CN115292820A - Method for predicting residual service life of urban rail train bearing - Google Patents
Method for predicting residual service life of urban rail train bearing Download PDFInfo
- Publication number
- CN115292820A CN115292820A CN202210992969.6A CN202210992969A CN115292820A CN 115292820 A CN115292820 A CN 115292820A CN 202210992969 A CN202210992969 A CN 202210992969A CN 115292820 A CN115292820 A CN 115292820A
- Authority
- CN
- China
- Prior art keywords
- bearing
- energy
- lstm
- network
- lstm network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/15—Vehicle, aircraft or watercraft design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
Abstract
The invention discloses a method for predicting the residual service life of an urban rail train bearing, which comprises the following steps of setting the rotating speed and sampling frequency of the bearing, collecting the original vibration signal of the bearing in the full life cycle when the bearing runs to failure, extracting the time domain characteristics of the original vibration signal of the bearing, and measuring the similarity degree between the characteristics by adopting a similarity measurement method; carrying out variation modal decomposition on the original vibration signal of the bearing to obtain modal component characteristics, and adopting an energy entropy discrimination method according to the principle that the larger the energy entropy is, the larger the information uncertainty is; combining the time domain characteristics with the obtained modal component characteristics, and dividing the time domain characteristics into a training set and a verification set; and establishing an LSTM prediction network by using the optimized hyper-parameters, inputting the characteristic matrix into the LSTM network, updating the weight through a back propagation algorithm, updating the gradient through an Adam optimizer, and outputting the RUL prediction value of the bearing after calculation. The method can scientifically, efficiently and comprehensively predict the RUL of the urban rail train bearing, and has the advantages of quick and effective network hyper-parameter selection, good prediction performance and the like.
Description
Technical Field
The invention belongs to the technical field of fault prediction of rotating part bearings of rail transit trains, and particularly relates to a method for predicting the residual service life of an urban rail train bearing.
Background
In recent years, the rapid development of urban rail transit becomes an important component of public transportation, and the driving reliability and safety of urban rail trains are widely regarded by various social circles. Bearings are indispensable parts in the rotating machinery of trains, and studies have shown that about 50% of motor failures are caused by rolling bearing failures. Therefore, the prediction of the residual service life of the bearing has important significance on the operation and maintenance work of the train. At present, the industrial industry mostly adopts periodical preventive maintenance for train parts, and maintenance and inspection are carried out by combining major repair, middle repair and minor repair, so that faults can be reduced as much as possible, but the phenomena of excessive maintenance and insufficient maintenance still exist. In order to ensure reliable operation of trains, reduce the number and frequency of failures, and reduce economic loss and even personal safety hazards caused by failure shutdown, academic and industrial fields are focusing on maintaining trains by using a emerging technology of failure Prediction and Health Management (PHM). The PHM monitors and acquires system running state data by using a sensor, performs data processing and feature extraction, evaluates the health condition of a monitored object through a fault diagnosis and prediction model, predicts the fault occurrence of the monitored object, and provides guidance and decision for the overhaul and maintenance of the monitored object. The remaining service life (RUL) is an important basis for evaluating the health condition of a subject, the PHM effectively displays the degradation trend of the PHM by monitoring the bearing state and predicting the RUL of the PHM, and dynamically establishes and optimizes a maintenance strategy according to the running state, so that the faults of rotating parts such as a motor and the like are avoided, and the running reliability and safety of a train are improved.
The bearing life prediction mainly comprises two stages of data processing, feature extraction and RUL prediction. At present, bearing characteristics can be extracted by adopting a signal processing method such as time-frequency decomposition, modal decomposition and the like, and the problem of insufficient degradation state information caused by adopting single domain characteristics exists. In the RUL prediction method, the model fitting degradation state depends on expert prior knowledge, and the prediction precision and the applicability are difficult to reach higher levels. The data-driven method maps the relationship between the monitoring signal and the RUL value by mining degradation information in the state data through machine learning, deep learning, and the like, but the RUL prediction method such as the conventional support vector machine has a great limitation in using time series information of the bearing vibration signal. Long-short term memory networks (LSTMs) are suitable for addressing the long-term dependency problem and are widely used in bearing RUL prediction. The setting of the LSTM hyper-parameter can influence the prediction precision, the workload of manually selecting the hyper-parameter is large, the accuracy is low, the optimization of the hyper-parameter by some optimization algorithms is easy to fall into local optimization, the optimization capability is poor, the precision is not high enough, and the residual service life of the bearing is difficult to predict comprehensively and accurately.
Disclosure of Invention
The invention aims to: aiming at the existing problems, the invention provides a prediction method of the residual service life of an urban rail train bearing, the prediction method of the invention reserves the time sequence characteristic of a bearing vibration signal in the data processing and characteristic extraction stage, and selects the hyper-parameter of a neural network in the RUL prediction stage through Harris eagle optimization algorithm (HHO) self-adaptive optimization so as to improve the prediction precision; the method can scientifically, efficiently and comprehensively predict the RUL of the urban rail train bearing, and has the advantages of various characteristic information, quick and effective selection of the LSTM network hyper-parameters, good prediction performance and the like. In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
the invention provides a method for predicting the residual service life of an urban rail train bearing, which comprises the following steps:
step 1: setting the rotating speed and sampling frequency of the bearing, and acquiring the original vibration signal of the bearing in the full life cycle when the bearing runs to failure;
and 2, step: extracting time domain characteristics of the original vibration signals of the bearing, measuring the similarity degree between the characteristics by adopting a similarity measurement method, removing the characteristics with lower discrimination degree, and selecting the characteristic parameters which represent the optimal degradation capability of the bearing;
step 3, carrying out variation modal decomposition on the original vibration signal of the bearing to obtain modal component characteristics, selecting modal components according to the information uncertainty principle of the larger energy entropy and the smallest energy entropy by adopting an energy entropy discrimination method, expanding frequency domain information on the basis of time domain characteristics, and respectively calculating the energy and the energy entropy of the modal components;
and 4, step 4: combining the time domain features obtained in the step 2 with the modal component features obtained in the step 3, wherein k feature vectors with the sequence length of n are total, and the ith feature vector is F i =[F 1 (i),F 2 (i),...,F n (i)]In which F n (i) Representing the eigenvalues corresponding to the nth sampling time point, and then combining the k eigenvectors into an eigenvector matrix F = [ F = 1 ,F 2 ,...,F k ](ii) a Removing abnormal values in each feature vector, normalizing the feature vectors to be between 0 and 1 after completing the abnormal values through cubic spline interpolation, and dividing a feature matrix into a training set and a verification set;
and 5: adopting an LSTM network as an RUL prediction main network, setting the super-parameter of the LSTM network as a population position, training the LSTM network by training set data, using the mean square error of a verification set data prediction result as a fitness function, optimizing the LSTM super-parameter by adopting a Harris eagle optimization algorithm, and iteratively updating an optimal super-parameter combination each time the LSTM network is trained; and establishing an LSTM prediction network by using the optimized hyper-parameters, inputting the characteristic matrix into the LSTM network, and outputting the RUL prediction value of the bearing after calculation.
Further preferably, in the above scheme, the energy in the energy entropy discrimination method in step 3 satisfies the following expression:
wherein u i For the ith modal component sequence, t = {0,1,. And n }, where n is the sequence length;
the energy entropy satisfies:
H i =-p i log 10 p i , (2);
wherein p is i Is the ratio of the energy of the ith modal component to the total energy,
where i = {0,1,. K }, where K is the total modal component number.
Preferably, in the step 4, the feature matrix is constructed by selecting the variation modal decomposition modal component by a time domain feature extraction and energy entropy discrimination method, and the constructed feature matrix contains the time domain, frequency domain and entropy domain feature information of the bearing degradation.
Preferably, in step 5, the specific process of optimizing the LSTM network hyper-parameters by using the harris eagle optimization algorithm is as follows:
step 5.1: selecting the hyper-parameter learning rate, the batch size, the iteration period, the LSTM layer unit number and the Dense layer unit number of the LSTM network as the group positions, and setting the group number and the maximum iteration times;
and step 5.2: initializing Harris eagle populations and positions and energies of prey, training an LSTM network by training set data, taking the mean square error of a data prediction result of a verification set as a fitness function, and comparing and updating the fitness and the positions of the optimal individuals;
step 5.3: calculating escape energy of a prey, searching for an updated position globally when the escape energy is greater than or equal to 1, and searching for a local position locally when the escape energy is less than 1; in the local exploration process, when the escape energy of a prey is more than or equal to 0.5, harris hawks consume the energy through soft attack, when the escape energy is less than 0.5, harris hawks directly capture the prey through hard attack, and when the prey cannot escape, a team fast dive strategy is adopted for attack;
step 5.4: calculating the fitness of the individual after the position is updated by the Harris eagle predation strategy, and comparing and updating the fitness and the position of the optimal individual;
and step 5.5: and judging whether the end condition of the optimization is met, namely whether the set maximum iteration number is reached or whether the fitness meets the requirement, if so, outputting the optimized fitness and the hyperparameter corresponding to the optimal population position, and if not, returning to the step 5.3 to continue the optimization.
The scheme is further preferable, the LSTM network adaptively iteratively updates the optimal hyper-parameter through a Harris eagle optimization algorithm in the training process, and the optimal LSTM network is directly constructed by the optimal hyper-parameter without retraining.
In summary, because the invention adopts the technical scheme, the invention has the following remarkable effects:
the invention comprehensively considers the problems existing in the signal processing and the RUL prediction network, adopts signal processing methods such as time domain analysis and variational modal decomposition to extract characteristics, synthesizes time domain and frequency domain information, avoids incomplete effective degradation information of single characteristics, adopts the LSTM network as an RUL prediction main body network, adopts the Harris eagle optimization algorithm to adaptively optimize the hyper-parameters in the network training process, and effectively solves the problems that the workload of the manual selection of the hyper-parameters is too large and the hyper-parameters with optimal prediction performance are not easy to find. The method of the invention can scientifically and comprehensively predict the RUL of the bearing of the urban rail train, is an effective and practical prediction method, and can be widely applied to prediction of the residual service life of the bearing of various rotating parts of the train.
Drawings
FIG. 1 is a flow chart of a method for predicting the residual service life of an urban rail train bearing;
FIG. 2 is a fitness curve for HHO optimized LSTM hyperparameters;
fig. 3 is a comparison of the prediction curves for HHO optimized LSTM and the original network.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail below with reference to the accompanying drawings by way of examples of preferred embodiments. It should be noted, however, that the numerous details set forth in the description are merely for the purpose of providing the reader with a thorough understanding of one or more aspects of the present invention, which may be practiced without these specific details.
As shown in fig. 1 to 3, the method for predicting the remaining service life of the urban rail train bearing according to the invention comprises the following steps:
step 1: setting the rotating speed and sampling frequency of the bearing, and acquiring the original vibration signal of the bearing in the full life cycle when the bearing runs to failure;
step 2: extracting time domain characteristics of the original vibration signals of the bearing, measuring the similarity degree between the characteristics by adopting a similarity measurement method, removing the characteristics with lower discrimination degree, and selecting the characteristic parameters which represent the optimal degradation capability of the bearing; wherein the time domain characteristics are shown in table 1;
TABLE 1 time Domain characterization
Step 3, carrying out variation modal decomposition on the original vibration signal of the bearing to obtain modal component (IMF component) characteristics, selecting the modal component according to the information uncertainty principle that the energy entropy is larger and the energy entropy is smallest by adopting an energy entropy discrimination method, expanding frequency domain information on the basis of time domain characteristics, and respectively calculating the energy and the energy entropy of the modal component;
the energy satisfies the following expression:
wherein u is i For the ith modal component sequence, t = {0,1,. And n }, where n is the sequence length;
the energy entropy satisfies:
H i =-p i log 10 p i , (2);
wherein p is i Is the ratio of the energy of the ith modal component to the total energy,
where i = {0,1,. K }, where K is the total modal component number.
And 4, step 4: combining the time domain characteristics obtained in the step 2 with the modal component characteristics obtained in the step 3 to obtain k characteristic vectors with the sequence length of n, wherein the ith characteristic vector is F i =[F 1 (i),F 2 (i),...,F n (i)]In which F is n (i) Representing the eigenvalues corresponding to the nth sampling time point, and then combining the k eigenvectors into an eigenvector matrix F = [ F ] 1 ,F 2 ,...,F k ](ii) a Eliminating abnormal values in each feature vector, normalizing the feature vectors to be between 0 and 1 after completing the abnormal values through cubic spline interpolation, and dividing a feature matrix into a training set and a verification set; the method combines a time domain feature extraction method and an energy entropy discrimination method to select variation modal decomposition modal components to construct a feature matrix, wherein the constructed feature matrix contains the time domain, frequency domain and entropy domain feature information of bearing degradation;
and 5: adopting an LSTM network as an RUL prediction main network, setting the super-parameter of the LSTM network as a population position, training the LSTM network by training set data, using the mean square error of a verification set data prediction result as a fitness function, optimizing the LSTM super-parameter by adopting a Harris eagle optimization algorithm, and iteratively updating an optimal super-parameter combination each time the LSTM network is trained; and establishing an LSTM prediction network by using the optimized hyper-parameters, inputting the characteristic matrix into the LSTM network, updating the weight through a back propagation algorithm, updating the gradient through an Adam optimizer, and outputting the RUL prediction value of the bearing after calculation. In the embodiment of the invention, the specific process of optimizing the super-parameters of the LSTM network by adopting the Harris eagle optimization algorithm comprises the following steps:
step 5.1: selecting the hyper-parameter learning rate, the batch size, the iteration period, the LSTM layer unit number and the Dense layer unit number of the LSTM network as the group positions, and setting the group number and the maximum iteration times;
step 5.2: initializing Harris eagle populations and positions and energies of prey, training an LSTM network by training set data, taking the mean square error of a verification set data prediction result as a fitness function, and comparing and updating the fitness and the positions of the optimal individuals;
step 5.3: calculating escape energy of a prey, searching for an updated position globally when the escape energy is greater than or equal to 1, and searching for a local position locally when the escape energy is less than 1; in the local exploration process, when the escaping energy of the prey is more than or equal to 0.5, harris hawks consume the energy of the prey through soft attack, when the escaping energy is less than 0.5, harris hawks directly capture the prey through hard attack, and when the prey cannot escape, a team fast dive strategy is adopted for attacking;
step 5.4: calculating the fitness of the individuals after the position is updated by the Harris eagle predation strategy, and comparing and updating the fitness and the position of the optimal individual;
step 5.5: and judging whether the end condition of the optimization is met, namely whether the set maximum iteration number is reached or whether the fitness meets the requirement, if so, outputting the optimized fitness and the hyperparameter corresponding to the optimal population position, and if not, returning to the step 5.3 to continue the optimization.
The method comprises the steps of monitoring and collecting bearing vibration signals, extracting time domain characteristics of the vibration signals, selecting through similarity measurement, then adopting variational modal decomposition and combining an energy entropy discrimination method to select modal component characteristics, and combining a plurality of characteristic vectors F i =[F 1 (i),F 2 (i),...,F n (i)]To construct a feature matrix F = [ F = 1 ,F 2 ,...,F k ]Dividing training set, verifying set and then training setInputting a Harris eagle optimized LSTM network for model training, taking the mean square error of a verification set data prediction result as a fitness function self-adaptive optimization searching super parameter, and establishing an LSTM prediction model to predict the bearing RUL by using the iteratively updated optimal super parameter. Taking the residual service life of the rolling bearing as an example, the load of a test bench for testing and collecting the vibration signal of the bearing is 12kN, the rotating speed is 2100rpm, the sampling frequency is 25.6kHz, the sampling interval is 1min, and the collecting time is 1.28s each time; and predicting according to the following steps;
1) Monitoring and acquiring a bearing vibration signal in real time, sampling once every 1min, sampling for 1.28s at 25.6kHz every time to obtain 32768 data points, and sampling 123 groups of samples in total until the bearing fails;
2) Carrying out time domain analysis on the collected bearing vibration signals to obtain 14 time domain characteristics, and selecting a characteristic type with smaller similarity to represent a degradation state according to the similarity among the characteristics;
3) Carrying out variation modal decomposition on the collected bearing vibration signal to obtain 5 modal components, and selecting the modal components with the energy as large as possible and the energy entropy as small as possible to represent a degradation state according to an energy entropy discrimination method;
4) Respectively carrying out normalization, abnormal value elimination and interpolation completion on each obtained feature vector, constructing a feature matrix, and dividing the first 90 samples into a training set and the 91 st to 120 th samples into a verification set;
5) The training set data is used as the input of the LSTM network, the mean square error predicted by the verification set data is used as a fitness function of Harris eagle optimization, the number of population is set to be 10, the maximum optimizing frequency is set to be 10, the population position corresponds to the hyperparameters such as learning rate, batch size and iteration period, the optimal hyperparameter combination is adaptively optimized while the LSTM network is trained, and the prediction network is established by the hyperparameters after optimizing iteration is completed. The fitness curve in the optimization super-parameter process of the Harris eagle optimization algorithm is shown in figure 2, so that the method has the advantages of strong global search capability and high convergence speed, and the fitness value can be converged after four times of optimization.
TABLE 2 out of parameter ranges
6) And for characteristic sample data, the self-adaptive HHO optimized super-parameter long-short term memory network is used for predicting the residual service life of the bearing, compared with LSTM and original LSTM network models under optimization of other algorithms, the prediction precision is improved, the Mean Square Error (MSE) index is reduced by 27.3% -62.8%, and the Mean Absolute Error (MAE) index is reduced by 19.0% -40.9%. In contrast, the prediction performance of the method is obviously improved. The HHO optimized LSTM network is plotted against the original network prediction results in fig. 3.
The foregoing is only a preferred embodiment of the present invention, and it should be noted that those skilled in the art can make various improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be construed as the protection scope of the present invention.
Claims (5)
1. A method for predicting the residual service life of an urban rail train bearing is characterized by comprising the following steps: the method comprises the following steps:
step 1: setting the rotating speed and sampling frequency of the bearing, and acquiring the original vibration signal of the bearing in the full life cycle when the bearing runs to failure;
and 2, step: extracting time domain characteristics of an original vibration signal of the bearing, measuring the similarity degree between the characteristics by adopting a similarity measurement method, removing the characteristics with lower discrimination degree, and selecting the characteristic parameters which represent the optimal degradation capability of the bearing;
step 3, carrying out variation modal decomposition on the original vibration signal of the bearing to obtain modal component characteristics, selecting modal components according to the principle that the larger the energy entropy is, the larger the information uncertainty is, by adopting an energy entropy discrimination method and by using the standard of the highest energy and the smallest energy entropy, expanding frequency domain information on the basis of time domain characteristics, and respectively calculating the energy and the energy entropy of the modal components;
and 4, step 4: combining the time domain characteristics obtained in the step 2 with the modal component characteristics obtained in the step 3 to obtain k characteristic vectors with the sequence length of n and the ith characteristic vectorIs F i =[F 1 (i),F 2 (i),...,F n (i)]In which F is n (i) Representing the eigenvalues corresponding to the nth sampling time point, and then combining the k eigenvectors into an eigenvector matrix F = [ F ] 1 ,F 2 ,...,F k ](ii) a Removing abnormal values in each feature vector, normalizing the feature vectors to be between 0 and 1 after completing the abnormal values through cubic spline interpolation, and dividing a feature matrix into a training set and a verification set;
and 5: adopting an LSTM network as an RUL prediction main network, setting the super-parameter of the LSTM network as a population position, training the LSTM network by training set data, using the mean square error of a verification set data prediction result as a fitness function, optimizing the LSTM super-parameter by adopting a Harris eagle optimization algorithm, and iteratively updating an optimal super-parameter combination each time the LSTM network is trained; and establishing an LSTM prediction network by using the optimized hyper-parameters, inputting the characteristic matrix into the LSTM network, updating the weight through a back propagation algorithm, updating the gradient through an Adam optimizer, and outputting the RUL prediction value of the bearing after calculation.
2. The method for predicting the residual service life of the urban rail train bearing according to claim 1, characterized by comprising the following steps: the energy in the energy entropy discrimination method in the step 3 satisfies the following expression:
wherein u is i For the ith modal component sequence, t = {0,1,. And n }, where n is the sequence length;
the energy entropy satisfies:
H i =-p i log 10 p i , (2);
wherein p is i Is the ratio of the energy of the ith modal component to the total energy,
where i = {0,1,. K }, where K is the total modal component number.
3. The method for predicting the residual service life of the urban rail train bearing according to claim 1, characterized by comprising the following steps: and 4, selecting variation modal decomposition modal components by a time domain feature extraction and energy entropy discrimination method to construct a feature matrix in the step 4, wherein the constructed feature matrix contains time domain, frequency domain and entropy domain feature information of bearing degradation.
4. The method for predicting the residual service life of the urban rail train bearing according to claim 1, characterized by comprising the following steps: in the step 5, the specific process of optimizing the super-parameters of the LSTM network by adopting the Harris eagle optimization algorithm comprises the following steps:
step 5.1: selecting the hyper-parameter learning rate, the batch size, the iteration period, the LSTM layer unit number and the Dense layer unit number of the LSTM network as the group positions, and setting the group number and the maximum iteration times;
step 5.2: initializing Harris eagle populations and positions and energies of prey, training an LSTM network by training set data, taking the mean square error of a data prediction result of a verification set as a fitness function, and comparing and updating the fitness and the positions of the optimal individuals;
step 5.3: calculating escape energy of the prey, performing global search to update the position when the escape energy is greater than or equal to 1, and performing local exploration to update the position when the escape energy is less than 1; in the local exploration process, when the escaping energy of the prey is more than or equal to 0.5, harris hawks consume the energy of the prey through soft attack, when the escaping energy is less than 0.5, harris hawks directly capture the prey through hard attack, and when the prey cannot escape, a team fast dive strategy is adopted for attacking;
step 5.4: calculating the fitness of the individual after the position is updated by the Harris eagle predation strategy, and comparing and updating the fitness and the position of the optimal individual;
step 5.5: and judging whether the end condition of the optimization is met, namely whether the set maximum iteration number is reached or whether the fitness meets the requirement, if so, outputting the optimized fitness and the hyperparameter corresponding to the optimal population position, and if not, returning to the step 5.3 to continue the optimization.
5. The method for predicting the residual service life of the urban rail train bearing according to claim 4, characterized by comprising the following steps: the LSTM network adaptively iterates and updates the optimal hyper-parameters through a Harris eagle optimization algorithm in the training process, and the optimal LSTM network is directly constructed according to the optimal hyper-parameters without retraining.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210992969.6A CN115292820A (en) | 2022-08-18 | 2022-08-18 | Method for predicting residual service life of urban rail train bearing |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210992969.6A CN115292820A (en) | 2022-08-18 | 2022-08-18 | Method for predicting residual service life of urban rail train bearing |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115292820A true CN115292820A (en) | 2022-11-04 |
Family
ID=83830771
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210992969.6A Pending CN115292820A (en) | 2022-08-18 | 2022-08-18 | Method for predicting residual service life of urban rail train bearing |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115292820A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116502049A (en) * | 2023-06-25 | 2023-07-28 | 山东科技大学 | Rolling bearing residual service life prediction method, system, equipment and storage medium |
-
2022
- 2022-08-18 CN CN202210992969.6A patent/CN115292820A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116502049A (en) * | 2023-06-25 | 2023-07-28 | 山东科技大学 | Rolling bearing residual service life prediction method, system, equipment and storage medium |
CN116502049B (en) * | 2023-06-25 | 2023-09-08 | 山东科技大学 | Rolling bearing residual service life prediction method, system, equipment and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110378052B (en) | Equipment residual life prediction method considering future working conditions based on cyclic neural network | |
CN109492193B (en) | Abnormal network data generation and prediction method based on deep machine learning model | |
CN108984893B (en) | Gradient lifting method-based trend prediction method | |
Kwan et al. | A novel approach to fault diagnostics and prognostics | |
CN110598851A (en) | Time series data abnormity detection method fusing LSTM and GAN | |
CN111220387B (en) | Vehicle bearing residual life prediction method based on multi-feature-quantity correlation vector machine | |
CN112629863A (en) | Bearing fault diagnosis method for dynamic joint distribution alignment network under variable working conditions | |
CN112328588B (en) | Industrial fault diagnosis unbalanced time sequence data expansion method | |
CN115034248A (en) | Automatic diagnostic method, system and storage medium for equipment | |
Zhao et al. | Bearing health condition prediction using deep belief network | |
CN112179691A (en) | Mechanical equipment running state abnormity detection system and method based on counterstudy strategy | |
CN108961460B (en) | Fault prediction method and device based on sparse ESGP (Enterprise service gateway) and multi-objective optimization | |
CN115187832A (en) | Energy system fault diagnosis method based on deep learning and gram angular field image | |
CN114048688A (en) | Method for predicting service life of bearing of wind power generator | |
CN109145373B (en) | Residual life prediction method and device based on improved ESGP and prediction interval | |
CN113792758A (en) | Rolling bearing fault diagnosis method based on self-supervision learning and clustering | |
CN114091349A (en) | Multi-source field self-adaption based rolling bearing service life prediction method | |
CN112529053A (en) | Short-term prediction method and system for time sequence data in server | |
CN115292820A (en) | Method for predicting residual service life of urban rail train bearing | |
CN115422687A (en) | Service life prediction method of rolling bearing | |
CN116702076A (en) | Small sample migration learning fault diagnosis method, system, computer and storage medium based on CNN feature fusion | |
Sadoughi et al. | A deep learning approach for failure prognostics of rolling element bearings | |
CN114881087A (en) | Building robot bearing performance degradation assessment method | |
CN111400964B (en) | Fault occurrence time prediction method and device | |
CN113554229A (en) | Three-phase voltage unbalance abnormality detection method and device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |