An adaptive prediction method and device for the remaining life of bearings of coal mine machinery and equipment

Abstract

The present invention provides an adaptive prediction method and device for the remaining life of bearings of coal mine machinery and equipment, belonging to the technical field of bearing remaining life prediction; solves the problem of how to more accurately predict the development rate of bearing damage and the remaining life of the bearing after a bearing fault is detected; comprises the following steps: collecting the vibration signal and rotation speed of the bearing, sequentially calculating the bearing vibration characteristic-energy level and energy level; smoothing the bearing vibration characteristic-energy level and energy level; sequentially calculating the actual damage degree of the bearing based on the smoothed vibration characteristic-energy level and energy level; updating the model parameters of the bearing physical model based on the deviation between the actual damage degree of the bearing and the bearing damage degree predicted by the bearing physical model; and providing a prediction of the bearing damage degree and the remaining life of the bearing based on the updated model parameters of the bearing physical model; the present invention is applied to the prediction of the remaining life of the bearing.

Description

Technical Field

[0001] The present invention provides a method and device for adaptively predicting the remaining life of a bearing of coal mine machinery and equipment, belonging to the technical field of remaining life prediction of bearings. Background Art

[0002] Predicting the remaining life of bearings in coal mining machinery and equipment has long been a technical challenge urgently needed to be addressed in the field of intelligent coal mining. Due to the harsh operating environment of bearings in coal mining machinery and equipment and the inherent uncertainty of bearing fatigue growth, bearing damage progression exhibits a high degree of statistical randomness, which directly leads to significant variability in the remaining life of different bearings in coal mining machinery and equipment. Traditional deterministic models based on material fracture mechanics cannot accurately describe the progression of bearing damage. This is because the model parameters of these prediction models are predetermined based on experimental data or historical operating data, while the actual operating environment of bearings is unpredictable. Furthermore, even small fluctuations in the operating environment accumulate during bearing operation, resulting in significant uncertainty in remaining life prediction. Bearing condition monitoring can estimate the actual extent of bearing damage by measuring signals related to bearing health, such as vibration, acoustic emission, and temperature. The remaining life of the bearing is then predicted by analyzing the trends of these damage levels. Because this method utilizes real-time measured signals, it can accurately estimate the current extent of bearing damage. However, the uncertainty of the future operating environment and the development of bearing damage in practice means that it can only reliably predict the remaining life of the bearing within a short timeframe. Therefore, how to effectively integrate the above-mentioned information from the two levels of real-time bearing measurement signals and physical models of the bearings, accurately predict the development rate of bearing damage and the remaining life of the bearing after a bearing fault is detected, enable the operation and maintenance personnel of coal mine machinery and equipment to effectively plan the maintenance and replacement of coal mine machinery and equipment to avoid secondary damage, and reasonably arrange the use of coal mine machinery and equipment to improve utilization efficiency, is a key technical link that needs to be solved for the implementation of condition-based maintenance of coal mine machinery and equipment.

[0003] Physical models of bearing damage progression can reasonably predict the remaining life of a bearing over a longer timeframe. However, these predictions often have significant uncertainty due to changes or fluctuations in the bearing's operating environment. Prediction methods based on real-time bearing measurement signals can accurately estimate the current extent of bearing damage, but in practice, uncertainties in the bearing's future operating environment and the progression of bearing damage mean they can only reliably predict the remaining life over a shorter timeframe. Summary of the Invention

[0004] In order to solve the problem of more accurately predicting the development rate of bearing damage and the remaining life of the bearing after a bearing fault is detected, the present invention proposes an adaptive prediction method and device for the remaining life of the bearing of coal mine machinery and equipment.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: an adaptive prediction method for the remaining life of bearings of coal mine machinery and equipment, comprising the following steps:

[0006] S1: Bearing vibration feature extraction: Collect the bearing vibration signal and speed, segment them by frame, and sequentially calculate the bearing vibration features - energy level RMSband and energy level Etol;

[0007] S2: Data preprocessing: Smoothing the bearing vibration characteristics - energy level RMSband and energy level Etol, and sequentially calculating the smoothed vibration characteristics - energy level and energy levels

[0008] S3: Quantitative estimation of bearing damage extent: based on smoothed vibration signature-energy level and energy levels Sequentially calculate the actual damage degree of the bearing;

[0009] S4: Adaptive filtering: updating the model parameters of the bearing physical model based on the deviation between the actual bearing damage degree and the bearing damage degree predicted by the bearing physical model;

[0010] S5: Prediction of the remaining life of the bearing: Based on the model parameters of the updated bearing physical model, the prediction of the bearing damage degree and the remaining life of the bearing are sequentially given.

[0011] The feature extraction of bearing vibration in step S1 specifically includes:

[0012] S11: Extract the vibration signal of the bearing resonance frequency band using bandpass filtering;

[0013] S12: calculating the energy level of the bearing resonance frequency band according to the vibration signal of the bearing resonance frequency band;

[0014] S13: Extract the envelope spectrum of the bearing resonance frequency band vibration signal using Hilbert transform and Fourier transform;

[0015] S14: Calculating the energy level of the bearing characteristic frequency according to the envelope spectrum of the vibration signal of the bearing resonance frequency band.

[0016] In step S3, the quantitative estimation of the bearing damage degree uses a correlation vector machine to provide a probability distribution of the quantitative estimation of the bearing damage degree.

[0017] In step S4, unscented particle filtering is used to calculate the posterior probability distribution of the model parameters of the bearing physical model, so as to update the model parameters of the bearing physical model.

[0018] The prediction of the remaining life of the bearing in step S5 specifically includes:

[0019] S51: Establish a deterministic empirical model between the bearing damage development rate at a certain moment and the damage extent at that moment under constant operating conditions - the bearing damage development model;

[0020] S52: Based on the bearing damage development model, a prediction of the current bearing damage level and the remaining bearing life is given.

[0021] An adaptive prediction device for the remaining life of a bearing of coal mine machinery and equipment, comprising:

[0022] Bearing vibration feature extraction module: This module collects the bearing vibration signal and speed, segments the data into frames, and sequentially calculates the bearing vibration features (RMSband and Etol), which are then sent to the data preprocessing module.

[0023] Data preprocessing module: used to smooth the bearing vibration characteristics - energy level RMSband and energy level Etol, and sequentially calculate the smoothed vibration characteristics - energy level and energy levels Send to the quantitative estimation module of bearing damage degree;

[0024] Quantitative estimation module of bearing damage degree: used to estimate the energy level based on the smooth vibration characteristics and energy levels Sequentially calculate the actual damage degree of the bearing and send it to the adaptive filtering module;

[0025] Adaptive filtering module: used to update the model parameters of the bearing physical model based on the deviation between the actual bearing damage degree and the bearing damage degree predicted by the bearing physical model, and send the updated model parameters of the bearing physical model to the bearing remaining life prediction module;

[0026] Bearing remaining life prediction module: used to sequentially predict the bearing damage degree and the bearing remaining life based on the model parameters of the updated bearing physical model, and send the bearing damage degree predicted by the bearing physical model to the adaptive filtering module.

[0027] The bearing vibration feature extraction module includes the following steps when calculating the bearing vibration features - energy level RMSband and energy level Etol:

[0028] The bearing vibration feature extraction module uses bandpass filtering to extract the vibration signal of the bearing resonance frequency band;

[0029] The bearing vibration feature extraction module calculates the energy level of the bearing resonance frequency band based on the vibration signal of the bearing resonance frequency band;

[0030] The bearing vibration feature extraction module uses Hilbert transform and Fourier transform to extract the envelope spectrum of the bearing resonance frequency band vibration signal;

[0031] The bearing vibration feature extraction module calculates the energy level of the bearing characteristic frequency based on the envelope spectrum of the bearing resonance frequency band vibration signal.

[0032] The quantitative estimation module for the bearing damage degree uses a correlation vector machine to provide a probability distribution for the quantitative estimation of the bearing damage degree.

[0033] The adaptive filtering module uses unscented particle filtering to calculate the posterior probability distribution of model parameters of the bearing physical model based on the deviation between the actual damage degree of the bearing and the bearing damage degree predicted by the bearing physical model, thereby updating the model parameters of the bearing physical model.

[0034] The prediction module for the remaining life of the bearing first establishes a deterministic empirical model between the damage development rate of the bearing at a certain moment and the damage extent at that moment under constant operating conditions - a bearing damage development model;

[0035] The bearing remaining life prediction module provides a prediction of the current bearing damage degree and the bearing remaining life according to the bearing damage development model.

[0036] The present invention has the following beneficial effects compared to the prior art: the adaptive prediction method for the remaining life of bearings of coal mine machinery and equipment provided by the present invention combines the advantages of two strategies: a physical model of bearing damage development and a prediction method based on real-time bearing measurement signals. After a bearing fault is detected, the development rate of bearing damage and the remaining life of the bearing can be more accurately predicted, enabling operators and maintenance personnel of coal mine machinery and equipment to effectively plan the maintenance and replacement of coal mine machinery and equipment to avoid secondary damage, and rationally arrange the use of coal mine machinery and equipment to improve utilization efficiency.

[0037] The feature extraction, bearing damage quantitative estimation, and bearing physical model parameter update algorithms adopted in the present invention are simple and have low requirements on computing resources. They facilitate coal mine machinery and equipment manufacturers and users to perform online configuration of fault prediction capabilities as needed, thereby realizing state-based maintenance of coal mine machinery and equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The present invention will be further described below with reference to the accompanying drawings:

[0039] Figure 1 is a flow chart of the method of the present invention;

[0040] Figure 2 This is a logic diagram for adaptively updating model parameters of the bearing physical model of the present invention. DETAILED DESCRIPTION

[0041] like Figure 1-2 As shown, to combine the advantages of real-time bearing measurement signals and physical bearing models, and to more accurately predict the bearing damage progression rate and remaining life after a bearing fault is detected, this invention constructs a set of complementary vibration signatures to estimate the actual damage level of mining equipment bearings in real time. The deviation between this estimate and the damage level predicted by the bearing physical model captures the impact of changes or fluctuations in the bearing operating environment on the development of bearing damage. Based on this deviation, an adaptive filtering algorithm is then used to update the model parameters of the bearing physical model. This updated physical model limits the uncertainty in bearing fault prediction, thereby achieving a more accurate prediction of the bearing's remaining life and providing the necessary decision-making basis for condition-based maintenance of coal mining machinery and equipment.

[0042] like Figure 1 As shown, this embodiment provides an adaptive prediction method for the remaining life of a bearing of coal mine machinery and equipment, comprising the following steps:

[0043] S1, the bearing vibration feature extraction module, collects the bearing vibration signal and speed, divides them into frames, sequentially calculates the bearing vibration features - energy level RMSband and energy level Etol, and sends them to the data preprocessing module.

[0044] The bearing vibration feature extraction module in step S1 includes the following steps when calculating the bearing vibration features - energy level RMSband and energy level Etol:

[0045] S11, a bearing vibration feature extraction module uses bandpass filtering to extract the vibration signal of the bearing resonance frequency band;

[0046] The cut-off frequency of the band-pass filter is set according to the resonance frequency band range of the bearing and the supporting structure in actual application;

[0047] S12 , the bearing vibration feature extraction module calculates the energy level RMSband of the bearing resonance frequency band according to the vibration signal of the bearing resonance frequency band.

[0048] The energy level RMSband of the bearing resonance frequency band is realized by the following formula:

[0049]

[0050] In the above formula: x(i) is the bandpass filtered signal, is the mean of the signal, and K is the number of samples of the signal.

[0051] S13, a bearing vibration feature extraction module uses Hilbert transform and Fourier transform to extract the envelope spectrum of the bearing resonance frequency band vibration signal;

[0052] S14, the bearing vibration feature extraction module calculates the energy level Etol of the bearing characteristic frequency according to the envelope spectrum of the bearing resonance frequency band vibration signal.

[0053] The energy level Etol of the bearing characteristic frequency is realized by the following formula:

[0054] Etol=E(f bpfi )+E(f bpf o)+E(2f bsf )(2);

[0055] In the above formula: E(f bpfi ), E(f bpf o), E(2f bsf ) are the energies near the characteristic frequencies of the bearing inner ring, outer ring and rolling element in the envelope spectrum.

[0056] The energy level E(f bpfi ), E(f bpf o) and E(2f bsf ) is achieved through the following formula:

[0057]

[0058] Where Env(f) is the envelope spectrum of the vibration signal in the bearing resonance frequency band, and △f / 2 is the neighborhood radius of the bearing characteristic frequency f.

[0059] The bearing corresponds to the characteristic frequency f of the inner ring, outer ring and rolling element bpfi , f bpf o and 2f bsf This is achieved through the following formula:

[0060]

[0061]

[0062]

[0063] In the above formula: f s is the frequency of the rotating shaft, ζ is the contact angle of the bearing rolling element, D is the diameter of the bearing rolling element, d m is the bearing pitch diameter and Z is the number of bearing rolling elements.

[0064] S2, data preprocessing module, smoothes the bearing vibration characteristics - energy level RMSband and energy level Etol, and sequentially calculates the smoothed vibration characteristics - energy level and energy levels Sent to the quantitative estimation module of bearing damage extent.

[0065] The smooth vibration characteristics are achieved by the following formula:

[0066] csig(t i )=ε·sig(t i )+(1-ε)·csig(t i-1 ) (7);

[0067] In the above formula: sig(t i ) is t i The original signal at time csig(t i ) is t i The smoothed signal at the moment, ε is the smoothing parameter of exponential smoothing, which is set according to the actual application environment.

[0068] S3, a quantitative estimation module for the bearing damage degree, is based on the smoothed vibration characteristics - energy level and energy levels The actual damage degree of the bearing is calculated sequentially and sent to the adaptive filtering module.

[0069] The quantitative estimation module for the bearing damage degree in step S3 includes the following steps when calculating the actual damage degree of the bearing:

[0070] S31, the quantitative estimation module of the bearing damage degree uses a correlation vector machine to provide a probability distribution of the quantitative estimation of the bearing damage degree.

[0071] The quantitative estimation model of bearing damage degree is in the following form:

[0072]

[0073] In the above formula: w=[w1,..,w M ] T 、 are the weights and basis functions of the quantitative estimation model, respectively. The input x is the smoothed bearing vibration eigenvector Target t is the damage degree of the bearing.

[0074] The quantitative estimation of the bearing damage degree is achieved by the following recursive algorithm:

[0075] initialization:

[0076] (a) Model parameters {w i} corresponding to all hyperparameters {a i} and noise variance σ 2 .

[0077] Iteration:

[0078] (a) Calculate the posterior distribution of parameter w p{w|t, a, σ 2}statistics μ and ∑, where p{w|t, a, σ 2} obeys Gaussian distribution with mean μ=σ -2 ∑Φ T t, the variance is ∑=(σ -2 Φ T Φ+A) -1 , Φ is the N×M design matrix, N is the number of samples in the training set, M is the number of hyperparameters of the model, and A=diag(a1,...,a M ) is a diagonal matrix consisting of all hyperparameters a;

[0079] (b) Calculate all {γ i}, and then re-estimate {a i} and σ 2 ;

[0080] γ i =1-a i ∑ ii (8);

[0081]

[0082]

[0083] where γ i It is an intermediate variable introduced in the calculation process;

[0084] (c) Repeat (a)-(b) until convergence, that is, obtain the hyperparameter a and noise variance σ 2 The posterior distribution p{a, σ 2 The mode a of |t} MP and

[0085] estimate:

[0086] (a) Delete the hyperparameter {a i}, middle a i =∞ corresponding model parameter w i and basis function φ i (x);

[0087] (b) For the new target t * , is estimated by the following posterior distribution:

[0088] The distribution is Gaussian with mean g(x * ; μ), the variance is

[0089] S4, an adaptive filtering module, updates the model parameters of the bearing physical model according to the deviation between the actual damage degree of the bearing and the bearing damage degree predicted by the bearing physical model, and sends the updated model parameters of the bearing physical model to the prediction module of the remaining life of the bearing.

[0090] In step S4, when the adaptive filtering module updates the model parameters of the bearing physical model according to the deviation between the actual damage degree of the bearing and the bearing damage degree predicted by the bearing physical model, the following steps are included:

[0091] S41, an adaptive filtering module, uses an unscented particle filter to calculate the posterior probability distribution of the model parameters of the bearing physical model based on the deviation between the actual damage degree of the bearing and the bearing damage degree predicted by the bearing physical model, thereby updating the model parameters of the bearing physical model.

[0092] The bearing physical model can be transformed into the following state space model:

[0093] x t =x t-1 +v t (11);

[0094] y t =h(n t , x t )+τ t (12);

[0095] in is the state of the system at time t, n x is the system state x t The latitude value, y t is the observation of the system at time t, is the input of the system, n n is the system input n t The dimension value of is the process noise, n v is the process noise v t Dimension value, and assuming v t Obeying a Gaussian distribution with a mean of 0 and a variance of Q, τ t is the measurement noise, and assuming that τ t It follows a Gaussian distribution with mean 0 and variance R.

[0096] like Figure 2 As shown, the extended state variable is set to the union of the system state and the noise variable The unscented particle filter updates the model parameters of the bearing physical model through the following recursive algorithm:

[0097] Initialization: t = 0;

[0098] For i=1,...,N, sample states (particles) from the prior distribution p(x0) And set:

[0099]

[0100]

[0101]

[0102]

[0103] in They are the sampling states (particles) of the prior distribution p(x0) The mean and variance of They are the prior distribution p(x0) sampling state (particle) The mean and variance of the corresponding weight samples or sigma points.

[0104] For t=1,2,…, the filtering process is as follows:

[0105] (a) Importance Sampling

[0106] For i=1,…N, use the unscented Kalman filter to update the particles and calculate the sigma point:

[0107]

[0108] In the above formula: is the weighted sampling or sigma point of the sampling state (particle) at time t-1, λ is the composite scaling parameter, n a =n x +n v +n n .

[0109] Transfer particles to the next moment (time update):

[0110]

[0111]

[0112]

[0113]

[0114]

[0115] In the above formula: They are the state of the system at time t calculated by the state transfer model based on the data at time t-1, as well as the corresponding mean and variance, They are the observation of the system at time t and the corresponding mean calculated by the state observation model through the data at time t-1;

[0116] in

[0117]

[0118]

[0119]

[0120] Integrate new observations (measurement updates):

[0121]

[0122]

[0123]

[0124]

[0125]

[0126] In the above formula: K t is an intermediate variable introduced in the calculation process. They are respectively the mean and variance of the state of the system at time t obtained by comprehensive observation of the system at time t.

[0127] sampling:

[0128] set up and

[0129] For i=1,…,N, the importance weight is estimated and the calculation formula is as follows:

[0130]

[0131] In the above formula: is the importance weight of the sampling state (particle), are the state transfer distribution and state observation distribution of the system at time t, is the posterior distribution of the system state at time t.

[0132] For i=1,…,N, the normalized importance weight is calculated as follows:

[0133]

[0134] In the above formula: is the normalized importance weight of the sampled state (particle).

[0135] (b) Resampling

[0136] Based on high / low importance weight Duplicate / compress particles individually To obtain N approximate distributions Random particles

[0137] (c) Output: The algorithm output is a set of samples that approximate the posterior distribution, as shown in the following formula:

[0138]

[0139] In the above formula: p(x 0：t |y 1：t ) is the posterior distribution of the system state at time t:0, is an approximation of the posterior distribution of the system state at time 0:t given by the sampling importance resampling algorithm, is the Dirac function.

[0140] S5, a bearing remaining life prediction module, sequentially gives a prediction of the bearing damage degree and the bearing remaining life according to the model parameters of the updated bearing physical model, and sends the bearing damage degree predicted by the bearing physical model to the adaptive filtering module.

[0141] The prediction module for the remaining life of the bearing in step S5 includes the following steps when providing a prediction of the bearing damage degree and the remaining life of the bearing:

[0142] S51, the prediction module of the remaining life of the bearing establishes a deterministic empirical model between the damage development rate dD / dN of the bearing at a certain moment under constant operating conditions and the damage degree D at this moment - the bearing damage development model.

[0143] The bearing damage development model is established by the following formula:

[0144]

[0145] Where D is the damaged area of ​​the bearing surface at a certain moment, υ and τ are two model parameters related to the operating environment of the bearing, and different constants are used according to different situations.

[0146] Given the initial bearing damage level, the bearing damage development model can be transformed into the following form:

[0147] InD=θ1+θ2In(n+θ0) (34);

[0148] Among them, θ1, θ2, θ0 are parameters related to the bearing operating environment. Set the system state to x = [θ1, θ2, θ0] T , the system output is y(n)=InD(n)=h(n,x), and the bearing damage development model can be further transformed into the state space model shown in formula (11) and formula (12).

[0149] S52, the bearing remaining life prediction module provides a prediction of the current bearing damage degree and the bearing remaining life according to the bearing damage development model.

[0150] The current bearing damage degree is predicted by the following formula:

[0151]

[0152] Among them g t (x 0：t ) is the system observation or output at time t, given by formula (34), that is, g t (x 0：t )=lnD,p(x 0：t |y 1：t ) is the posterior distribution of the system state, given by formula (32).

[0153] Given the actual bearing damage degree D i And the allowable damage degree of the bearing D c Then, the current prediction of the remaining life of the bearing is realized by the following formula:

[0154]

[0155] where υ and τ are the model parameters of the current bearing damage development model.

[0156] The present invention also proposes an adaptive prediction device for the remaining life of bearings of coal mine machinery and equipment, comprising a bearing vibration feature extraction module, a data preprocessing module, a bearing damage degree quantitative estimation module, an adaptive filtering module, and a bearing remaining life prediction module. In response to the harsh operating environment of coal mine machinery and equipment bearings and the resulting high uncertainty in the development of bearing damage, a set of vibration features with complementary information is constructed through the bearing vibration feature extraction module. These features, as a whole, can reliably track the development of bearing damage. After smoothing them in the data preprocessing module, the actual bearing damage degree is estimated through the bearing damage degree quantitative estimation module. The adaptive filtering module then updates the model parameters in the bearing remaining life prediction module based on the deviation between this estimated value and the bearing damage degree predicted by the bearing remaining life prediction module. Finally, the bearing remaining life prediction module, after the model parameters have been updated, provides an accurate prediction of the bearing remaining life. The present invention uses information fusion technology to integrate the real-time measurement signals of the bearings with the physical model of the bearings, effectively limiting the uncertainty of the bearing physical model in predicting the remaining life. This enables the operation and maintenance personnel of coal mine machinery and equipment to accurately predict the development rate of bearing damage and the remaining life of the bearing after detecting a bearing fault, effectively plan the maintenance and replacement of coal mine machinery and equipment to avoid secondary damage, and reasonably arrange the use of coal mine machinery and equipment to improve utilization efficiency.

[0157] The present invention uses real-time measurement signals that reflect the current health status of the bearing to update the a priori model parameters of the bearing damage development model. In this way, the a posteriori model parameters take into account the impact of changes or fluctuations in the operating environment of the bearings of coal mine machinery and equipment on the development of bearing damage. Therefore, the updated model can provide a more accurate prediction of the bearing damage development rate and the remaining life of the bearing, providing the necessary decision-making basis for condition-based maintenance of coal mine machinery and equipment.

[0158] Regarding the specific structure of the present invention, it should be noted that the connection relationship between the various component modules adopted in the present invention is definite and feasible. Except for those specifically described in the embodiments, the specific connection relationship can bring about corresponding technical effects and solve the technical problems raised by the present invention without relying on the execution of corresponding software programs. The components, modules, models of specific components appearing in the present invention, the connection methods between each other, and the conventional usage methods and expected technical effects brought about by the above-mentioned technical features, except for those specifically described, all belong to the disclosed contents in patents, journal articles, technical manuals, technical dictionaries, and textbooks that can be obtained by technical personnel in this field before the application date, or belong to the existing technologies such as conventional technology and common knowledge in this field. There is no need to elaborate, so that the technical solution provided in this case is clear, complete, and feasible, and the corresponding physical products can be reproduced or obtained based on this technical means.

[0159] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An adaptive prediction method for the remaining life of bearings of coal mine machinery and equipment, characterized by: The steps include: S1: Bearing vibration feature extraction: Collect the bearing vibration signal and speed, segment them by frame, and sequentially calculate the bearing vibration features - energy level RMSband and energy level Etol; The energy level RMSband refers to the energy level RMSband of the bearing resonance frequency band, which is calculated based on the vibration signal of the bearing resonance frequency band. The calculation formula is as follows: ； In the above formula: x(i) is the bandpass filtered signal, is the mean of this signal, K is the number of samples of this signal; The energy level Etol refers to the energy level Etol of the bearing characteristic frequency, which is calculated based on the envelope spectrum of the vibration signal in the bearing resonance frequency band. The calculation formula is as follows: Etol=E(f bpfi )+E(f bpfo )+E(2f bsf ); In the above formula: E(f bpfi )、E(f bpfo )、E(2f bsf ) are the energies near the characteristic frequencies of the bearing inner ring, outer ring and rolling element in the envelope spectrum respectively; S2: Data preprocessing: Smoothing the bearing vibration characteristics - energy level RMSband and energy level Etol, and sequentially calculating the smoothed vibration characteristics - energy level and energy levels ; S3: Quantitative estimation of bearing damage extent: based on smoothed vibration signature-energy level and energy levels , sequentially calculate the actual damage degree of the bearing; S4: Adaptive filtering: updating the model parameters of the bearing physical model based on the deviation between the actual bearing damage degree and the bearing damage degree predicted by the bearing physical model; S5: Prediction of the remaining life of the bearing: Based on the model parameters of the updated bearing physical model, the prediction of the bearing damage degree and the remaining life of the bearing are sequentially given.

2. The adaptive prediction method for the remaining life of a bearing of coal mine machinery and equipment according to claim 1, characterized in that: The feature extraction of bearing vibration in step S1 specifically includes: S11: Extract the vibration signal of the bearing resonance frequency band using bandpass filtering; S12: calculating the energy level of the bearing resonance frequency band according to the vibration signal of the bearing resonance frequency band; S13: Extract the envelope spectrum of the bearing resonance frequency band vibration signal using Hilbert transform and Fourier transform; S14: Calculating the energy level of the bearing characteristic frequency according to the envelope spectrum of the vibration signal of the bearing resonance frequency band.

3. The adaptive prediction method for the remaining life of a bearing of coal mine machinery and equipment according to claim 2, characterized in that: In step S3, the quantitative estimation of the bearing damage degree uses a correlation vector machine to provide a probability distribution of the quantitative estimation of the bearing damage degree.

4. The adaptive prediction method for the remaining life of a bearing of coal mine machinery and equipment according to claim 3 is characterized by: In step S4, unscented particle filtering is used to calculate the posterior probability distribution of the model parameters of the bearing physical model, so as to update the model parameters of the bearing physical model.

5. The adaptive prediction method for the remaining life of a bearing of coal mine machinery and equipment according to claim 4, characterized in that: The prediction of the remaining life of the bearing in step S5 specifically includes: S51: Establish a deterministic empirical model between the bearing damage development rate at a certain moment and the damage extent at that moment under constant operating conditions - the bearing damage development model; S52: Based on the bearing damage development model, a prediction of the current bearing damage level and the remaining bearing life is given.

6. An adaptive prediction device for the remaining life of bearings of coal mine machinery and equipment, characterized by: include: Bearing vibration feature extraction module: This module collects the bearing vibration signal and speed, segments the data into frames, and sequentially calculates the bearing vibration features (RMSband and Etol), which are then sent to the data preprocessing module. The energy level RMSband refers to the energy level RMSband of the bearing resonance frequency band, which is calculated based on the vibration signal of the bearing resonance frequency band. The calculation formula is as follows: ； In the above formula: x(i) is the bandpass filtered signal, is the mean of this signal, K is the number of samples of this signal; The energy level Etol refers to the energy level Etol of the bearing characteristic frequency, which is calculated based on the envelope spectrum of the vibration signal in the bearing resonance frequency band. The calculation formula is as follows: Etol=E(f bpfi )+E(f bpfo )+E(2f bsf ); In the above formula: E(f bpfi )、E(f bpfo )、E(2f bsf ) are the energies near the characteristic frequencies of the bearing inner ring, outer ring and rolling element in the envelope spectrum respectively; Data preprocessing module: used to smooth the bearing vibration characteristics - energy level RMSband and energy level Etol, and sequentially calculate the smoothed vibration characteristics - energy level and energy levels , sent to the quantitative estimation module of bearing damage degree; Quantitative estimation module of bearing damage degree: used to estimate the energy level based on the smooth vibration characteristics and energy levels , sequentially calculate the actual damage degree of the bearing and send it to the adaptive filtering module; Adaptive filtering module: used to update the model parameters of the bearing physical model based on the deviation between the actual bearing damage degree and the bearing damage degree predicted by the bearing physical model, and send the updated model parameters of the bearing physical model to the bearing remaining life prediction module; Bearing remaining life prediction module: used to sequentially predict the bearing damage degree and the bearing remaining life based on the model parameters of the updated bearing physical model, and send the bearing damage degree predicted by the bearing physical model to the adaptive filtering module.

7. The adaptive prediction device for the remaining life of a bearing of coal mine machinery and equipment according to claim 6, characterized in that: The bearing vibration feature extraction module includes the following steps when calculating the bearing vibration features - energy level RMSband and energy level Etol: The bearing vibration feature extraction module uses bandpass filtering to extract the vibration signal of the bearing resonance frequency band; The bearing vibration feature extraction module calculates the energy level of the bearing resonance frequency band based on the vibration signal of the bearing resonance frequency band; The bearing vibration feature extraction module uses Hilbert transform and Fourier transform to extract the envelope spectrum of the bearing resonance frequency band vibration signal; The bearing vibration feature extraction module calculates the energy level of the bearing characteristic frequency based on the envelope spectrum of the bearing resonance frequency band vibration signal.

8. The adaptive prediction device for the remaining life of a bearing of coal mine machinery and equipment according to claim 6, characterized in that: The quantitative estimation module for the bearing damage degree uses a correlation vector machine to provide a probability distribution for the quantitative estimation of the bearing damage degree.

9. The adaptive prediction device for the remaining life of a bearing of coal mine machinery and equipment according to claim 6, characterized in that: The adaptive filtering module uses unscented particle filtering to calculate the posterior probability distribution of model parameters of the bearing physical model based on the deviation between the actual damage degree of the bearing and the bearing damage degree predicted by the bearing physical model, thereby updating the model parameters of the bearing physical model.

10. The adaptive prediction device for the remaining life of a bearing of coal mine machinery and equipment according to claim 6, characterized in that: The prediction module for the remaining life of the bearing first establishes a deterministic empirical model between the damage development rate of the bearing at a certain moment and the damage extent at that moment under constant operating conditions - a bearing damage development model; The bearing remaining life prediction module provides a prediction of the current bearing damage degree and the bearing remaining life according to the bearing damage development model.

Images