Method for predicting service life of rolling bearing

Abstract

The invention discloses a service life prediction method for a rolling bearing, and relates to the technical field of equipment service life prediction. The method comprises the following steps of: remarkably improving the dynamic characterization capability of a bearing degradation process by fusing a dispersion entropy and an amplitude feature; and a health state index is constructed, and the degradation process of the rolling bearing is accurately represented. Although the dispersion entropy method can effectively distinguish different health states of the bearing, continuous prediction of the degradation trend cannot be realized, the dispersion entropy introducing amplitude information can fully utilize amplitude characteristics contained in vibration signals, and the capability of identifying fault occurrence and the degradation trend is remarkably improved. In combination with an experimental platform, bearing life test data is collected for verification, and experimental results show that the method has higher accuracy and stability in the aspects of fault trend recognition and health assessment. The method is suitable for early life prediction and residual life prediction of the rolling bearing, and has a good engineering application prospect.

Description

Technical Field

[0001] This application relates to the field of equipment life prediction technology, and in particular to a method for predicting the life of rolling bearings. Background Technology

[0002] Rotating machinery is widely used in aerospace, energy equipment, rail transportation, and industrial manufacturing. Rolling bearings, as key support and transmission components, operate under high speeds, alternating loads, and complex conditions for extended periods, making them highly susceptible to degradation failures such as wear and fatigue spalling. Bearing failures are often sudden and irreversible; a malfunction can lead to system shutdown or even a safety accident. Therefore, continuously monitoring the degradation state of rolling bearings based on vibration signals collected during operation and predicting their remaining service life is a crucial technical approach for achieving condition-based maintenance, reducing operating costs, and improving system reliability. Extracting health status indicators from vibration signals that stably reflect the bearing degradation evolution process is one of the core issues in rolling bearing life prediction research.

[0003] In existing technologies, health status indicators are typically obtained by feature extraction from vibration signals to characterize the evolution of bearings from a healthy state to a failure state. Traditional methods often employ time-domain statistical features such as root mean square (RMS) values ​​and kurtosis. These indicators can reflect the energy changes and impact characteristics of vibration signals to some extent and have certain application value in early bearing fault detection. However, due to the prevalence of background noise and random interference in vibration signals, and the nonlinear and non-stationary characteristics of bearing degradation, the aforementioned statistical features are highly sensitive to noise and are prone to problems such as increased fluctuations and discontinuous trends during long-term operation, making them unsuitable as stable and reliable input features for life prediction. To address this, researchers have proposed complexity indicators such as spread entropy, based on symbolization and pattern statistics. By performing probability mapping and pattern distribution analysis on vibration signals, they improve the ability to characterize changes in the structure of nonlinear signals. While spread entropy improves the insufficient anti-interference capability of traditional statistical features to some extent, it mainly focuses on the symbol arrangement pattern of vibration signals, with limited utilization of signal amplitude variations. When a bearing enters the mid-to-late stage of degradation, the vibration amplitude often shows a continuous increasing trend. However, the spread entropy does not respond sufficiently to the global amplitude evolution, which can easily lead to insufficient monotonicity of health indicators and affect the accuracy of remaining service life prediction. Summary of the Invention

[0004] Therefore, it is necessary to provide a method for predicting the life of rolling bearings to address the aforementioned technical problems.

[0005] The present invention adopts the following technical solution: This invention provides a method for predicting the life of rolling bearings, comprising: Acquire vibration signals of rolling bearings during operation; Determine the spread entropy of the rolling bearing vibration signal based on the vibration signal. The vibration signal is added to a preset constant, and the reciprocal is taken to construct an embedding matrix for the rolling bearing vibration signal. The norm of each embedding vector in the embedding matrix is ​​determined, and the norm is normalized to a normal value using a nonlinear mapping function. The interval is used to obtain the normalized modulus values ​​of all embedded vectors; based on the preset number of amplitude partitions, the normalized modulus values ​​are then... The interval is divided into several amplitude intervals at equal intervals; the normalized modulus value of each embedded vector is assigned to the corresponding amplitude interval to obtain the amplitude mode of the amplitude interval to which each embedded vector belongs; the probability distribution of the amplitude mode of the amplitude interval to which each embedded vector belongs is statistically analyzed, and the amplitude entropy of the rolling bearing vibration signal is determined based on the probability distribution of the amplitude mode of each embedded vector. Based on the dispersion entropy and amplitude entropy, the health status index of the rolling bearing within a specified time period is determined, and the degradation curve of the rolling bearing is generated.

[0006] Preferably, acquiring the vibration signal of the rolling bearing during operation specifically includes: Install the rolling bearing system onto the experimental platform and configure a rolling bearing vibration signal data acquisition system; The rolling bearing health indicators were set and the initial working state of the experimental platform was simulated, and the rolling bearing system was ensured to operate at room temperature; the health indicators were used to characterize the degree of degradation of the rolling bearing during operation. Using a rolling bearing vibration signal data acquisition system, the rolling bearing is continuously monitored from its initial brand-new state, and the original vibration signals of the rolling bearing are continuously collected throughout its entire life cycle from operation to failure. The original vibration signal is denoised and standardized to obtain the vibration signal of the rolling bearing during operation.

[0007] Preferably, the dispersion entropy of the rolling bearing vibration signal is determined based on the vibration signal, specifically including: The vibration signal is then subjected to standard normalization and cumulative distribution function normalization in sequence to obtain the normalized vibration signal. The normalized vibration signal is discretized according to a preset number of symbol categories to generate a vibration signal symbol sequence. Based on the embedding dimension and time delay parameters of the vibration signal symbol sequence, an embedding vector of the vibration signal symbol sequence is constructed, and the scattering entropy of the rolling bearing vibration signal is calculated based on the probability distribution of the embedding vector.

[0008] Preferably, the vibration signal is sequentially subjected to standard normalization and cumulative distribution function normalization to obtain a normalized vibration signal, specifically including: The vibration signal is normalized to obtain the mean and standard deviation of the vibration signal. Based on the mean and standard deviation, the vibration signal is normalized to the interval [0,1] using the cumulative distribution function, resulting in the normalized vibration signal, as shown in the formula: ; In the formula, For normalized vibration signals, The cumulative distribution function is... It is a vibration signal. and These are the mean and standard deviation, respectively.

[0009] Preferably, the formula for calculating the vibration signal symbol sequence is: ; In the formula, It is a sequence of vibration signal symbols. The preset number of symbol categories, For normalized vibration signals, This is a function for generating random numbers.

[0010] Preferably, determining the spread entropy of the rolling bearing vibration signal based on the embedding vector specifically includes: Discretize the embedding vector of the symbol sequence, and based on the discretized embedding vector, convert the embedding vector into a unique index pattern, as shown in the formula: ; In the formula, A unique index for each embedded vector. It is the discretized embedding vector. It is the embedding dimension. This is the preset number of symbol categories; The expression for the embedding vector of the vibration signal symbol sequence is: ; In the formula, For the embedding vector of the symbol sequence, Let the embedding dimension be the number of symbols in the sequence. The time delay of the symbol sequence; Count the number of times the unique index of each embedding vector appears in the preset number of symbol categories. Calculate the probability of each embedding pattern. : ; In the formula, It is the total number of embedded vectors. It is the first The probability of the occurrence of a unique index of an embedded vector; The scatter entropy is calculated based on the probability of occurrence of the unique index of the embedded vector, using the following formula: ; In the formula, DE is the spread entropy. Let c be the probability of each embedding pattern occurring, and c be the preset number of symbol categories. Let be the embedding dimension of the symbol sequence.

[0011] Preferably, based on the vibration signal, the amplitude mode probability distribution of the amplitude interval to which each embedded vector belongs is statistically analyzed, and the amplitude entropy of the rolling bearing vibration signal is determined based on the amplitude mode probability distribution of each embedded vector, specifically including: Count the number of times the magnitude pattern appears in all embedded vectors. And calculate the probability of each pattern occurring. : ; In the formula, This represents the total number of all embedding vectors in the rolling bearing vibration signal embedding matrix. For the first The probability of each amplitude pattern; the probability distribution formed by the probabilities of amplitude patterns characterizes the energy distribution features of the vibration signal at the amplitude variation level; The amplitude entropy of the rolling bearing vibration signal is calculated based on the amplitude pattern probability distribution of each embedded vector, using the following formula: ; In the formula, The amplitude entropy of the rolling bearing vibration signal. This represents the number of partitions in the amplitude mode. This is the preset number of amplitude zones.

[0012] Preferably, the formula for calculating the health status index is: ; In the formula, MDE is the health status index of the rolling bearing vibration signal, and DE is the dispersion entropy of the rolling bearing vibration signal. The amplitude entropy of the rolling bearing vibration signal. This serves as the benchmark for the maximum entropy normalization of DE.

[0013] The above-mentioned at least one technical solution adopted in this invention can achieve the following beneficial effects: In the rolling bearing life prediction method provided by this invention, a high sampling rate vibration sensor is used to collect vibration signals of the rolling bearing throughout its entire life cycle. Based on these vibration signals, a dispersion entropy index (MDE) incorporating amplitude information is constructed as a health indicator for the rolling bearing. The MDE index, while characterizing the temporal structure complexity of the vibration signal using dispersion entropy, introduces embedded vector amplitude information. Through nonlinear mapping and probabilistic modeling of the vibration signal amplitude variation characteristics, it achieves enhanced characterization of the dynamic behavior of the vibration signal. This index can effectively amplify the subtle changes in the vibration signal during the early stages of rolling bearing degradation, improving the sensitivity and stability of the health indicator in the early stages of degradation. This allows for a more accurate characterization of the degradation process of the rolling bearing from a healthy state to a failure state, providing a reliable feature basis for subsequent life prediction.

[0014] In addition, the health indicators proposed in this invention are applicable to various condition monitoring scenarios and can provide high-quality health trend inputs for subsequent intelligent diagnosis or remaining life prediction models, significantly improving the accuracy of fault prediction and system reliability. Attached Figure Description

[0015] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0016] Figure 1 A flowchart illustrating a method for predicting the life of a rolling bearing provided by the present invention; Figure 2 A bearing test bench for a method of predicting the life of rolling bearings provided by the present invention; Figure 3 Vibration signals from experimental data for a rolling bearing life prediction method provided by this invention; Figure 4 The life prediction result diagram is provided by the life prediction method for rolling bearings according to the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments in the specification without creative effort are within the scope of protection of this application.

[0018] As a critical component in rotating machinery, the operating condition of rolling bearings directly affects the reliability and safety performance of the equipment. Accurate monitoring of bearing health and prediction of remaining service life (RUL) are of great significance, providing a basis for maintenance planning decisions.

[0019] Although various methods exist for predicting bearing life based on vibration signals, such as wavelet packet decomposition, empirical mode decomposition (EMD), time-frequency domain analysis, and spectral analysis, these methods still have limitations when dealing with the nonlinear and non-stationary characteristics of signals. Furthermore, while deep learning methods such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs) excel in feature extraction and pattern recognition, their model training relies on large amounts of labeled data and lacks good interpretability. Current research and engineering practice often employ time-domain statistical features and entropy-based features based on signal complexity to construct bearing health indicators. For example, time-domain indicators such as root mean square (RMS), peak value, and kurtosis can intuitively reflect the amplitude distribution and impact components of vibration signals; entropy indicators such as sample entropy, permutation entropy, and dispersion entropy are used to characterize the nonlinear complexity of signals. The health indicators extracted using DE as the core not only improve the convergence efficiency of the prediction model but also significantly enhance its adaptability to different degradation stages, providing a solid foundation for accurate remaining life prediction of rolling bearings. With the acceleration of industrial intelligence, the application value of such health indicators, which integrate trend perception and amplitude perception, will become increasingly significant in practical engineering.

[0020] This invention addresses the problem that dispersion entropy (DE) relies solely on discrete symbol information and is insensitive to amplitude changes in the early stages of failure. It proposes a method for identifying rolling bearing failure trends and assessing lifespan by fusing dispersion entropy (MDE) with amplitude information and multidimensional health indicators. By incorporating amplitude information to enhance the modeling of signal dynamic behavior and combining it with a multidimensional health indicator fusion strategy, the method significantly improves the sensitivity and accuracy of characterizing the deterioration process of rolling bearings, providing higher-quality health indicators for remaining life prediction.

[0021] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.

[0022] Figure 1 This is a schematic diagram of a method for predicting the life of a rolling bearing according to the present invention, which specifically includes the following steps: S101: Acquire the vibration signal of the rolling bearing during operation.

[0023] Optionally, acquiring the vibration signal of the rolling bearing during operation specifically includes: installing the rolling bearing system onto an experimental platform and configuring a rolling bearing vibration signal data acquisition system; setting rolling bearing health indicators and simulating the initial working state of the experimental platform, and ensuring that the rolling bearing system operates at room temperature; the health state is used to characterize the degree of degradation of the rolling bearing during operation; using the rolling bearing vibration signal data acquisition system, continuously monitoring the rolling bearing from its initial new state, and continuously acquiring the original vibration signal of the rolling bearing throughout its entire life cycle until failure; denoising and standardizing the original vibration signal to obtain the vibration signal of the rolling bearing during operation.

[0024] Specifically, see Figure 2 To establish an accelerated testing platform for the entire lifespan of rolling bearings, the rolling bearing system was debugged; the initial working state of the mechanical fault simulation platform was set to ensure that the bearing system operated at room temperature (20.0℃ in this implementation); the wheel-driven platform was operated stably at a specified speed, and the bearing was monitored online using only vibration sensors; when the characteristic value of the vibration signal reached a preset threshold or the running time reached a preset condition, vibration signal data acquisition began; the acquisition ended after the data acquisition reached a predetermined sequence length or sample number, thus obtaining the vibration time-domain signal within the bearing's lifespan.

[0025] S102: Determine the spread entropy of the rolling bearing vibration signal based on the vibration signal.

[0026] Optionally, the dispersion entropy of the gearbox rolling bearing vibration signal is determined based on the vibration signal, specifically including: performing standard normalization and cumulative distribution function normalization on the vibration signal sequentially to obtain a normalized vibration signal; discretizing the normalized vibration signal according to a preset number of symbol categories to generate a vibration signal symbol sequence; constructing an embedding vector of the vibration signal symbol sequence based on the embedding dimension and time delay parameter of the vibration signal symbol sequence, and calculating the dispersion entropy of the rolling bearing vibration signal based on the probability distribution of the embedding vector.

[0027] The normalization process for the vibration signal specifically includes: performing standard normalization on the vibration signal to obtain the mean and standard deviation of the vibration signal; and, based on the mean and standard deviation, normalizing the vibration signal to the interval [0,1] using a cumulative distribution function to obtain the normalized vibration signal, as shown in the formula: ; In the formula, For normalized vibration signals, The cumulative distribution function is... It is a vibration signal. and These are the mean and standard deviation, respectively.

[0028] Specifically, the formula for calculating the vibration signal symbol sequence is as follows: ; In the formula, It is a sequence of vibration signal symbols. The preset number of symbol categories, For normalized vibration signals, This is a function for generating random numbers.

[0029] Specifically, the expression for the embedding vector of the vibration signal symbol sequence is: ; In the formula, For the embedding vector of the symbol sequence, Let the embedding dimension be the number of symbols in the sequence. This is the time delay parameter for the symbol sequence.

[0030] Specifically, determining the spread entropy of the rolling bearing vibration signal based on the embedding vector includes: Discretize the embedding vector of the symbol sequence, and based on the discretized embedding vector, convert the embedding vector into a unique index pattern, as shown in the formula: ; In the formula, A unique index for each embedded vector. It is the discretized embedding vector. It is the embedding dimension. This is the preset number of symbol categories; The expression for the embedding vector of the vibration signal symbol sequence is: ; In the formula, For the embedding vector of the symbol sequence, Let the embedding dimension be the number of symbols in the sequence. The time delay of the symbol sequence; Count the number of times the unique index of each embedding vector appears in the preset number of symbol categories. Calculate the probability of each embedding pattern. : ; In the formula, It is the total number of embedded vectors. It is the first The probability of the occurrence of a unique index of an embedded vector; The scatter entropy is calculated based on the probability of occurrence of the unique index of the embedded vector, using the following formula: ; In the formula, DE is the spread entropy. Let c be the probability of each embedding pattern occurring, and c be the preset number of symbol categories. Let be the embedding dimension of the symbol sequence.

[0031] S103: Add the vibration signal to a preset constant and take the reciprocal to construct an embedding matrix for the rolling bearing vibration signal; determine the norm of each embedding vector in the embedding matrix, and normalize the norm to a normal value using a nonlinear mapping function. The interval is used to obtain the normalized modulus values ​​of all embedded vectors; based on the preset number of amplitude partitions, the normalized modulus values ​​are then... The interval is divided into several amplitude intervals at equal intervals; the normalized modulus value of each embedded vector is assigned to the corresponding amplitude interval to obtain the amplitude mode of the amplitude interval to which each embedded vector belongs; the probability distribution of the amplitude mode of the amplitude interval to which each embedded vector belongs is statistically analyzed, and the amplitude entropy of the rolling bearing vibration signal is determined based on the probability distribution of the amplitude mode of each embedded vector.

[0032] Specifically, based on the vibration signal, an embedding matrix is ​​constructed and the norm of each embedding vector in the embedding matrix is ​​calculated. This includes adding the vibration signal to a preset constant and taking the reciprocal to construct the vibration signal embedding matrix, as shown in the formula: ; In the formula, For the vibration signal embedding matrix, This is a preset constant value. This is the time delay parameter for the symbol sequence.

[0033] Calculate the norm of each column of the embedding vector in the embedding matrix and normalize it, using the following formula: ; In the formula, The norm of the normalized embedding vector. represents the norm of each column of the embedding vector in the embedding matrix.

[0034] Optionally, based on the vibration signal, the amplitude mode probability distribution of the amplitude interval to which each embedded vector belongs is statistically analyzed, and the amplitude entropy of the rolling bearing vibration signal is determined based on the amplitude mode probability distribution of each embedded vector, specifically including: Count the number of times the magnitude pattern appears in all embedded vectors. And calculate the probability of each pattern occurring. : ; In the formula, This represents the total number of all embedding vectors in the rolling bearing vibration signal embedding matrix. For the first The probability of each amplitude pattern; the probability distribution formed by the probabilities of amplitude patterns characterizes the energy distribution features of the vibration signal at the amplitude variation level; The amplitude entropy of the rolling bearing vibration signal is calculated based on the amplitude pattern probability distribution of each embedded vector, using the following formula: ; In the formula, The amplitude entropy of the rolling bearing vibration signal. This represents the number of partitions in the amplitude mode. This is the preset number of amplitude zones.

[0035] Specifically, firstly, the norm of each embedding vector in the embedding matrix constructed from the vibration signals is calculated. Subsequently, the norm value is normalized to the interval using a nonlinear mapping function. After obtaining the normalized modulus values ​​of all embedded vectors, the magnitude is partitioned according to a preset number of amplitude partitions. The entire interval is divided into several amplitude intervals at equal intervals. The modulus value of each embedding vector will be assigned to a corresponding amplitude range, thus forming the "amplitude pattern" to which the embedding vector belongs. Specifically, assuming the modulus value... Belongs to the interval Then the embedding vector will be assigned to the amplitude mode. :

[0036] ; Amplitude patterns specifically include all amplitude partitions. A defined set of interval patterns. If the amplitude interval is divided into... If there are, then there exists. There are several amplitude patterns, each representing a different amplitude range where the magnitude value of the embedded vector falls. Next, the frequency of each amplitude pattern in all embedded vectors is counted. And calculate the probability of each pattern occurring.

[0037] Optionally, the formula for calculating the amplitude entropy of the rolling bearing vibration signal is: ; In the formula, The amplitude entropy of the rolling bearing vibration signal. This represents the number of partitions in the amplitude mode. This is the preset number of amplitude zones.

[0038] S104: Based on the dispersion entropy and amplitude entropy, determine the health status index of the rolling bearing within a specified time period, and form the degradation curve of the rolling bearing.

[0039] Optionally, the formula for calculating the health status index is: ; In the formula, MDE is the health status index of the rolling bearing vibration signal, and DE is the dispersion entropy of the rolling bearing vibration signal. The amplitude entropy of the rolling bearing vibration signal. This serves as the benchmark for the maximum entropy normalization of DE.

[0040] This embodiment uses bearing vibration degradation data collected from a bearing accelerated life test platform as experimental data to further verify and explain the effectiveness of the method proposed in this invention.

[0041] Specifically, the bearing parameters are shown in Table 1. Operating condition 2-2 (Bearing 2_2) was selected as the experimental sample, and the vibration signal is as follows: Figure 3 As shown in the figure. Under this operating condition, the bearing speed is 2250 r / min and the radial load is 11 kN. The signal sampling frequency is 25.6 kHz, the sampling period is 1 min, the sampling duration is 1.28 s, and the corresponding number of sampling points is 32768. See Table 2 for details of the vibration signal sampling settings.

[0042] Table 1 LDK UER204 Bearing Parameters Table 2 Bearing2_2 Information Table Experimental Methods: Based on the original vibration signals of rolling bearings collected under operating condition 2-2, health status indicators were extracted at different time stages of bearing operation to characterize the degradation evolution process. The Magnitude-informed Dispersion Entropy (MDE) was primarily used to analyze the vibration signals, comprehensively depicting the evolution of the temporal structure complexity and amplitude variation characteristics of the vibration signals during the degradation process. To evaluate the effectiveness of the MDE index in characterizing the degradation trend of rolling bearings, Root Mean Square (RMS), Kurtosis, and Dispersion Entropy (DE) were selected as comparative indicators. RMS and Kurtosis are commonly used time-domain statistical features, mainly reflecting the overall energy level and impulse characteristics of the vibration signal; Dispersion Entropy (DE) is used to describe the temporal structure complexity of the vibration signal. The sensitivity and stability of the MDE index in characterizing the bearing degradation process were verified through comparative analysis of the changing trends of different health indicators over operating time.

[0043] Experimental Results: The trends of the above four health indicators over time were plotted, and the results are as follows: Figure 4 As shown. From Figure 4 It can be clearly observed that as bearing service time increases, the MDE index exhibits a more stable and monotonous degradation trend, making it more sensitive to bearing failure. In contrast, the traditional RMS and Kurtosis show greater volatility, while the DE index's trend is not significant enough and is easily affected by noise.

[0044] Conclusion: This experiment verifies the effectiveness and superiority of the MDE method proposed in this patent in reflecting bearing degradation trends. MDE can accurately reflect the degradation process of a bearing from normal to faulty state by combining amplitude variation information in vibration signals, providing a reliable basis for subsequent remaining life prediction and health status assessment. This demonstrates the applicability and practical value of this patented method in actual engineering life prediction scenarios.

[0045] In summary, this invention proposes a rolling bearing life prediction method based on amplitude spread entropy. It introduces a joint modeling mechanism of vibration signal amplitude information on the basis of spread entropy to enhance the sensitivity and stability of health status indicators to degradation trends. This method constructs a spread pattern distribution while performing adaptive amplitude transformation and phase space reconstruction on the vibration signal. By calculating the L2 norm of the reconstructed vector amplitude information and combining it with a nonlinear mapping method, it achieves a smooth expression of amplitude characteristics, effectively integrating the gradually increasing amplitude changes in the vibration signal with degradation into the health status assessment process. Compared to the spread entropy method, amplitude spread entropy not only reflects the changes in the structural complexity of the vibration signal but also characterizes the overall evolution law of vibration amplitude. While suppressing random noise interference, it significantly improves the continuity and monotonicity of health status indicators throughout the entire life cycle, providing a more stable and reliable feature basis for predicting the remaining service life of rolling bearings.

[0046] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this invention.

Claims

1. A method for predicting the life of a rolling bearing, characterized in that, include: Acquire vibration signals of rolling bearings during operation; Determine the spread entropy of the rolling bearing vibration signal based on the vibration signal. The vibration signal is added to a preset constant, and the reciprocal is taken to construct an embedding matrix for the rolling bearing vibration signal. The norm of each embedding vector in the embedding matrix is ​​determined, and the norm is normalized to a normal value using a nonlinear mapping function. The interval is used to obtain the normalized modulus values ​​of all embedded vectors; Based on the preset number of amplitude zones, The interval is divided into several amplitude intervals at equal intervals; The normalized modulus value of each embedded vector is assigned to the corresponding amplitude interval to obtain the amplitude mode of the amplitude interval to which each embedded vector belongs; the probability distribution of the amplitude mode of the amplitude interval to which each embedded vector belongs is statistically analyzed, and the amplitude entropy of the rolling bearing vibration signal is determined based on the probability distribution of the amplitude mode of each embedded vector. Based on the dispersion entropy and amplitude entropy, the health status index of the rolling bearing within a specified time period is determined, and the degradation curve of the rolling bearing is generated.

2. The method for predicting the life of a rolling bearing as described in claim 1, characterized in that, The acquisition of vibration signals of the rolling bearing during operation specifically includes: Install the rolling bearing system onto the experimental platform and configure a rolling bearing vibration signal data acquisition system; The rolling bearing health indicators were set and the initial working state of the experimental platform was simulated, and the rolling bearing system was ensured to operate at room temperature; the health indicators were used to characterize the degree of degradation of the rolling bearing during operation. Using a rolling bearing vibration signal data acquisition system, the rolling bearing is continuously monitored from its initial brand-new state, and the original vibration signals of the rolling bearing are continuously collected throughout its entire life cycle from operation to failure. The original vibration signal is denoised and standardized to obtain the vibration signal of the rolling bearing during operation.

3. The method for predicting the life of a gearbox rolling bearing as described in claim 1, characterized in that, The determination of the dispersion entropy of the gearbox rolling bearing vibration signal based on the vibration signal specifically includes: The vibration signal is then subjected to standard normalization and cumulative distribution function normalization in sequence to obtain the normalized vibration signal. The normalized vibration signal is discretized according to a preset number of symbol categories to generate a vibration signal symbol sequence. Based on the embedding dimension and time delay parameters of the vibration signal symbol sequence, an embedding vector of the vibration signal symbol sequence is constructed, and the scattering entropy of the rolling bearing vibration signal is calculated based on the probability distribution of the embedding vector.

4. The method for predicting the life of a rolling bearing as described in claim 3, characterized in that, The process of sequentially performing standard normalization and cumulative distribution function normalization on the vibration signal to obtain a normalized vibration signal specifically includes: The vibration signal is normalized to obtain the mean and standard deviation of the vibration signal. Based on the mean and standard deviation, the vibration signal is normalized to the interval [0,1] using the cumulative distribution function, resulting in the normalized vibration signal, as shown in the formula: ； In the formula, For normalized vibration signals, The cumulative distribution function is... It is a vibration signal. and These are the mean and standard deviation, respectively.

5. The method for predicting the life of a rolling bearing as described in claim 3, characterized in that, The formula for calculating the vibration signal symbol sequence is: ； In the formula, It is a sequence of vibration signal symbols. The preset number of symbol categories, For normalized vibration signals, This is a function for generating random numbers.

6. The method for predicting the life of a rolling bearing as described in claim 3, characterized in that, Based on the embedding vector, the dispersion entropy of the rolling bearing vibration signal is determined, specifically including: Discretize the embedding vector of the symbol sequence, and based on the discretized embedding vector, convert the embedding vector into a unique index pattern, as shown in the formula: ； In the formula, A unique index for each embedded vector. It is the discretized embedding vector. It is the embedding dimension. This is the preset number of symbol categories; The expression for the embedding vector of the vibration signal symbol sequence is: ； In the formula, For the embedding vector of the symbol sequence, Let the embedding dimension be the number of symbols in the sequence. The time delay of the symbol sequence; Count the number of times the unique index of each embedding vector appears in the preset number of symbol categories. Calculate the probability of each embedding pattern. : ； In the formula, It is the total number of embedded vectors. It is the first The probability of the occurrence of a unique index of an embedded vector; The scatter entropy is calculated based on the probability of occurrence of the unique index of the embedded vector, using the following formula: ； In the formula, DE is the spread entropy. Let c be the probability of each embedding pattern occurring, and c be the preset number of symbol categories. Let be the embedding dimension of the symbol sequence.

7. The method for predicting the life of a rolling bearing as described in claim 1, characterized in that, The step of statistically analyzing the amplitude mode probability distribution of each embedded vector within its amplitude range based on the vibration signal, and determining the amplitude entropy of the rolling bearing vibration signal based on the amplitude mode probability distribution of each embedded vector, specifically includes: Count the number of times the magnitude pattern appears in all embedded vectors. And calculate the probability of each pattern occurring. : ； In the formula, This represents the total number of all embedding vectors in the rolling bearing vibration signal embedding matrix. For the first The probability of each amplitude pattern; the probability distribution formed by the probabilities of amplitude patterns characterizes the energy distribution features of the vibration signal at the amplitude variation level; The amplitude entropy of the rolling bearing vibration signal is calculated based on the amplitude pattern probability distribution of each embedded vector, using the following formula: ； In the formula, The amplitude entropy of the rolling bearing vibration signal. This represents the number of partitions in the amplitude mode. This is the preset number of amplitude zones.

8. The method for predicting the life of a rolling bearing as described in claim 1, characterized in that, The formula for calculating the health status index is as follows: ； In the formula, MDE is the health status index of the rolling bearing vibration signal, and DE is the dispersion entropy of the rolling bearing vibration signal. The amplitude entropy of the rolling bearing vibration signal. This serves as the benchmark for the maximum entropy normalization of DE.

Images