A bearing failure assessment method and system

Abstract

The application belongs to the technical field of rotating machinery condition monitoring and fault diagnosis. A bearing failure evaluation method and system are provided. The bearing vibration signal is obtained and processed by mean removal and absolute value to obtain a to-be-processed signal. Then, the lower threshold of the tail part is determined, and the tail sample set with an amplitude greater than the threshold is screened out. Then, a soft-truncated power-law distribution model is constructed based on the tail sample set, and the distribution parameters including the power-law tail index, the truncated scale parameter and the truncated smooth index are solved by maximum likelihood estimation. Then, the analytical expression of the improved Gini index is constructed according to the distribution parameters, and the bearing degradation degree index is calculated. Finally, the bearing degradation curve is constructed according to the degradation degree index of different time periods. The application accurately describes the impact characteristics of the tail part of the vibration signal by the soft-truncated power-law model, and quantifies the degradation degree by the improved Gini index, which can effectively evaluate the health status and failure trend of the rolling bearing.

Description

Technical Field

[0001] This invention relates to the field of rotating machinery condition monitoring and fault diagnosis technology, specifically to a bearing failure assessment method and system. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] As a critical supporting component in rotating machinery systems, the operating condition of rolling bearings directly affects the safety and reliability of the equipment. In aerospace, wind power equipment, rail transportation, and intelligent manufacturing, real-time monitoring and degradation assessment of bearing health status has become a core task of predictive maintenance. Current mainstream methods typically construct health indicators (HI) based on vibration signal analysis to quantify the bearing degradation process. With the increasing complexity of industrial equipment, vibration signals exhibit significant nonlinearity, non-stationarity, and sparse impact characteristics, especially in the early stages of failure, where impact events, though weak, exhibit heavy-tailed distribution characteristics. This has prompted researchers to explore modeling methods that better fit the statistical characteristics of actual signals to achieve accurate characterization and trend tracking of the degradation process.

[0004] However, existing methods for constructing health indicators generally rely on Gaussian or finite variance distribution assumptions, making it difficult to accurately describe the power-law heavy-tailed characteristics of vibration signals in early failures. When the signal tail decays slowly and the theoretical variance may not exist, traditional statistics such as kurtosis or root mean square values ​​are prone to drastic fluctuations, resulting in discontinuous and non-monotonic degradation trends, which affects the reliability of lifespan prediction. At the same time, most methods require relative comparisons based on health stage data, and the indicators lack absolute quantitative significance and are greatly affected by changes in operating conditions. In addition, although data-driven methods such as deep learning have certain effects, they are highly dependent on complete life cycle label data, making them difficult to promote and apply in actual industrial scenarios. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a bearing failure assessment method and system. By statistically modeling the tail samples of vibration signals, an exponential truncation term is introduced to construct a soft-truncation power-law model, ensuring the existence of the distribution moment while maintaining the power-law tail characteristics. Furthermore, the distribution parameters are obtained through maximum likelihood estimation, and the Gini exponent is expressed as a parametric function. An incomplete Gamma function is used to calculate the closed-form solution of the integral term, thereby constructing an analytical and normalizable degradation index and improving the accuracy of rolling bearing degradation assessment.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a bearing failure assessment method.

[0007] A bearing failure assessment method includes the following procedures: The vibration signal during the operation of the rolling bearing is acquired, and the vibration signal is processed by removing the mean and absolute value to obtain the signal to be processed; Determine the lower threshold of the tail of the signal to be processed, and select a set of tail samples from the signal to be processed whose amplitude is greater than the lower threshold of the tail. A soft-truncated power-law distribution model is constructed based on the tail sample set, and the distribution parameters of the soft-truncated power-law distribution model are solved by maximum likelihood estimation. The distribution parameters include the power-law tail exponent, the truncation scale parameter, and the truncation smoothing exponent. An analytical expression for the improved Gini index is constructed based on the distribution parameters, and the degradation index of rolling bearings is calculated. Degradation curves for rolling bearings are constructed based on degradation indicators for different time periods.

[0008] In one implementation of the first aspect of the present invention, determining the lower threshold of the tail of the signal to be processed includes: Acquire health signals of rolling bearings during normal operation; Calculate the quantile values ​​of the health signal and determine the quantile values ​​as the lower threshold of the tail.

[0009] In one implementation of the first aspect of the present invention, a soft-truncation power-law distribution model is constructed based on the tail sample set, including: Assume that the random variable of the impact magnitude in the tail sample set follows a soft-truncated power-law distribution in the form of a survival function, which includes a normalization constant, a power-law tail exponent, a truncation scale parameter, a truncation smoothing exponent, and a tail lower bound threshold. The normalization constant is solved by algebraic operations based on the lower bound threshold of the tail, the truncation scale parameter, the truncation smoothing exponent, and the power-law tail exponent.

[0010] In one implementation of the first aspect of the present invention, solving for the distribution parameters of the soft-truncated power-law distribution model by maximum likelihood estimation includes: Based on the magnitude variable, normalization constant, power-law tail exponent, truncation scale parameter, and truncation smoothing exponent of the random variable of impact magnitude, the probability density function of the soft-truncation power-law distribution model is constructed. Construct a log-likelihood function containing each sample point in the tail sample set. The log-likelihood function is the sum of the logarithmic values ​​of each sample point under the probability density function. Find the parameter vector that maximizes the log-likelihood function, and use the parameter vector as the estimated distribution parameters.

[0011] In one implementation of the first aspect of the present invention, constructing an analytical expression for the improved Gini exponent based on the distribution parameters includes: ; in, To improve the Gini index, The mean, For survival function, The magnitude of the impact is a random variable; For amplitude variables, This is the lower bound threshold for the samples at the tail end of the vibration signal.

[0012] In one implementation of the first aspect of the present invention, the degradation degree index of the rolling bearing is calculated, including: ; in, Indicates the degree of degradation; The mathematical expectation of the random variable representing the amplitude of vibration and impact; The power-law tail exponent represents the soft-truncation power-law distribution; The truncation smoothing exponent represents the soft-truncation power-law distribution; The cutoff scale parameter represents the soft-truncation power-law distribution; The lower bound threshold representing the sample at the tail end of the vibration signal; This represents an incomplete Gamma function.

[0013] In one implementation of the first aspect of the present invention, constructing the degradation curve of the rolling bearing includes: The vibration signal sample segments are divided into multiple segments in chronological order. For each vibration signal sample segment, the steps of screening, model building, solving distribution parameters and calculating degradation index are performed to obtain the degradation index sequence at the corresponding time point. The degradation index sequence is arranged in chronological order to form a degradation curve characterizing the failure trend of rolling bearings.

[0014] Secondly, the present invention provides a bearing failure assessment system.

[0015] A bearing failure assessment system, comprising: The signal processing unit is configured to: acquire the vibration signal during the operation of the rolling bearing, and perform mean and absolute value removal processing on the vibration signal to obtain the signal to be processed; The sample filtering unit is configured to: determine the lower tail threshold of the signal to be processed, and filter out a set of tail samples from the signal to be processed whose amplitude is greater than the lower tail threshold. The parameter estimation unit is configured to: construct a soft-truncated power-law distribution model based on the tail sample set, and solve the distribution parameters of the soft-truncated power-law distribution model through maximum likelihood estimation. The distribution parameters include the power-law tail exponent, the truncation scale parameter, and the truncation smoothing exponent. The index calculation unit is configured to: construct an analytical expression for the improved Gini index based on the distribution parameters, and calculate the degradation index of the rolling bearing; The curve construction unit is configured to construct the degradation curve of the rolling bearing based on the degradation degree index at different time periods.

[0016] Thirdly, the present invention provides a computer-readable storage medium storing a computer program adapted to be loaded by a processor and executed by the bearing failure assessment method of the first aspect of the present invention.

[0017] Fourthly, the present invention provides a computer device, comprising: a processor and a computer-readable storage medium; A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the bearing failure assessment method of the first aspect of the present invention.

[0018] Compared with the prior art, the beneficial effects of the present invention are: This invention effectively solves the problem that existing technologies struggle to accurately characterize the heavy-tailed impact characteristics of vibration signals in early rolling bearing failures, leading to large fluctuations in health indicators and a lack of absolute quantitative significance. By constructing a degradation assessment method based on a soft-truncation power-law model, this invention addresses the issue that existing technologies struggle to accurately depict the heavy-tailed impact characteristics of vibration signals in early rolling bearing failures, resulting in large fluctuations in health indicators and a lack of absolute quantitative significance. Specifically, this invention first performs mean and absolute value removal processing on the original vibration signal to highlight the impact component. Then, it focuses on the tail samples of the signal, accurately modeling the statistical characteristics of the coexistence of heavy tails and truncation using a soft-truncated power-law distribution. Physically interpretable distribution parameters (including the power-law tail exponent, truncation scale parameter, and truncation smoothing exponent) are obtained through maximum likelihood estimation. Based on this, an analytical expression for the improved Gini index is innovatively derived. This index relies only on the tail distribution parameters, requires no health baseline data, and possesses absolute quantification capabilities. As bearing degradation intensifies, the tail characteristics of the impact amplitude distribution undergo a regular evolution. The proposed index can stably and monotonically reflect this process, significantly improving the continuity and robustness of the degradation trend. Finally, a complete degradation curve is constructed through the time-series degradation index, providing a reliable basis for predictive maintenance and overcoming the shortcomings of traditional methods in unlabeled, variable-condition scenarios with poor adaptability.

[0019] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0020] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0021] Figure 1 A schematic flowchart of a bearing failure assessment method provided as an exemplary embodiment of the present invention; Figure 2 A schematic diagram of the time-domain signals of the first six samples in the health stage provided as an exemplary embodiment of the present invention, wherein, Figure 2 (a) in the diagram is a schematic diagram of the time-domain signal of sample 1. Figure 2 (b) in the diagram is a schematic diagram of the time-domain signal of sample 2. Figure 2 (c) in the diagram is a schematic diagram of the time-domain signal of sample 3. Figure 2 (d) in the diagram is a schematic diagram of the time-domain signal of sample 4. Figure 2 (e) in the diagram is a time-domain signal diagram of sample 5. Figure 2 (f) in the diagram is a time-domain signal diagram of sample 6; Figure 3 A schematic diagram illustrating the fitting effect of the STPL model at different stages is provided as an exemplary embodiment of the present invention, wherein, Figure 3 (a) in the figure is a schematic diagram of the fitting effect of STPL in the healthy stage. Figure 3 (b) in the figure is a schematic diagram of the fitting effect of STPL in the early fault stage. Figure 3 (c) in the figure is a schematic diagram of the fitting effect of STPL in the severe fault stage; Figure 4 This is a schematic diagram comparing the proposed index of the present invention with other indexes, provided as an exemplary embodiment of the present invention. Figure 4 (a) in the middle is Exponential curve, Figure 4 (b) in the figure is the original nonparametric Gini exponential curve. Figure 4 (c) in the figure represents the kurtosis index curve. Figure 4 In the figure, (d) represents the RMS curve; Figure 5 A schematic diagram of a bearing failure assessment system provided as an exemplary embodiment of the present invention; Figure 6 A schematic diagram of a computer device provided for an exemplary embodiment of the present invention. Detailed Implementation

[0022] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0023] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0024] Existing methods for constructing bearing health indicators mainly suffer from the following technical shortcomings: the Gaussian distribution assumption does not match the characteristics of heavy-tailed impacts; some methods use Gaussian mixture models or finite variance probability distributions to model vibration signals. However, in the early stages of bearing degradation, vibration signals often exhibit power-law tail characteristics, meaning that the probability of extreme impact amplitudes occurring is higher than expected by the Gaussian distribution. When the signal tail exponent is less than 1... When the variance is less than 2, the theoretical variance may not exist, and traditional Gaussian models struggle to accurately describe the tail structure, leading to significant fluctuations in health indicators or insensitivity to early shocks. Health indicators are highly dependent on the baseline, being relative measures. Some methods construct health indicators by calculating the difference between the current state distribution and the baseline distribution of the health stage. For example, methods based on Gaussian distribution overlap or reconstruction error require pre-selecting health stage samples as a reference. This method is inherently a relative measure, and indicator values ​​are significantly affected by the selection of health samples and changes in operating conditions, making it difficult to establish a unified absolute quantitative standard. Deep learning methods rely on large amounts of prior data. Convolutional neural networks, long short-term memory networks, and bidirectional gated recurrent units are used to construct health indicators. These methods typically require complete lifecycle data and residual lifespan labels for supervised training. Model performance is highly dependent on the size and quality of training samples. In real-world industrial scenarios, equipment often lacks complete lifecycle data or accurate labels, limiting the widespread application of these models.

[0025] To address the issues of mismatch in heavy-tail modeling, dependence on health benchmarks, and dependence on supervised data in existing technologies, this implementation proposes a bearing failure assessment method. It constructs a soft-truncated power-law tail distribution model, achieving statistical matching of the heavy-tail. While preserving the characteristics of the power-law tail, it uses an exponential truncation term to smooth the extreme impact amplitude, ensuring that the distribution can both characterize the sparsity of early impacts and guarantee the existence of higher-order moments, thus satisfying the statistical inference conditions. The Gini exponent is transformed into a parametric function for absolute quantitative expression. Addressing the problem that existing health indicators rely on relative comparisons with health benchmarks, this invention no longer uses healthy samples as a control to calculate distribution differences, but instead expresses the Gini exponent as a function of distribution parameters. After obtaining the parameter vector through maximum likelihood estimation, the degradation index is directly calculated, achieving a normalized quantitative expression of the degradation degree. This method essentially transforms "distribution difference comparison" into "distribution structure function calculation," upgrading from a relative measurement mechanism to an absolute parameter measurement mechanism. The index value is independent of the sample selection method in the health phase, maintaining a uniform quantitative scale across different devices and operating conditions, improving the comparability and universality of the index. A closed-loop solution calculation framework for the Gamma function is constructed, eliminating the dependence on numerical integration. Addressing the issue of the improved Gini index requiring infinite interval numerical integration in heavy-tailed environments, this invention transforms the integral term into an upper-incomplete Gamma function form through variable substitution, obtaining an analytical closed-loop solution expression. This closed-loop solution calculation method avoids multiple numerical integration iterations, reducing computational complexity.

[0026] More specifically, this invention uses the XJTU-SY rolling bearing accelerated life test dataset Bearing 1_3 as an example to demonstrate the specific implementation method of this invention, such as... Figure 1 As shown, the process includes the following: S1: Vibration signal acquisition and preprocessing.

[0027] S101: Obtain vibration acceleration Each sample was collected for 1.2 seconds, and each sample contained 30,720 data points. Represents the time-domain vibration acceleration signal; Represents the total number of sampling points; the sampling frequency is .

[0028] S102: Perform mean and absolute value removal processing on the signal: (1); Here, mean() is used to calculate the mean.

[0029] S103: Threshold determination (lower bound of the tail), acquire the first 10 samples as the health phase signal (no less than 5 seconds), denoted as... Calculate the 90th percentile, then the lower bound threshold for the normal operation phase. ,but: (2); like Figure 2 The image shows a schematic diagram of the time-domain signals of the first six samples in the healthy phase. Figure 2 (a) in the diagram is a schematic diagram of the time-domain signal of sample 1. Figure 2 (b) in the diagram is a schematic diagram of the time-domain signal of sample 2. Figure 2 (c) in the diagram is a schematic diagram of the time-domain signal of sample 3. Figure 2 (d) in the diagram is a schematic diagram of the time-domain signal of sample 4. Figure 2(e) in the diagram is a time-domain signal diagram of sample 5. Figure 2 (f) in the figure is a schematic diagram of the time domain signal of sample 6.

[0030] S104: Filter tail samples, the set of tail samples used for modeling. (i=1,2,…,n).

[0031] S2: Modeling of the soft-truncation power-law distribution of vibration signals.

[0032] S201: Assuming the tail samples follow a soft-truncated power-law distribution (STPL), the survival function is: (3); in, The magnitude of the impact is a random variable; For amplitude variables; The power-law tail exponent, with a value range set to [value range missing]. >0; The range of values ​​for the truncation scale parameter is set to [value range]. >1.027; To truncate the smoothing exponent, the range of values ​​is set to 1. >0; This is the normalization constant.

[0033] S202: Algebraic solution of the normalization constant, normalization constant Calculate using the following formula: (4); This formula avoids calculating the normalization constant through integration, thus improving computational efficiency.

[0034] S203: Maximum likelihood estimation, with the probability density function as follows: (5); in, .

[0035] S204: Constructing the log-likelihood function: (6); Solution: (7); Figure 3 The fitting effect of the STPL model at different stages and the corresponding results are shown. Numerical values, among which, Figure 3 (a) in the figure is a schematic diagram of the fitting effect of STPL in the healthy stage. Figure 3 (b) in the figure is a schematic diagram of the fitting effect of STPL in the early fault stage. Figure 3(c) in the figure is a schematic diagram of the fitting effect of STPL in the severe fault stage.

[0036] S3: Improved calculation of closed-loop solutions for the Gini exponent.

[0037] S301: The improved Gini index is defined as follows: (8); in, To improve the Gini index, This is the mean.

[0038] S302: The integral term in equation (8) can be further expressed as: (9); S303: Variable Substitution (10); We obtain a closed-form solution: (11); in, For an incomplete Gamma function, , .

[0039] S303: Final Degradation Level It can be represented as: (12); This invention improves the calculation of closed-form solutions for the Gini exponent, avoiding numerical integration.

[0040] S4: Bearing failure trend construction.

[0041] S401: Calculation of vibration signal samples at different stages ; S402: Forming a bearing degradation curve: It is used to assess the degree and trend of failure.

[0042] like Figure 4 As shown, the method of the present invention is used to calculate the full life cycle data of the same bearing. The index, the original Gini index, the kurtosis index, and the effective value index were compared and analyzed.

[0043] Figure 4 In (a), the horizontal axis represents the file number (corresponding to the time series), and the vertical axis represents... The index shows that it remains relatively stable in the early stages of degradation, exhibits a continuous upward trend in the middle and late stages, and shows a significant jump near the failure stage. The overall curve exhibits continuity and strong monotonicity. This trend reflects the statistical evolution process of the impact energy distribution gradually intensifying at the tail end.

[0044] Figure 4 (b) in the figure represents the original nonparametric Gini exponential curve, which exhibits significant fluctuations and multiple declines during the intermediate stage, resulting in a weak trend. This is because nonparametric estimation is highly sensitive to extreme samples under heavy-tailed distribution conditions, leading to a large estimation variance.

[0045] Figure 4 (c) in the figure represents the kurtosis index curve. It can be observed that the kurtosis exhibits multiple sharp peaks during the degradation process, with numerous local extreme values. Such high-frequency fluctuations are not conducive to establishing a stable degradation trend model and may lead to false alarms or misjudgments of the trend.

[0046] Figure 4 (d) in the figure represents the effective value (RMS) curve. This index changes slowly in the early stages of degradation and is not sensitive to shock-induced degradation. It only shows a significant upward trend when the overall energy increases in the later stages, making it difficult to use for early degradation identification.

[0047] according to Figure 4 (a) to Figure 4 As can be seen from (d) in this paper, the parameterized Gini index constructed by this invention exhibits a smooth increasing trend throughout the degradation process. Compared with traditional kurtosis and RMS indices, it has clearer stage characteristics and smaller fluctuation amplitude compared with the original Gini index, which is more conducive to establishing degradation trend models and remaining lifetime prediction models.

[0048] In summary, the method of the present invention has the following technical effects: (1) This invention achieves an absolute measurement of bearing degradation. The invention parameterizes the Gini index using a soft-truncation power-law model, mapping the degree of rolling bearing degradation to the interval [0,1]. A value close to 1 indicates highly concentrated impact energy and a high degree of failure, while a value close to 0 indicates relatively uniform energy distribution and a healthy bearing. Unlike traditional statistical indicators such as kurtosis, mean, and effective value, which rely on sample distribution fluctuations and amplitude fluctuations and only reflect relative trends, this method achieves an absolute quantitative expression of bearing degradation.

[0049] (2) The use of a closed-form solution using the Gamma function improves computational efficiency and stability. In the calculation of the parameterized Gini exponent, this invention transforms the integral term into an incomplete Gamma function form through variable substitution, obtaining an analytical closed-form expression. This avoids numerical integration of the integral term. Traditional improved Gini exponents, under heavy-tailed distribution conditions, typically require multiple infinite interval integration operations, resulting in high computational cost and potential instability in integration accuracy or convergence difficulties when tail decay is slow. This method directly calculates the integral term using the closed-form expression, significantly reducing computational complexity while avoiding the accumulation of integration errors and improving numerical stability. It is suitable for online real-time degradation assessment scenarios.

[0050] (3) Applicable to heavy-tail impact signal environments in early bearing failures. In the early stages of bearing degradation, the distribution of vibration signals at the tail typically exhibits power-law decay characteristics. When the tail is exponential, the theoretical variance of the signal may not exist, and traditional statistical estimations are unstable. This method constructs a soft-truncation power-law model by introducing an exponential truncation term, ensuring that the distribution at the tail exhibits smooth decay characteristics, thereby guaranteeing the existence of all moments. This modeling approach can maintain estimation stability under heavy-tail impact signal conditions, avoiding drastic fluctuations in indicators due to extreme values ​​at the tail, and improving the robustness of degradation assessment.

[0051] (4) Enhanced monotonicity of degradation trend is beneficial for lifetime prediction. Since the parameters of the soft-truncation power-law model change regularly with crack propagation, and the parameterized Gini exponent essentially reflects the gradual change trend of the tail structure, the constructed degradation curve G(t) exhibits strong continuity and monotonicity over time. Compared with the high-frequency fluctuations of traditional kurtosis or non-parametric Gini under heavy-tailed noise conditions, the degradation curve obtained by this method has a more stable trend, which is beneficial for constructing a remaining lifetime prediction model and trend extrapolation analysis.

[0052] (5) Reduced dependence on manual empirical parameters. In the process of distribution modeling and parameter estimation, the present invention does not require manual setting of impact intensity threshold, noise amplitude or empirical filtering parameters. The modeling process can be completed only through tail threshold and maximum likelihood estimation, which reduces the impact of manual parameter tuning on the evaluation results and improves the adaptability of the method under different working conditions.

[0053] Figure 5 A bearing failure assessment system is shown, comprising: The signal processing unit 501 is configured to: acquire the vibration signal during the operation of the rolling bearing, and perform mean and absolute value removal processing on the vibration signal to obtain the signal to be processed; The sample filtering unit 502 is configured to: determine the lower threshold of the tail of the signal to be processed, and filter out a set of tail samples from the signal to be processed whose amplitude is greater than the lower threshold of the tail. The parameter estimation unit 503 is configured to: construct a soft-truncated power-law distribution model based on the tail sample set, and solve the distribution parameters of the soft-truncated power-law distribution model through maximum likelihood estimation. The distribution parameters include the power-law tail exponent, the truncation scale parameter, and the truncation smoothing exponent. The index calculation unit 504 is configured to: construct an analytical expression for the improved Gini index based on the distribution parameters, and calculate the degradation index of the rolling bearing; Curve construction unit 505 is configured to construct the degradation curve of the rolling bearing based on the degradation degree index for different time periods.

[0054] It is understood that the aforementioned units can be individually or entirely merged into one or more other units, or some of the units can be further divided into multiple functionally smaller units. This achieves the same operation without affecting the technical effects of the embodiments of the present invention. The aforementioned units are based on logical functional division. In practical applications, the function of one unit can be implemented by multiple units, or the function of multiple units can be implemented by one unit. In other embodiments of the present invention, the system may also include other units. In practical applications, these functions can also be implemented with the assistance of other units, and can be implemented collaboratively by multiple units.

[0055] According to another embodiment of the present invention, the system of this embodiment can be constructed by running a computer program (including program code) capable of performing the steps involved in the corresponding method of the present invention on a general-purpose computing device, such as a computer, which includes processing elements and storage elements such as a central processing unit (CPU), random access memory (RAM), and read-only memory (ROM). The computer program can be recorded on, for example, a computer-readable recording medium, loaded into the aforementioned computing device through the computer-readable recording medium, and run therein.

[0056] Figure 6 A computer device is shown, which includes a processor 601, a communication interface 602, and a computer-readable storage medium 603. The processor 601, communication interface 602, and computer-readable storage medium 603 can be connected via a bus or other means.

[0057] The communication interface 602 is used to receive and send data. The computer-readable storage medium 603 can be stored in the memory of the electronic device. The computer-readable storage medium 603 is used to store computer programs, which include program instructions. The processor 601 is used to execute the program instructions stored in the computer-readable storage medium 603.

[0058] The processor 601 is the computing and control core of an electronic device. It is suitable for implementing one or more instructions, specifically for loading and executing one or more instructions to achieve the corresponding method flow or corresponding function.

[0059] Processor 601 is configured to perform the following procedure: The vibration signal during the operation of the rolling bearing is acquired, and the vibration signal is processed by removing the mean and absolute value to obtain the signal to be processed; Determine the lower threshold of the tail of the signal to be processed, and select a set of tail samples from the signal to be processed whose amplitude is greater than the lower threshold of the tail. A soft-truncated power-law distribution model is constructed based on the tail sample set, and the distribution parameters of the soft-truncated power-law distribution model are solved by maximum likelihood estimation. The distribution parameters include the power-law tail exponent, the truncation scale parameter, and the truncation smoothing exponent. An analytical expression for the improved Gini index is constructed based on the distribution parameters, and the degradation index of rolling bearings is calculated. Degradation curves for rolling bearings are constructed based on degradation indicators for different time periods.

[0060] This invention also provides a computer-readable storage medium, which is a memory device in an electronic device for storing programs and data. It is understood that the computer-readable storage medium here may include both built-in storage media in the electronic device and extended storage media supported by the electronic device. The computer-readable storage medium provides storage space for storing the processing system of the electronic device.

[0061] Furthermore, this storage space also contains one or more instructions suitable for loading and execution by the processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory; alternatively, it can also be at least one computer-readable storage medium located remotely from the aforementioned processor.

[0062] In one embodiment, the computer-readable storage medium stores one or more instructions; the processor loads and executes the one or more instructions stored in the computer-readable storage medium to perform the following process: The vibration signal during the operation of the rolling bearing is acquired, and the vibration signal is processed by removing the mean and absolute value to obtain the signal to be processed; Determine the lower threshold of the tail of the signal to be processed, and select a set of tail samples from the signal to be processed whose amplitude is greater than the lower threshold of the tail. A soft-truncated power-law distribution model is constructed based on the tail sample set, and the distribution parameters of the soft-truncated power-law distribution model are solved by maximum likelihood estimation. The distribution parameters include the power-law tail exponent, the truncation scale parameter, and the truncation smoothing exponent. An analytical expression for the improved Gini index is constructed based on the distribution parameters, and the degradation index of rolling bearings is calculated. Degradation curves for rolling bearings are constructed based on degradation indicators for different time periods.

[0063] The present invention also provides a computer program product or computer program comprising computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the following process: The vibration signal during the operation of the rolling bearing is acquired, and the vibration signal is processed by removing the mean and absolute value to obtain the signal to be processed; Determine the lower threshold of the tail of the signal to be processed, and select a set of tail samples from the signal to be processed whose amplitude is greater than the lower threshold of the tail. A soft-truncated power-law distribution model is constructed based on the tail sample set, and the distribution parameters of the soft-truncated power-law distribution model are solved by maximum likelihood estimation. The distribution parameters include the power-law tail exponent, the truncation scale parameter, and the truncation smoothing exponent. An analytical expression for the improved Gini index is constructed based on the distribution parameters, and the degradation index of rolling bearings is calculated. Degradation curves for rolling bearings are constructed based on degradation indicators for different time periods.

[0064] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can implement the described functions using different methods for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0065] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of the present invention is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic cable, digital cable) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data processing device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive), etc.

[0066] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A bearing failure assessment method, characterized in that, The process includes the following: The vibration signal during the operation of the rolling bearing is acquired, and the vibration signal is processed by removing the mean and absolute value to obtain the signal to be processed; Determine the lower threshold of the tail of the signal to be processed, and select a set of tail samples from the signal to be processed whose amplitude is greater than the lower threshold of the tail. A soft-truncated power-law distribution model is constructed based on the tail sample set, and the distribution parameters of the soft-truncated power-law distribution model are solved by maximum likelihood estimation. The distribution parameters include the power-law tail exponent, the truncation scale parameter, and the truncation smoothing exponent. An analytical expression for the improved Gini index is constructed based on the distribution parameters, and the degradation index of rolling bearings is calculated. Construct an expression for the improved Gini exponent based on the distribution parameters, including: ； in, To improve the Gini index, The mean, For survival function, The magnitude of the impact is a random variable; For amplitude variables, This is the lower bound threshold for the samples at the tail end of the vibration signal; The degradation index of the rolling bearing is calculated, including: ； in, Indicates the degree of degradation; The mathematical expectation of the random variable representing the amplitude of vibration and impact; The power-law tail exponent represents the soft-truncation power-law distribution; The truncation smoothing exponent represents the soft-truncation power-law distribution; The cutoff scale parameter represents the soft-truncation power-law distribution; The lower bound threshold representing the sample at the tail end of the vibration signal; This represents an incomplete Gamma function; Degradation curves for rolling bearings are constructed based on degradation indicators for different time periods.

2. The bearing failure assessment method as described in claim 1, characterized in that, Determine the lower threshold of the tail of the signal to be processed, including: Acquire health signals of rolling bearings during normal operation; Calculate the quantile values ​​of the health signal and determine the quantile values ​​as the lower tail threshold.

3. The bearing failure assessment method as described in claim 1, characterized in that, A soft-truncation power-law distribution model is constructed based on the tail sample set, including: Assume that the random variable of the impact magnitude in the tail sample set follows a soft-truncated power-law distribution in the form of a survival function, which includes a normalization constant, a power-law tail exponent, a truncation scale parameter, a truncation smoothing exponent, and a tail lower bound threshold. The normalization constant is solved by algebraic operations based on the lower bound threshold of the tail, the truncation scale parameter, the truncation smoothing exponent, and the power-law tail exponent.

4. The bearing failure assessment method as described in claim 1, characterized in that, Solving for the distribution parameters of the soft-truncated power-law distribution model using maximum likelihood estimation includes: Based on the magnitude variable, normalization constant, power-law tail exponent, truncation scale parameter, and truncation smoothing exponent of the random variable of impact magnitude, the probability density function of the soft-truncation power-law distribution model is constructed. Construct a log-likelihood function containing each sample point in the tail sample set. The log-likelihood function is the sum of the logarithmic values ​​of each sample point under the probability density function. Find the parameter vector that maximizes the log-likelihood function, and use the parameter vector as the estimated distribution parameters.

5. The bearing failure assessment method as described in claim 1, characterized in that, Constructing the degradation curve of the rolling bearing includes: Divide the vibration signal sample segments into multiple time segments and calculate the degradation degree index sequence at the corresponding time points; The degradation index sequence is arranged in chronological order to form a degradation curve characterizing the failure trend of rolling bearings.

6. A bearing failure assessment system, characterized in that, The bearing failure assessment method as described in any one of claims 1-5 includes: The signal processing unit is configured to: acquire the vibration signal during the operation of the rolling bearing, and perform mean and absolute value removal processing on the vibration signal to obtain the signal to be processed; The sample filtering unit is configured to: determine the lower tail threshold of the signal to be processed, and filter out a set of tail samples from the signal to be processed whose amplitude is greater than the lower tail threshold. The parameter estimation unit is configured to: construct a soft-truncated power-law distribution model based on the tail sample set, and solve the distribution parameters of the soft-truncated power-law distribution model through maximum likelihood estimation. The distribution parameters include the power-law tail exponent, the truncation scale parameter, and the truncation smoothing exponent. The index calculation unit is configured to: construct an analytical expression for the improved Gini index based on the distribution parameters, and calculate the degradation index of the rolling bearing; The curve construction unit is configured to construct the degradation curve of the rolling bearing based on the degradation degree index at different time periods.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded by a processor and executed as described in any one of claims 1 to 5 for bearing failure assessment.

8. A computer device, characterized in that, include: Processor and computer-readable storage media; A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program that, when executed by the processor, implements the bearing failure assessment method as described in any one of claims 1 to 5.

Images