A method and system for predicting bearing degradation using a deep latent variable state-space model

Abstract

This invention discloses a method and system for predicting bearing degradation using a deep latent variable state-space model. The method uses the Box-Cox transformation function to correct simulated and real bearing degradation data, fusing multiple time-domain degradation features to obtain degradation health indicators. A degradation state characterizing the bearing degradation rate is obtained through differential transformation. Within the state-space model framework, a prediction model based on a deep latent variable state-space model is constructed. Features are extracted, fused, and differencing are performed on the simulated and real data to obtain their respective degradation states. The deep latent variable state-space model is pre-trained using the degradation states obtained from the simulated degradation data, and the model weights are initialized. Then, the prediction model based on the deep latent variable state-space model is used to predict the degradation states obtained from the real data. The predicted degradation values ​​are obtained by accumulating the predicted degradation states. This method solves the problems of existing processes being complex and time-consuming.

Description

Technical Field

[0001] This invention belongs to the field of bearing degradation prediction technology, specifically relating to a bearing degradation prediction method and system using a deep latent variable state-space model. Background Technology

[0002] Bearings, due to their unique function, play a crucial role in maintaining the normal working position and rotational accuracy of shafts, and are considered one of the most important components of rotating machinery. They are widely used in many fields such as machining, transportation, hydropower generation, and aerospace engineering. However, the operating environment of bearings is very complex, affected by various factors such as temperature, humidity, and load variations, making them prone to failure. Timely and reasonable predictive maintenance is essential for the normal operation of industrial production. Therefore, accurately predicting the degree of bearing degradation and carrying out maintenance and inspection at the appropriate time is crucial for avoiding business downtime and reducing personnel casualties.

[0003] Currently, traditional methods for predicting bearing degradation mainly rely on physical and statistical models. Physical model-based methods are complex and require significant time and resources. Statistical model-based methods mostly depend on finite-form nonlinear assumptions such as power laws and exponential models, and are often limited to the measurement equations of time-series models. This significantly limits their ability to capture the strong nonlinear behavior and long-term dependencies in bearing degradation data.

[0004] Therefore, researchers began exploring deep learning-based methods to enhance the nonlinear modeling capabilities and adaptability of models. While deep learning models have achieved some success in time series forecasting, many challenges remain in bearing degradation prediction. For example, deep learning models cannot provide probabilistic prediction capabilities like statistical models; they heavily rely on large-scale labeled state data, exhibiting weak learning and generalization abilities with small sample sizes; and they face difficulties in modeling skewed distributions and non-constant variance phenomena in degradation data. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method and system for predicting bearing degradation using a deep latent variable state space model, which addresses the shortcomings of the prior art and solves the technical problem that the existing methods are complex and require a lot of time and resources.

[0006] The present invention adopts the following technical solution:

[0007] A method for predicting bearing degradation using a deep latent variable state-space model includes the following steps:

[0008] Based on the bearing parameters and corresponding operating conditions predicted by the requirements, a second-order pseudo-cyclic steady bearing fault simulation model based on analytical formulas is established to obtain the bearing simulation degradation data.

[0009] The simulated degradation data and the real degradation data are corrected by using the data transformation function Box-Cox. Multiple time-domain degradation features are fused using the Mahalanobis distance fusion method and the 3-sigma stage division method to obtain degradation health indicators. Then, the degradation state characterizing the bearing degradation rate is obtained through differential transformation.

[0010] Within the state-space model framework, a recurrent neural network is used as the state transition equation, and a variational autoencoder is combined as the degenerate observation equation to construct a prediction model based on a deep latent variable state-space model.

[0011] Feature extraction, fusion, and differencing are performed on simulated degradation data and real data to obtain their degradation states. A pre-trained deep latent variable state space model is obtained using the simulated degradation data, and the weights of the state space model are initialized. Then, a prediction model based on the deep latent variable state space model is used to predict the degradation states obtained from the real data. By accumulating the predicted degradation states, the bearing degradation prediction value is obtained.

[0012] Preferably, the second-order pseudo-cyclic steady bearing fault simulation model based on the analytical formula is as follows:

[0013]

[0014] in, T f A represents the time when a local bearing defect occurs. i The amplitude modulation gradually increases during the attenuation process, which is derived from the Paris formula, B. i For the amplitude distribution of transient pulses that follows a normal distribution, r is the pulse attenuation coefficient. j The resonant frequency of the rotating system For the time of transfer, The resonant frequency, Rotating harmonics under variable speed conditions The harmonic order of the rotating harmonic. Let be the occurrence time of the i-th pulse. The amplitude of the rotating harmonic. It is Gaussian noise. This represents the final vibration response of the system during the defect stage.

[0015] Preferably, the vibration model under the healthy phase is as follows:

[0016] in, The resonant frequency, Rotating harmonics under variable speed conditions The harmonic order of the rotating harmonic. Let be the occurrence time of the i-th pulse. The amplitude of the rotating harmonic. It is Gaussian noise.

[0017] Preferably, the obtained simulated degradation data and the real degradation data are corrected using the Box-Cox transformation function, as follows:

[0018]

[0019] Where y is the value after the Box-Cox transformation; To adjust the parameters.

[0020] Preferably, the various time-domain degradation characteristics include: indicators sensitive to overall damage, including the root mean square reflecting signal energy; indicators sensitive to local damage, including kurtosis reflecting the degree of impact; and dimensionless indicators independent of operating conditions and sensitive to damage and faults, including waveform factor and crest factor.

[0021] Preferably, the prediction model based on the deep latent variable state-space model is as follows:

[0022]

[0023] Model Priors:

[0024]

[0025]

[0026] Model posterior:

[0027]

[0028] in, Let be the joint distribution of the model, representing the entire process from time 1 to time T in which the model generates predicted values. These are the predicted degradation values ​​from time 1 to time T. Let these be the latent variables from time 1 to time T. Let T be the intermediate variable in the transition distribution from time 1 to time T. These are the initial intermediate variables for the recurrent neural network. The input degradation state is from time 1 to time T. The emission distribution represents the process of generating predicted values ​​from latent variables. This provides the prior distribution and the initial hypothetical distribution for the latent variables. The transition distribution represents the prior distribution of states at each time step given the previous state and the current input of the model, learning the temporal correlation in the bearing degradation data.

[0029] Preferably, the degradation observation equation is:

[0030]

[0031]

[0032] Where z is a latent variable and y is the value after the Box-Cox transformation. For the weights and biases of the neural network, It follows a Gaussian distribution. for t Latent variables of time step, This represents the process of generating the prior mean and prior variance, from... Obtained through neural network fitting. For neural network parameters, These are predicted values.

[0033] Secondly, embodiments of the present invention provide a bearing degradation prediction system based on a deep latent variable state-space model, comprising:

[0034] The data module, based on the required predicted bearing parameters and corresponding operating conditions, establishes a second-order pseudo-cyclic steady bearing fault simulation model based on analytical formulas to obtain the bearing's simulation degradation data.

[0035] The correction module uses the Box-Cox data transformation function to correct the obtained simulated degradation data and real degradation data. It uses the Mahalanobis distance fusion method and the 3-sigma stage division method to fuse multiple time-domain degradation features to obtain degradation health indicators. Then, it uses differential transformation to obtain the degradation state characterizing the bearing degradation rate.

[0036] The equation module, within the state-space model framework, uses recurrent neural networks as state transition equations and combines them with variational autoencoders as degenerate observation equations to construct a prediction model based on a deep latent variable state-space model.

[0037] The prediction module extracts, fuses, and differs features from simulated degradation data and real data to obtain their degradation states. It uses the simulated degradation data to obtain a pre-trained deep latent variable state space model of the degradation state, initializes the weights of the state space model, and then uses the prediction model based on the deep latent variable state space model to predict the degradation state obtained from the real data. By accumulating the predicted degradation states, the predicted value of bearing degradation is obtained.

[0038] Thirdly, a computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described deep latent variable state-space model bearing degradation prediction method.

[0039] Fourthly, embodiments of the present invention provide a computer-readable storage medium including a computer program, which, when executed by a processor, implements the steps of the above-described deep latent variable state-space model bearing degradation prediction method.

[0040] Fifthly, a chip includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described deep latent variable state-space model bearing degradation prediction method.

[0041] In a sixth aspect, embodiments of the present invention provide an electronic device, including a computer program, wherein when the computer program is executed by the electronic device, it implements the steps of the above-described deep latent variable state space model bearing degradation prediction method.

[0042] Compared with the prior art, the present invention has at least the following beneficial effects:

[0043] A deep latent variable state-space model method for bearing degradation prediction is proposed. This method leverages the nonlinear mapping capabilities of deep neural networks and combines them with the uncertainty structure modeling features provided by state-space models, improving nonlinear modeling capabilities under complex working conditions. It alleviates the problem of data scarcity during degradation prediction and introduces the physical mechanism of bearing degradation during pre-training to help the model better understand degradation patterns, accelerating the model training process. It uses knowledge provided by a fault model based on analytical formulas as prior knowledge to guide the prediction model in learning system characteristics and to more accurately capture system state changes when predicting degradation behavior, further alleviating data scarcity, accelerating model training, and improving prediction accuracy. By performing Box-Cox transformation on the original data, data skewness and non-normality can be eliminated, and the degradation rate prediction after difference transformation effectively improves the model's long-term predictive ability. It is suitable for predicting the degradation trend of mechanical parts, especially bearings, providing a promising solution for industrial applications. The degradation trend prediction accuracy is high, even for mechanical parts under complex working conditions.

[0044] Furthermore, in predicting bearing degradation trends, the collected experimental data may contain environmental noise, measurement errors, and minor defects. In addition, deep learning models with numerous neural network layers typically require a large amount of data for training. However, real-world industrial datasets containing bearing degradation states may not provide a sufficient number of samples for training comprehensive models. To address this issue, this study analyzes the vibration signal characteristics of bearings in both healthy and faulty states, and investigates a second-order pseudo-cyclic stationary bearing fault simulation model based on analytical formulas, generating mechanistic degradation data encompassing both healthy and faulty states. Pre-training using simulation data fully leverages the prior knowledge of the mechanistic data, allowing the parameters of the deep latent variable state-space model to acquire bearing degradation characteristics in advance, accelerating model convergence and reducing dependence on real data. This method alleviates the problem of data scarcity and improves prediction accuracy.

[0045] Furthermore, the Box-Cox transform is a method that adjusts the distribution of data through power function transformation. For degenerate data, the difference in distribution between simulated and real data may stem from inconsistencies in noise levels, skewness, or dynamic range. The Box-Cox transform can make the statistical characteristics of the two types of data more similar without altering the core trends, thereby enhancing the representativeness and adaptability of simulated data in model training. This method further optimizes the pre-training effect of the model, enabling the predictive model to more accurately capture degenerate features.

[0046] Furthermore, to accurately depict the "health-degradation" stage and minimize interference, the raw data needs to be processed. Currently, various degradation characteristic indicators have been proposed in the field of bearing fault diagnosis, all of which can reflect bearing degradation under a specific mode. However, considering that the characteristic parameters and characteristic frequencies of each component of the tested bearing are unknown, and that damage to each component may occur in combination, time-domain statistical features are chosen as the input features for the extraction method. Time-domain statistical indicators can effectively reduce the noise present in the raw signal, and many indicators themselves have clear physical meaning. We selected 14 indicators to assess the health status of the bearing, including indicators sensitive to overall damage, such as the root mean square reflecting signal energy, and indicators sensitive to local damage, such as kurtosis reflecting the degree of impact. We also selected dimensionless indicators that are independent of operating conditions and sensitive to damage and faults, such as waveform factor and crest factor, to comprehensively reflect the bearing degradation caused by various forms of faults.

[0047] It is understood that the beneficial effects of the second to sixth aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here.

[0048] In summary, this invention constructs a bearing degradation prediction model based on a deep latent variable state-space model, which not only significantly improves nonlinear modeling capabilities but also incorporates the uncertainty structure modeling provided by the SSM. Furthermore, addressing the issue of insufficient sample size in industrial datasets, a second-order pseudo-cyclic stationary bearing fault simulation model based on analytical formulas provides mechanistic degradation information, alleviating the sample scarcity problem. The introduction of the Box-Cox transform reduces the difference between simulation and original data, improving the model's prediction accuracy and robustness, and enhancing the reliability of life prediction. The use of difference transform to obtain the degradation state characterizing the bearing degradation rate, and subsequent model training and prediction, effectively improves the model's long-term predictive ability.

[0049] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of the process of the present invention;

[0051] Figure 2 This is a framework diagram of the deep latent variable state space model of the present invention;

[0052] Figure 3 This is a graphical model diagram of the deep latent variable state space model of the present invention;

[0053] Figure 4 This is a framework diagram of the bearing failure model of the present invention;

[0054] Figure 5 This is a simulation degradation data diagram of the bearing fault model of the present invention;

[0055] Figure 6 shows the prediction results of the PHM2012 bearing dataset in this invention, where (a) is Bearing1_1 and (b) is Bearing1_3;

[0056] Figure 7 A schematic diagram of a computer device provided in an embodiment of the present invention;

[0057] Figure 8 This is a block diagram of a chip provided according to an embodiment of the present invention. Detailed Implementation

[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] In the description of this invention, it should be understood that the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0060] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0061] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Additionally, the character " / " in this invention generally indicates that the preceding and following objects have an "or" relationship.

[0062] It should be understood that although terms such as first, second, third, etc., may be used in the embodiments of the present invention to describe the preset range, these preset ranges should not be limited to these terms. These terms are only used to distinguish the preset ranges from one another. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0063] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0064] The accompanying drawings illustrate various structural schematic diagrams according to embodiments disclosed in this invention. These drawings are not to scale, and some details have been enlarged for clarity, and some details may have been omitted. The shapes of the various regions and layers shown in the drawings, as well as their relative sizes and positional relationships, are merely exemplary and may deviate from reality due to manufacturing tolerances or technical limitations. Furthermore, those skilled in the art can design regions / layers with different shapes, sizes, and relative positions as needed.

[0065] This invention provides a deep latent variable state-space model method for predicting bearing degradation. First, within the state-space model framework, a recurrent neural network is used as the state transition equation, combined with a variational autoencoder as the degradation observation equation, to construct a deep latent variable state-space model. Second, based on the bearing's intrinsic parameters and corresponding operating conditions, a second-order pseudo-cyclic stationary bearing fault simulation model is established to obtain bearing degradation data. Box-Cox transformation is used to correct the simulated degradation data and the actual degradation data. A Mahalanobis distance fusion method and a 3-sigma stage partitioning method are used to fuse multiple features to obtain health indicators. Then, a difference transformation is used to obtain the degradation state characterizing the bearing degradation rate. The model is pre-trained using data obtained from simulated degradation data to initialize weights and capture key features of the degradation data. Data obtained from actual bearing degradation data is used for further training and prediction of the model. Compared with existing methods, this invention can use less historical data, significantly enhances nonlinear modeling capabilities, and achieves higher accuracy and stronger adaptability in prediction.

[0066] Example 1

[0067] Please see Figure 1 This invention discloses a bearing degradation prediction method using a deep latent variable state-space model, comprising the following steps:

[0068] S1. Bearing degradation data preprocessing;

[0069] The collected bearing vibration signals were transformed using the Box-Cox data transformation function to eliminate data skewness and non-normality. Multiple time-domain degradation features were selected and fused using the Mahalanobis distance fusion method and the 3-sigma stage division method. The fused data was smoothed to generate clear indicators representing the health and degradation stages. The health indicators were processed using differential transformation to obtain the degradation state characterizing the bearing degradation rate. The specific steps are as follows:

[0070] S101. For the collected bearing vibration signal, use the data transformation function Box-Cox to transform and process it to eliminate the skewness and non-normality of the data;

[0071] The transformation process is as follows:

[0072]

[0073] Where y is the value after the Box-Cox transformation; To adjust the parameters, adjust The value makes the Box-Cox transform into a logarithmic and power-law transform.

[0074] S102. Select multiple temporal degradation features;

[0075] Indicators sensitive to overall damage, such as the root mean square (RMS) reflecting signal energy, and indicators sensitive to local damage, such as kurtosis reflecting the degree of impact, are also selected. Dimensionless indicators independent of operating conditions and sensitive to damage and faults, such as waveform factor and crest factor, are also chosen.

[0076] S103. Obtain degenerative health indicators using Mahalanobis distance fusion method;

[0077] The phase division adopts the 3-sigma standard. Finally, the fused data obtained using Mahalanobis distance is smoothed to produce a clear "health degradation" indicator.

[0078] S104. Use differential transformation to process health indicators to obtain the degradation state characterizing the bearing degradation rate.

[0079] S2. Construct a prediction model based on a deep latent variable state space model;

[0080] Within the framework of traditional statistical learning state-space models, a variational autoencoder is used to construct the observation equations for the state-space model. A recurrent neural network is used to temporally extend the latent variables of the variational autoencoder, and the prior parameters of the variational autoencoder are updated sequentially based on the output of the recurrent neural network. This allows the learning of the temporal correlations between bearing degradation data, thus constructing a degradation model based on a deep latent variable state-space model.

[0081] Within the framework of traditional statistical learning state-space models, the observation equations of the state-space model are constructed using variational autoencoders;

[0082] The degradation observation equation is:

[0083]

[0084]

[0085] in, z These are latent variables, and they are relatively independent in the VAE. The predicted value is calculated using a neural network through the latent variable z. These are the weights and biases of the neural network.

[0086] A recurrent neural network is used to temporally extend the latent variables of the variational autoencoder, and the prior parameters of the variational autoencoder are updated sequentially according to the output of the recurrent neural network. This allows the learning of the temporal correlations between bearing degradation patterns. (See below:)

[0087]

[0088] in, This represents the computational process of a recurrent neural network. uTo input the degradation state, h This is an intermediate layer in a recurrent neural network. z These are latent variables.

[0089] In summary, the joint distribution of the model can be expressed as:

[0090]

[0091] Prior distribution:

[0092]

[0093]

[0094] Among them, in the prior distribution Represents latent variables z Generated by a Gaussian distribution, Represents a Gaussian distribution. for t Latent variables of time step, prior Indicates prior knowledge, The prior mean, For the prior variance, This represents the process of generating the prior mean and prior variance, from... Obtained through neural network fitting. These are the parameters of the neural network.

[0095] Model posterior:

[0096]

[0097] Wherein, the posterior distribution represents the latent variables derived from the observed data. l With intermediate variables h get, h For intermediate variables in a recurrent neural network that contain past information, the above process is broken down into independent iterative processes to obtain the emission distribution at time step t. Prior distribution , transfer distribution , This indicates that the latent variables in the transition distribution are generated by a Gaussian distribution. Represents a Gaussian distribution. Let be the latent variable at time step t, and prior represent the prior. The prior mean, For the prior variance, This represents the process of generating the prior mean and prior variance, from... Obtained through neural network fitting. These are the parameters of the neural network.

[0098] S3. A pre-training method for models that integrate domain knowledge.

[0099] Based on the bearing's own parameters and corresponding operating conditions, vibration signals of the bearing in the healthy and defective stages are constructed to form a second-order pseudo-cyclic steady bearing fault simulation model based on analytical formulas under variable speed conditions, and to obtain the bearing's simulation degradation data.

[0100] S301. Construct a bearing fault model based on analytical formulas;

[0101] In the healthy phase, the subtle resonance phenomenon caused by random vibration excitation is ignored, and a vibration model containing rotating harmonics and Gaussian noise is constructed.

[0102] The vibration model is as follows:

[0103]

[0104] Wherein, the harmonic order q(n) of the rotating harmonic is approximately an integer, i.e., q(n) ≈ n.wgn(t) represents Gaussian noise, and R(t) is the rotating harmonic; C n This indicates the amplitude of the rotating harmonic.

[0105] The rotational speed R(t) under variable speed conditions is expressed as follows: From this, the instantaneous shaft phase is derived.

[0106]

[0107] During the defect stage, a vibration model is established, including rotating harmonics, repetitive transient pulses of varying intensities caused by the defect, and constant-intensity Gaussian noise. The repetitive pulse is established based on pseudo-cyclic stationarity, with the following formula:

[0108]

[0109] in, It is the slip time, which varies between 1% and 3% of the instantaneous duration of the fault pulse. Indicates the natural frequency.

[0110] The Paris formula is used to simulate the degradation process of bearings. The formula is as follows:

[0111]

[0112] The unit impulse response A i The final vibration response of the system in the defect stage is generated by convolving the time-domain product of the amplitude distribution and the pulse sequence d(t) with the vibration decay function e(t), as shown in the following formula:

[0113]

[0114] Among them, A iThis indicates that the amplitude modulation gradually increases during the attenuation process, B. i This represents the amplitude distribution of a transient pulse that follows a normal distribution. It is the pulse attenuation coefficient, r j It is the resonant frequency of the rotating system.

[0115] Based on the formulas established by integrating healthy and defective states, the analytical formula for the pseudo-cyclic steady-state fault mechanism model of bearings under variable speed conditions is derived as follows:

[0116]

[0117] in, T f Indicates the time when the local bearing defect occurred. During the bearing's healthy phase, the vibration signal only contains rotating harmonics and Gaussian noise. During the bearing failure phase, the vibration signal includes repetitive transient pulses of varying intensities caused by defects, rotating harmonics, and Gaussian noise of constant intensity; the simulation data generated by this formula, which integrates domain knowledge, can be used for model pre-training.

[0118] S302, Model pre-training integrating domain knowledge.

[0119] The proposed method of this invention was validated using data obtained from the PHM 2012 prediction challenge. The specific steps are as follows:

[0120] S3021. Collect vibration signals from the bearing;

[0121] S3022. Based on the actual bearing parameters and corresponding operating conditions given in the dataset, establish a bearing fault simulation model based on analytical formulas to obtain simulation degradation data of the rolling bearing, such as... Figure 5 As shown;

[0122] S3023. The simulation data and PHM2012 bearing degradation data are processed by Box-Cox transformation. Fourteen time-domain statistical features are selected as input features for the extraction method. The Mahalanobis distance fusion method and the 3-sigma standard stage division method are used to fuse the 14 features, and then smoothing is performed to obtain a clear "health degradation" index.

[0123] S3024. Use differential transformation to process health indicators to obtain the degradation state characterizing the bearing degradation rate.

[0124] Example 2

[0125] This invention provides a bearing degradation prediction system based on a deep latent variable state-space model. This system can be used to implement the aforementioned bearing degradation prediction method based on a deep latent variable state-space model. Specifically, the bearing degradation prediction system based on a deep latent variable state-space model includes a data module, a correction module, an equation module, and a prediction module.

[0126] The data module, based on the bearing parameters and corresponding operating conditions predicted according to the requirements, establishes a second-order pseudo-cyclic steady bearing fault simulation model based on analytical formulas to obtain the bearing simulation degradation data.

[0127] The correction module corrects the obtained simulated degradation data and real degradation data, integrates multiple time-domain degradation features to obtain degradation health indicators, and then obtains the degradation state characterizing the bearing degradation rate through differential transformation.

[0128] The equation module, within the state-space model framework, uses recurrent neural networks as state transition equations and combines them with variational autoencoders as degenerate observation equations to construct a prediction model based on a deep latent variable state-space model.

[0129] The prediction module extracts, fuses, and differs features from simulated degradation data and real data to obtain their degradation states. It uses the simulated degradation data to obtain a pre-trained deep latent variable state space model of the degradation state, initializes the weights of the state space model, and then uses the prediction model based on the deep latent variable state space model to predict the degradation state obtained from the real data. By accumulating the predicted degradation states, the predicted value of bearing degradation is obtained.

[0130] Example 3

[0131] This invention provides a terminal device comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions to achieve a corresponding method flow or corresponding function. The processor described in this embodiment can be used for the operation of a bearing degradation prediction method based on a deep latent variable state-space model, including:

[0132] Based on the required bearing parameters and corresponding operating conditions, a second-order pseudo-cyclic steady-state bearing fault simulation model based on analytical formulas is established to obtain simulated bearing degradation data. The obtained simulated degradation data and real degradation data are corrected, and multiple time-domain degradation features are fused to obtain degradation health indicators. Then, a degradation state characterizing the bearing degradation rate is obtained through difference transformation. Within the state-space model framework, a recurrent neural network is used as the state transition equation, combined with a variational autoencoder as the degradation observation equation, to construct a prediction model based on a deep latent variable state-space model. Feature extraction, fusion, and difference are performed on the simulated degradation data and real data to obtain their respective degradation states. The deep latent variable state-space model is pre-trained using the degradation states obtained from the simulated degradation data, and the space model weights are initialized. Then, the prediction model based on the deep latent variable state-space model is used to predict the degradation states obtained from the real data. By accumulating the predicted degradation states, the bearing degradation prediction value is obtained.

[0133] Please see Figure 7 The terminal device is a computer device. In this embodiment, the computer device 60 includes a processor 61, a memory 62, and a computer program 63 stored in the memory 62 and executable on the processor 61. When executed by the processor 61, the computer program 63 implements the fluid composition calculation method in the reservoir stimulation wellbore of this embodiment. To avoid repetition, these details are not elaborated here. Alternatively, when executed by the processor 61, the computer program 63 implements the functions of each model / unit in the bearing degradation prediction system of the deep latent variable state space model of this embodiment. To avoid repetition, these details are not elaborated here.

[0134] Computer device 60 can be a desktop computer, laptop, handheld computer, cloud server, or other computing device. Computer device 60 may include, but is not limited to, a processor 61 and a memory 62. Those skilled in the art will understand that... Figure 7 This is merely an example of computer device 60 and does not constitute a limitation on computer device 60. It may include more or fewer components than shown, or combine certain components, or different components. For example, computer device may also include input / output devices, network access devices, buses, etc.

[0135] The processor 61 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0136] The memory 62 can be an internal storage unit of the computer device 60, such as a hard disk or RAM of the computer device 60. The memory 62 can also be an external storage device of the computer device 60, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. equipped on the computer device 60.

[0137] Furthermore, the memory 62 may include both internal storage units of the computer device 60 and external storage devices. The memory 62 is used to store computer programs and other programs and data required by the computer device. The memory 62 can also be used to temporarily store data that has been output or will be output.

[0138] Please see Figure 8 The terminal device is an electronic device 600, which is manifested in the form of a general-purpose computing device. The components of the electronic device may include, but are not limited to: at least one processing unit 610, at least one storage unit 620, a bus 630 connecting different platform components (including storage unit 620 and processing unit 610), a display unit 640, etc.

[0139] The storage unit stores program code, which can be executed by the processing unit 610 to perform the steps described in the method section of this specification according to various exemplary embodiments of the present invention. For example, the processing unit 610 can perform actions such as... Figure 1 The steps are shown in the figure.

[0140] Storage unit 620 may include a readable medium in the form of a volatile storage unit, such as random access memory (RAM) 6201 and / or cache memory 6202, and may further include a read-only memory (ROM) 6203.

[0141] Storage unit 620 may also include a program / utility 6204 having a set (at least one) program module 6205, such program module 6205 including but not limited to: operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0142] Bus 630 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the multiple bus structures.

[0143] Electronic device 600 can also communicate with one or more external devices 700 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 600, and / or with any device that enables electronic device 600 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 650. Furthermore, electronic device 600 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 660. Network adapter 660 can communicate with other modules of electronic device 600 via bus 630. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage platforms.

[0144] Example 4

[0145] This invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a terminal device used to store programs and data. It is understood that the computer-readable storage medium here can include both built-in storage media in the terminal device and extended storage media supported by the terminal device; it can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, the storage space also stores one or more instructions suitable for loading and execution by a processor, which can be one or more computer programs (including program code). It should be noted that more specific examples (a non-exhaustive list) of the computer-readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0146] Computer-readable storage media also include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable storage medium can also be any readable medium other than a readable storage medium that can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium can be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0147] Program code for performing the operations of this invention can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0148] One or more instructions stored in a computer-readable storage medium can be loaded and executed by a processor to implement the corresponding steps of the bearing degradation prediction method of the deep latent variable state-space model in the above embodiments; one or more instructions in the computer-readable storage medium are loaded and executed by the processor in the following steps:

[0149] Based on the required bearing parameters and corresponding operating conditions, a second-order pseudo-cyclic steady-state bearing fault simulation model based on analytical formulas is established to obtain simulated bearing degradation data. The obtained simulated degradation data and real degradation data are corrected, and multiple time-domain degradation features are fused to obtain degradation health indicators. Then, a degradation state characterizing the bearing degradation rate is obtained through difference transformation. Within the state-space model framework, a recurrent neural network is used as the state transition equation, combined with a variational autoencoder as the degradation observation equation, to construct a prediction model based on a deep latent variable state-space model. Feature extraction, fusion, and difference are performed on the simulated degradation data and real data to obtain their respective degradation states. The deep latent variable state-space model is pre-trained using the degradation states obtained from the simulated degradation data, and the space model weights are initialized. Then, the prediction model based on the deep latent variable state-space model is used to predict the degradation states obtained from the real data. By accumulating the predicted degradation states, the bearing degradation prediction value is obtained.

[0150] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0151] Health indicators obtained from simulated degradation data were used to pre-train a deep latent variable state-space model, initializing the model's weights and capturing key features of the degradation data. The health indicators of the bearing degradation data in the PHM2012 dataset were divided into the first 60% of training data and the last 40% of validation data. The pre-trained model was further trained using the training data. When the training converged to a stable level, the validation model was used to predict the bearing degradation data, and the predicted degradation rates were accumulated to obtain the final bearing degradation prediction value. The prediction results for the Bearing1_1 and Bearing1_3 datasets are shown in Figure 6.

[0152] The results show that the method of the present invention has better prediction performance than the statistical model Switching SSM, the deep learning model LSTM, and the hybrid model Deepstate.

[0153] In summary, the present invention provides a deep latent variable state-space model method and system for predicting bearing degradation. This significantly improves the ability to capture the dynamic characteristics of the bearing degradation process, particularly in modeling complex nonlinear bearing degradation trends, and more accurately reflects the nonlinear behavior during degradation. Furthermore, the method possesses the ability to model uncertainties, effectively addressing model parameter uncertainties caused by random factors. By introducing fault simulation data, it overcomes the dependence of traditional deep learning models on large amounts of training sample data. Initializing model parameters with simulation data containing mechanistic information effectively improves prediction accuracy. Those skilled in the art will understand that, for ease of description and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of each functional unit and module are merely for ease of differentiation and are not intended to limit the scope of protection of this application. The specific working processes of the units and modules in the above system can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0154] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0155] 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 use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0156] In the embodiments provided by this invention, it should be understood that the disclosed devices / terminals and methods can be implemented in other ways. For example, the device / terminal embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0157] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0158] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0159] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random-access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0160] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus, and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0161] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0162] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0163] The above content is only for illustrating the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made to the technical solution based on the technical concept proposed in this invention shall fall within the scope of protection of the claims of this invention.

Claims

1. A method for predicting bearing degradation using a deep latent variable state-space model, characterized in that, Includes the following steps: Based on the bearing parameters and corresponding operating conditions predicted by the requirements, a second-order pseudo-cyclic steady bearing fault simulation model based on analytical formulas is established to obtain the bearing simulation degradation data. The obtained simulated degradation data and real degradation data are corrected, and multiple time-domain degradation features are integrated to obtain degradation health indicators. Then, the degradation state characterizing the bearing degradation rate is obtained through differential transformation. Within the state-space model framework, a recurrent neural network is used as the state transition equation, and a variational autoencoder is combined as the degenerate observation equation to construct a prediction model based on a deep latent variable state-space model. Feature extraction, fusion, and differencing are performed on simulated degradation data and real data to obtain their degradation states. A pre-trained deep latent variable state space model is obtained using the simulated degradation data, and the weights of the state space model are initialized. Then, a prediction model based on the deep latent variable state space model is used to predict the degradation states obtained from the real data. By accumulating the predicted degradation states, the bearing degradation prediction value is obtained.

2. The bearing degradation prediction method using a deep latent variable state-space model according to claim 1, characterized in that, The second-order pseudo-cyclic steady bearing fault simulation model based on analytical formulas is as follows: in, T f A represents the time when a local bearing defect occurs. i The amplitude modulation gradually increases during the attenuation process, which is based on the Paris formula, B i For the amplitude distribution of transient pulses that follows a normal distribution, r is the pulse attenuation coefficient. j The resonant frequency of the rotating system For the time of transfer, The resonant frequency, Rotating harmonics under variable speed conditions The harmonic order of the rotating harmonic. Let be the occurrence time of the i-th pulse. The amplitude of the rotating harmonic. It is Gaussian noise. This represents the final vibration response of the system during the defect stage.

3. The bearing degradation prediction method using a deep latent variable state-space model according to claim 1, characterized in that, The vibration model under the healthy phase is as follows: in, The resonant frequency, Rotating harmonics under variable speed conditions The harmonic order of the rotating harmonic. Let be the occurrence time of the i-th pulse. The amplitude of the rotating harmonic. It is Gaussian noise.

4. The bearing degradation prediction method using a deep latent variable state-space model according to claim 1, characterized in that, The Box-Cox transformation function is used to correct the obtained simulated degradation data and the actual degradation data, as follows: Where y is the value after the Box-Cox transformation; To adjust the parameters.

5. The bearing degradation prediction method using a deep latent variable state-space model according to claim 1, characterized in that, Various time-domain degradation characteristics include: indicators sensitive to overall damage, such as the root mean square reflecting signal energy; indicators sensitive to local damage, such as kurtosis reflecting the degree of impact; and dimensionless indicators independent of operating conditions and sensitive to damage and faults, such as waveform factor and crest factor.

6. The bearing degradation prediction method using a deep latent variable state-space model according to claim 1, characterized in that, The prediction model based on the deep latent variable state-space model is as follows: Model Priors: Model posterior: in, Let be the joint distribution of the model, representing the entire process from time 1 to time T in which the model generates predicted values. These are the predicted degradation values ​​from time 1 to time T. Let these be the latent variables from time 1 to time T. Let T be the intermediate variable in the transition distribution from time 1 to time T. These are the initial intermediate variables for the recurrent neural network. The input degradation state is from time 1 to time T. The emission distribution represents the process of generating predicted values ​​from latent variables. This provides the prior distribution and the initial hypothetical distribution for the latent variables. The transition distribution represents the prior distribution of states at each time step, given the previous state and the current input to the model. This is used to learn the temporal correlations in bearing degradation data. This represents the process of generating the prior mean and prior variance.

7. The bearing degradation prediction method using a deep latent variable state-space model according to claim 6, characterized in that, The degradation observation equation is: Where z is a latent variable. For the weights and biases of the neural network, It follows a Gaussian distribution. for t Latent variables of time step, This represents the process of generating the prior mean and prior variance, from... Obtained through neural network fitting. For neural network parameters, These are predicted values.

8. A bearing degradation prediction system based on a deep latent variable state-space model, characterized in that, include: The data module, based on the required predicted bearing parameters and corresponding operating conditions, establishes a second-order pseudo-cyclic steady bearing fault simulation model based on analytical formulas to obtain the bearing's simulation degradation data. The correction module corrects the obtained simulated degradation data and real degradation data, integrates multiple time-domain degradation features to obtain degradation health indicators, and then obtains the degradation state characterizing the bearing degradation rate through differential transformation. The equation module, within the state-space model framework, uses recurrent neural networks as state transition equations and combines them with variational autoencoders as degenerate observation equations to construct a prediction model based on a deep latent variable state-space model. The prediction module extracts, fuses, and differs features from simulated degradation data and real data to obtain their degradation states. It uses the simulated degradation data to obtain a pre-trained deep latent variable state space model of the degradation state, initializes the weights of the state space model, and then uses the prediction model based on the deep latent variable state space model to predict the degradation state obtained from the real data. By accumulating the predicted degradation states, the predicted value of bearing degradation is obtained.

9. A computer-readable storage medium for storing one or more programs, characterized in that, The one or more programs include instructions that, when executed by a computing device, cause the computing device to perform the method of any one of claims 1 to 7.

10. A computing device, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including steps for performing the method of any one of claims 1 to 7.

Images