time-frequency diagram; and determining the rotational speed information that changes with time on the time-frequency diagram based on the magnitude of the rotor speed at the time of drop.

[0007] According to an embodiment of the present invention, in step S3, the time-domain features include dimensional features and dimensionless features. The dimensional features include mean, variance, root mean square, root square amplitude, peak value or peak-to-peak value. The dimensionless features include kurtosis, skewness, impulse factor, margin factor, peak factor or waveform factor.

[0008] According to an embodiment of the present invention, in step S3, the rotation speed information is sequentially subjected to time-domain analysis and wavelet packet decomposition to obtain time-domain features and wavelet packet decomposition entropy, including: using the db5 mother wavelet to perform 7-level wavelet packet decomposition on the time-domain features to extract low-frequency band information below 625Hz; analyzing the low-frequency band information to obtain wavelet packet decomposition entropy, which includes wavelet packet decomposition energy entropy, wavelet packet decomposition singular value entropy, and fuzzy entropy corresponding to each frequency band.

[0009] According to an embodiment of the present invention, in step S3, determining a high-dimensional feature set based on temporal features and wavelet packet decomposition entropy includes: collecting four sensor features respectively, wherein each sensor feature includes temporal features and wavelet packet decomposition entropy; and merging the four sensor features into a high-dimensional feature set.

[0010] According to an embodiment of the present invention, step S4 specifically includes:

[0011] Suppose a high-dimensional feature set X = [x1, x2, ..., xn] n], x i Let i represent the i-th high-dimensional feature, i = 1, 2, ..., n. The transformation matrix A maps the high-dimensional feature set to a low-dimensional feature set Y = [y1, y2, ..., yn]. n The matrix transformation formula is expressed as:

[0012] y i =A T x i (1)

[0013] The transformation matrix A is obtained by minimizing the objective function, and can be expressed by the formula:

[0014] A=argmin∑ ij (y i -y j ) 2 W ij (2)

[0015] In the formula, y i Let W represent the i-th low-dimensional feature, j = 1, 2, ..., n; ij represents the weight between the i-th and j-th low-dimensional features; argmin represents the variable value that minimizes the objective function;

[0016] Substitute formula (1) into formula (2) and add the constraint condition y. T When Dy = 1, we can obtain...

[0017] A = argminA T XLX T A (3)

[0018] Then, we transform formula (3) into a problem of finding generalized eigenvalues, that is...

[0019] XLX T A=λXDX T A (4)

[0020] Therefore, the first l eigenvectors of formula (4) can be solved, i.e., A = [a0, a1, ..., a1]. l-1 ];

[0021] After obtaining the transformation matrix A, we can use formula (1) to solve for the low-dimensional feature set.

[0022] According to an embodiment of the present invention, step S5 specifically includes: denoteing the first low-dimensional feature of the low-dimensional feature set as the target feature, converting the target feature into a histogram and normalizing it, and then converting it into a probability distribution form; based on the target feature in the probability distribution form, calculating the JS divergence JS(i, i-1) between the probability distribution of the first low-dimensional feature of the i-th rotor displacement data and the probability distribution of the first low-dimensional feature of the (i-1)-th rotor displacement data.

[0023] According to an embodiment of the present invention, in step S6, the degradation index under different number of drops is calculated based on the JS divergence, including performing the following operations for any i-th rotor displacement data in n rotor displacement data:

[0024] Calculate the average value F(i) of the first i rotor displacement data and the average value F(i-1) of the first i-1 rotor displacement data for n rotor displacement data respectively;

[0025] Calculate the relative change of the average value F(i) with respect to the average value F(i-1), ΔF = F(i) / F(i-1);

[0026] The JS divergence JS(i, i-1) is corrected using the relative change ΔF to obtain the corrected JS divergence HC. JS (i) = JS(i, i-1)·ΔF;

[0027] The corrected JS divergence HC corresponding to the i-th rotor displacement data JS (i) Divide by the corrected JS divergence HC corresponding to the (i-1)th rotor displacement data. JS (i-1), to obtain the degradation index H(i) under the i-th fall.

[0028] Another aspect of the present invention provides a degradation assessment device suitable for long-life electromagnetic bearing protection bearings, comprising: a drop data acquisition module for acquiring n rotor displacement data during the drop process of the electromagnetic bearing protection bearing, where n ≥ 2 and is an integer; a speed information extraction module for extracting time-varying speed information from the rotor displacement data; a high-dimensional feature determination module for sequentially performing time-domain analysis and wavelet packet decomposition on the speed information to obtain time-domain features and wavelet packet decomposition entropy, and determining a high-dimensional feature set based on the time-domain features and wavelet packet decomposition entropy; a high-dimensional feature dimensionality reduction module for using a local preservation projection algorithm to reduce the dimensionality of the high-dimensional feature set to obtain a low-dimensional feature set; a JS divergence calculation module for extracting the first low-dimensional feature of the low-dimensional feature set, and calculating the JS divergence between the i-th rotor displacement data and the (i-1)-th rotor displacement data for any i-th rotor displacement data in the n-th rotor displacement data, where i = 2, 3, ..., n; and a failure threshold determination module for calculating degradation indices under different drop counts based on the JS divergence, and determining the failure threshold of the protection bearing based on the degradation indices.

[0029] Compared with existing technologies, the degradation assessment method and apparatus for long-life electromagnetic bearings provided by this invention collects multiple rotor displacement data during the drop test of the electromagnetic bearing. Based on the JS divergence of the probability distribution of the first low-dimensional feature of two adjacent rotor displacement data sets, the degradation index under different drop counts is determined, thereby determining the failure threshold. This allows for analysis of the bearing's degradation state from multiple drop displacement data, judgment of whether the bearing is about to fail, and suggestions for replacement. The method provided by this invention can effectively improve the reliability and economy of the unit. Attached Figure Description

[0030] The above and other objects, features and advantages of the present invention will become more apparent from the following description of embodiments of the invention with reference to the accompanying drawings, in which:

[0031] Figure 1 A flowchart illustrating a degradation assessment method for long-life electromagnetic bearings according to an embodiment of the present invention is shown schematically.

[0032] Figure 2 The schematic diagram illustrates a degradation assessment method for long-life electromagnetic bearings according to an embodiment of the present invention.

[0033] Figure 3 The diagram illustrates the results of analyzing drop rotor displacement data using the method of an embodiment of the present invention.

[0034] Figure 4 A block diagram of a degradation assessment apparatus for long-life electromagnetic bearings, according to an embodiment of the present invention, is shown schematically. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments and the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0036] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0037] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0038] Figure 1 A flowchart illustrating a degradation assessment method for long-life electromagnetic bearings according to an embodiment of the present invention is shown.

[0039] like Figure 1 As shown, the degradation assessment method for long-life electromagnetic bearing protection bearings according to this embodiment may include steps S1 to S6.

[0040] Step S1: Collect n rotor displacement data during the fall process of the electromagnetic bearing protective bearing, where n≥2 and is an integer.

[0041] Step S2: Extract the rotational speed information that changes over time from the rotor displacement data.

[0042] Step S3: Perform time-domain analysis and wavelet packet decomposition on the rotation speed information in sequence to obtain time-domain features and wavelet packet decomposition entropy. Determine the high-dimensional feature set based on the time-domain features and wavelet packet decomposition entropy.

[0043] Step S4: Use the local preserving projection algorithm to reduce the dimensionality of the high-dimensional feature set to obtain a low-dimensional feature set.

[0044] Step S5: Extract the first low-dimensional feature of the low-dimensional feature set. For any i-th rotor displacement data in the n-th rotor displacement data, calculate the JS divergence between the i-th rotor displacement data and the (i-1)-th rotor displacement data, i = 2, 3, ..., n.

[0045] Step S6: Calculate the degradation index under different drop counts based on JS divergence, and determine the failure threshold of the protective bearing based on the degradation index.

[0046] Through the above embodiments, this invention collects multiple rotor displacement data during the drop process of the electromagnetic bearing protection bearing. Based on the JS divergence of the probability distribution of the first low-dimensional feature of two adjacent rotor displacement data sets, it determines the degradation index under different drop counts, and thus determines the failure threshold. This allows for analysis of the bearing's degradation state from multiple drop displacement data, determining whether the bearing is about to fail, and proposing replacement recommendations. The method provided by this invention can effectively improve the reliability and economy of the unit.

[0047] The following is passed Figures 2-3 The degradation assessment method for protecting short-life electromagnetic bearings according to embodiments of the present invention is described in detail.

[0048] First, the degradation assessment method for electromagnetic bearing protection bearings with shorter lifespan provided in this embodiment of the invention uses multiple rotor displacement data during the drop process of the electromagnetic bearing protection bearing as the original vibration data. Each rotor displacement data includes low-speed segment data of high-speed rotor drop data (denoted as H) or low-speed drop data of the rotor tested before each drop accident (denoted as L).

[0049] It should be noted that steps S2 and subsequent steps S3 to S6 should either be based entirely on the low-speed data H of the high-speed rotor drop data or entirely on the low-speed drop data L of the rotor tested before each drop accident. The two should not be used interchangeably.

[0050] In this embodiment of the invention, step S2 specifically includes: performing a short-time Fourier transform on the rotor displacement data of a single drop to obtain a time-frequency diagram; and determining the rotational speed information that changes with time on the time-frequency diagram based on the magnitude of the rotor speed at the time of the drop.

[0051] For example, the rotational speed information that changes over time can be represented as a curve showing the change in rotational speed over time.

[0052] In this embodiment of the invention, in step S3, time-domain analysis is performed on the rotation speed information to obtain time-domain features. These time-domain features include dimensional features such as mean, variance, root mean square, root mean square amplitude, peak value, peak-to-peak value, etc., as well as dimensionless features such as kurtosis, skewness, impulse factor, margin factor, peak factor, waveform factor, etc.

[0053] In this embodiment of the invention, in step S3, the rotation speed information is sequentially subjected to time-domain analysis and wavelet packet decomposition to obtain time-domain features and wavelet packet decomposition entropy. This includes: using the db5 mother wavelet to perform 7-level wavelet packet decomposition on the time-domain features to extract low-frequency band information below 625Hz; analyzing the low-frequency band information to obtain wavelet packet decomposition entropy, which includes wavelet packet decomposition energy entropy, wavelet packet decomposition singular value entropy, and fuzzy entropy corresponding to each frequency band.

[0054] Figure 2 The schematic diagram illustrates a degradation assessment method for short-life electromagnetic bearings according to an embodiment of the present invention.

[0055] like Figure 2 As shown, in step S3, a high-dimensional feature set is determined based on the temporal features and wavelet packet decomposition entropy, including: collecting features from four sensors respectively, wherein each sensor feature includes temporal features and wavelet packet decomposition entropy; and merging the four sensor features into a high-dimensional feature set.

[0056] In this embodiment of the invention, in step S4, the high-dimensional feature set is reduced in dimensionality using a local preserving projection algorithm to obtain a low-dimensional feature set. The specific steps are as follows:

[0057] Suppose a high-dimensional feature set X = [x1, x2, ..., xn] n ], x i Let i represent the i-th high-dimensional feature, i = 1, 2, ..., n. The transformation matrix A maps the high-dimensional feature set to a low-dimensional feature set Y = [y1, y2, ..., yn]. n The matrix transformation formula is expressed as:

[0058] y i =A T x i (1)

[0059] The transformation matrix A is obtained by minimizing the objective function, and can be expressed by the formula:

[0060] A=argmin∑ ij (y i -y j ) 2 W ij (2)

[0061] In the formula, y i Let W represent the i-th low-dimensional feature, j = 1, 2, ..., n; ij represents the weight between the i-th and j-th low-dimensional features; argmin represents the variable value that minimizes the objective function;

[0062] Substitute formula (1) into formula (2) and add the constraint condition y. T When Dy = 1, we can obtain...

[0063] A = argminA T XLX T A (3)

[0064] Then, we transform formula (3) into a problem of finding generalized eigenvalues, that is...

[0065] XLX T A=λXDX T A (4)

[0066] Therefore, the first l eigenvectors of formula (4) can be solved, i.e., A = [a0, a1, ..., a1]. l-1 ];

[0067] After obtaining the transformation matrix A, we can use formula (1) to solve for the low-dimensional feature set.

[0068] Therefore, the low-dimensional feature set Y = [y1, y2, ..., y] is subsequently taken. n The first low-dimensional feature y1 is analyzed.

[0069] Please continue reading. Figure 2In this embodiment of the invention, step S5 specifically includes:

[0070] The first low-dimensional feature of the low-dimensional feature set is denoted as the target feature. The target feature is then converted into a histogram and normalized, and finally converted into a probability distribution form.

[0071] Based on the target features in the form of probability distributions, calculate the JS divergence JS(i, i-1) between the probability distribution of the first low-dimensional feature of the i-th rotor displacement data and the probability distribution of the first low-dimensional feature of the (i-1)-th rotor displacement data.

[0072] Therefore, the JS divergence is the JS divergence between the probability distributions of the first low-dimensional feature of two adjacent rotor displacement data.

[0073] After obtaining the JS divergence JS(i, i-1), in step S6, the degradation index under different number of drops is calculated based on the JS divergence, including for any i-th rotor displacement data in n rotor displacement data, the following operations are performed:

[0074] Calculate the average value F(i) of the first i rotor displacement data and the average value F(i-1) of the first i-1 rotor displacement data for n rotor displacement data respectively;

[0075] Calculate the relative change of the average value F(i) with respect to the average value F(i-1), ΔF = F(i) / F(i-1);

[0076] The JS divergence JS(i, i-1) is corrected using the relative change ΔF to obtain the corrected JS divergence HC. JS (i) = JS(i, i-1)·ΔF;

[0077] The corrected JS divergence HC corresponding to the i-th rotor displacement data JS (i) Divide by the corrected JS divergence HC corresponding to the (i-1)th rotor displacement data. JS (i-1), to obtain the degradation index H(i) = HC under the i-th drop. JS (i) / HC JS (i-1).

[0078] Since the degradation index H(i) mentioned above is a relative value obtained by division, it belongs to the comparison of different damage stages of the same unit. Therefore, it can eliminate the differences in rotor quality, speed, protective bearing size, performance and other factors between different units to a certain extent.

[0079] Therefore, based on the experimental data of a unit failure, the failure threshold can be determined, and then the drop data of similar units can be used to evaluate the degradation.

[0080] Figure 3The diagram illustrates the results of analyzing drop rotor displacement data using the method of an embodiment of the present invention.

[0081] like Figure 3 As shown, a total of 20 high-speed drops were conducted during the experiment. Before each high-speed drop, a low-speed drop test was performed. The data obtained were L1, L2, ..., L... 20 This example only analyzes low-speed test data. The first data point corresponds to the value of the 3rd test, and the last data point corresponds to the value of the 20th test. It can be seen that the indicators gradually decrease in the early stages, show an upward trend in the middle stages, and rapidly increase in the later stages, exceeding the initial value. Since the bearing did not ultimately fail in this test, a definitive failure threshold cannot be given. However, the trend of indicator changes shows that once the bearing fails, its indicators will be significantly higher than in other drops. Therefore, the indicator corresponding to the last safe drop can be used as the failure threshold. For example, if the bearing fails during the 20th high-speed drop, then the last safe drop would be the 19th. If analyzing high-speed accident data, the indicator corresponding to the 19th drop would be used as the failure threshold. However, in this example, since low-speed test data is analyzed, if the bearing fails during the 20th high-speed drop, the indicator corresponding to the 20th low-speed test can be used instead of this failure threshold. The above failure threshold can be used to guide the evaluation of similar long-life protective bearing units. "Long life" means that the protective bearing is predicted to withstand more than 10 drops during the design phase. Because the methods for calculating the lifespan of protective bearings are not yet mature during the design phase, there may be some errors in the predicted lifespan of protective bearings. For example, if the protective bearing is predicted to withstand 15 drops, it may suddenly fail and cause unit damage when the number of drops exceeds 10. In this case, the method proposed in this invention helps to determine whether the protective bearing needs to be replaced.

[0082] In summary, the present invention provides a degradation assessment method for long-life electromagnetic bearing protection bearings. By collecting multiple rotor displacement data during the drop test of the electromagnetic bearing protection bearing, and determining the degradation index under different drop counts based on the JS divergence of the probability distribution of the first low-dimensional feature of two adjacent rotor displacement data points, the failure threshold is determined. This method can analyze the bearing's degradation state from multiple drop displacement data, determine whether the bearing is about to fail, and propose replacement recommendations. The method provided by this invention can effectively improve the reliability and economy of the unit.

[0083] Based on the methods disclosed above, the present invention also provides a degradation assessment device suitable for protecting long-life electromagnetic bearings, which will be described below in conjunction with... Figure 4 The device is described in detail.

[0084] Figure 4A block diagram of a degradation assessment apparatus for long-life electromagnetic bearings, according to an embodiment of the present invention, is shown schematically.

[0085] like Figure 4 As shown, the degradation assessment device 400 for long-life electromagnetic bearing protection bearings according to this embodiment includes a drop data acquisition module 410, a rotational speed information extraction module 420, a high-dimensional feature determination module 430, a high-dimensional feature dimensionality reduction module 440, a JS divergence calculation module 450, and a failure threshold determination module 460.

[0086] The drop data acquisition module 410 is used to acquire n rotor displacement data during the drop process of the electromagnetic bearing protective bearing, where n≥2 and is an integer.

[0087] The rotational speed information extraction module 420 is used to extract time-varying rotational speed information from the rotor displacement data;

[0088] The high-dimensional feature determination module 430 is used to sequentially perform time-domain analysis and wavelet packet decomposition on the rotation speed information to obtain time-domain features and wavelet packet decomposition entropy, and determine a high-dimensional feature set based on the time-domain features and wavelet packet decomposition entropy.

[0089] The high-dimensional feature reduction module 440 is used to reduce the dimensionality of the high-dimensional feature set using a local preservation projection algorithm to obtain a low-dimensional feature set;

[0090] The JS divergence calculation module 450 is used to extract the first low-dimensional feature of the low-dimensional feature set, and to calculate the JS divergence between the i-th rotor displacement data and the (i-1)-th rotor displacement data for any i-th rotor displacement data in the n-th rotor displacement data, i = 2, 3, ..., n;

[0091] The failure threshold determination module 460 is used to calculate the degradation index under different drop counts based on the JS divergence, and determine the failure threshold of the protective bearing based on the degradation index.

[0092] It should be noted that the embodiments of the device section are similar to those of the method section, and the technical effects achieved are also similar. For specific details, please refer to the above-mentioned method embodiment section, which will not be repeated here.

[0093] According to embodiments of the present invention, any multiple of the drop data acquisition module 410, rotation speed information extraction module 420, high-dimensional feature determination module 430, high-dimensional feature dimensionality reduction module 440, JS divergence calculation module 450, and failure threshold determination module 460 can be combined into one module, or any one of these modules can be split into multiple modules. Alternatively, at least some of the functions of one or more of these modules can be combined with at least some of the functions of other modules and implemented in one module. According to embodiments of the present invention, at least one of the drop data acquisition module 410, rotation speed information extraction module 420, high-dimensional feature determination module 430, high-dimensional feature dimensionality reduction module 440, JS divergence calculation module 450, and failure threshold determination module 460 can be at least partially implemented as a hardware circuit, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or any other reasonable method of integrating or packaging the circuit, or implemented in hardware or firmware, or in any one of software, hardware, and firmware implementations, or in a suitable combination of any of these. Alternatively, at least one of the drop data acquisition module 410, rotation speed information extraction module 420, high-dimensional feature determination module 430, high-dimensional feature dimensionality reduction module 440, JS divergence calculation module 450, and failure threshold determination module 460 can be at least partially implemented as a computer program module, which can perform corresponding functions when the computer program module is run.

[0094] The accompanying drawings show some block diagrams and / or flowcharts. It should be understood that some blocks or combinations thereof in the block diagrams and / or flowcharts can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when executed by the processor, these instructions can create means for implementing the functions / operations described in these block diagrams and / or flowcharts.

[0095] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified. Furthermore, the word "a" or "an" preceding an element does not exclude the existence of multiple such elements. For example, in the specific content of step S5, based on the target feature in the form of a probability distribution, the JS divergence JS(i, i-1) between the probability distribution of the first low-dimensional feature of the i-th rotor displacement data and the probability distribution of the first low-dimensional feature of the (i-1)-th rotor displacement data is calculated. In other embodiments, the analysis may not be limited to the first low-dimensional feature, but may also be performed on multiple low-dimensional features in the dimensionality-reduced low-dimensional feature set.

[0096] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A degradation assessment method applicable to long-life electromagnetic bearings for protecting bearings, characterized in that, include: Step S1: Collect n rotor displacement data during the fall process of the electromagnetic bearing protective bearing, where n≥2 and is an integer; Step S2: Extract the rotational speed information that changes over time from the rotor displacement data; Step S3: Perform time-domain analysis and wavelet packet decomposition on the rotation speed information in sequence to obtain time-domain features and wavelet packet decomposition entropy, and determine the high-dimensional feature set based on the time-domain features and wavelet packet decomposition entropy; Step S4: Use the local preserving projection algorithm to reduce the dimensionality of the high-dimensional feature set to obtain a low-dimensional feature set; Step S5: Extract the first low-dimensional feature of the low-dimensional feature set. For any i-th rotor displacement data in the n rotor displacement data, calculate the JS divergence between the i-th rotor displacement data and the (i-1)-th rotor displacement data, i=2,3,…,n. Step S6: Calculate the degradation index under different drop counts based on the JS divergence, and determine the failure threshold of the protective bearing based on the degradation index.

2. The degradation assessment method for long-life electromagnetic bearings according to claim 1, characterized in that, Each rotor displacement data includes low-speed segment data of high-speed rotor drop data or low-speed drop data of rotor tested before each drop accident; Step S2 specifically includes: A short-time Fourier transform is performed on the rotor displacement data from a single drop to obtain a time-frequency diagram; Based on the rotor speed at the moment of impact, the rotational speed information as a function of time is determined on the time-frequency graph.

3. The degradation assessment method for long-life electromagnetic bearings according to claim 1, characterized in that, In step S3, the time-domain features include dimensional features and dimensionless features. The dimensional features include mean, variance, root mean square, root square amplitude, peak value or peak-to-peak value. The dimensionless features include kurtosis, skewness, impulse factor, margin factor, peak factor or waveform factor.

4. The degradation assessment method for long-life electromagnetic bearings according to claim 1, characterized in that, In step S3, the rotational speed information is sequentially subjected to time-domain analysis and wavelet packet decomposition to obtain time-domain features and wavelet packet decomposition entropy, including: The time-domain features were decomposed into 7-level wavelet packets using the db5 mother wavelet to extract low-frequency band information below 625Hz. The low-frequency band information is analyzed to obtain the wavelet packet decomposition entropy, which includes the wavelet packet decomposition energy entropy, the wavelet packet decomposition singular value entropy, and the fuzzy entropy corresponding to each frequency band.

5. The degradation assessment method for long-life electromagnetic bearings according to claim 1, characterized in that, In step S3, a high-dimensional feature set is determined based on the time-domain features and wavelet packet decomposition entropy, including: Four sensor features are collected respectively, wherein each sensor feature includes a time-domain feature and a wavelet packet decomposition entropy; The features of the four sensors are combined into a high-dimensional feature set.

6. The degradation assessment method for long-life electromagnetic bearings according to claim 1, characterized in that, Step S5 specifically includes: The first low-dimensional feature of the low-dimensional feature set is denoted as the target feature. The target feature is then converted into a histogram and normalized, and then converted into a probability distribution form. Based on the target features in the form of probability distributions, calculate the JS divergence between the probability distribution of the first low-dimensional feature of the i-th rotor displacement data and the probability distribution of the first low-dimensional feature of the (i-1)-th rotor displacement data. .

7. The degradation assessment method for long-life electromagnetic bearings according to claim 6, characterized in that, In step S6, calculating the degradation index for different number of drops based on the JS divergence includes performing the following operations for any i-th rotor displacement data in the n rotor displacement data: Calculate the average value of the first i rotor displacement data for each of the n rotor displacement data. and the average value of the rotor displacement data of the first i-1 times. ; Calculate the average value Relative to the average relative change ; Using the relative change Regarding the JS divergence After correction, the corrected JS divergence is obtained. ; The corrected JS divergence corresponding to the i-th rotor displacement data Divide by the corrected JS divergence corresponding to the (i-1)th rotor displacement data The degradation index under the i-th fall is obtained. .

8. A degradation assessment device suitable for protecting long-life electromagnetic bearings, characterized in that, include: The drop data acquisition module is used to collect n rotor displacement data during the drop process of the electromagnetic bearing protective bearing, where n≥2 and is an integer; The rotational speed information extraction module is used to extract time-varying rotational speed information from the rotor displacement data; The high-dimensional feature determination module is used to sequentially perform time-domain analysis and wavelet packet decomposition on the rotation speed information to obtain time-domain features and wavelet packet decomposition entropy, and determine a high-dimensional feature set based on the time-domain features and wavelet packet decomposition entropy. The high-dimensional feature reduction module is used to reduce the dimensionality of the high-dimensional feature set using a local preservation projection algorithm to obtain a low-dimensional feature set; The JS divergence calculation module is used to extract the first low-dimensional feature of the low-dimensional feature set, and to calculate the JS divergence between the i-th rotor displacement data and the (i-1)-th rotor displacement data for any i-th rotor displacement data in the n-th rotor displacement data, i=2,3,…,n; The failure threshold determination module is used to calculate the degradation index under different drop counts based on the JS divergence, and determine the failure threshold of the protective bearing based on the degradation index.

Images