CN110705181B - Rolling bearing residual life prediction method based on convolution length-time memory cyclic neural network - Google Patents
Rolling bearing residual life prediction method based on convolution length-time memory cyclic neural network Download PDFInfo
- Publication number
- CN110705181B CN110705181B CN201910970482.6A CN201910970482A CN110705181B CN 110705181 B CN110705181 B CN 110705181B CN 201910970482 A CN201910970482 A CN 201910970482A CN 110705181 B CN110705181 B CN 110705181B
- Authority
- CN
- China
- Prior art keywords
- convolution
- neural network
- rolling bearing
- encoder
- cyclic neural
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000005096 rolling process Methods 0.000 title claims abstract description 38
- 125000004122 cyclic group Chemical group 0.000 title claims abstract description 34
- 230000015654 memory Effects 0.000 title claims abstract description 31
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 28
- 230000036541 health Effects 0.000 claims abstract description 36
- 238000011156 evaluation Methods 0.000 claims abstract description 20
- 238000012549 training Methods 0.000 claims abstract description 15
- 238000003062 neural network model Methods 0.000 claims abstract description 11
- 230000001133 acceleration Effects 0.000 claims abstract description 4
- 230000006870 function Effects 0.000 claims description 17
- 230000008569 process Effects 0.000 claims description 13
- 230000007423 decrease Effects 0.000 claims description 8
- 238000011176 pooling Methods 0.000 claims description 5
- 230000003213 activating effect Effects 0.000 claims description 3
- 238000013527 convolutional neural network Methods 0.000 claims 1
- 238000006731 degradation reaction Methods 0.000 abstract description 13
- 230000015556 catabolic process Effects 0.000 abstract description 11
- 238000012360 testing method Methods 0.000 description 5
- 230000007787 long-term memory Effects 0.000 description 4
- 230000000306 recurrent effect Effects 0.000 description 4
- 238000005070 sampling Methods 0.000 description 4
- 230000006403 short-term memory Effects 0.000 description 4
- 238000000605 extraction Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000005562 fading Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000007634 remodeling Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention provides a method for predicting the residual life of a rolling bearing based on a convolution length time memory cyclic neural network, which comprises the following steps: acquiring a vibration acceleration signal of a rolling bearing as a data sample; establishing a convolution self-encoder, taking a data sample as input, and training to obtain a multilayer convolution self-encoder; acquiring characteristic data of the data sample according to the data sample and the multilayer convolution self-encoder; determining health evaluation indexes of the rolling bearing according to the characteristic data of the data sample by using a fuzzy C-means algorithm; establishing a convolution long-time and short-time memory cyclic neural network model, taking a bearing data sample in a decay period and a corresponding health evaluation index as the input of the convolution long-time and short-time memory cyclic neural network, and training to obtain a prediction model; the method reduces the dependence on manual experience and professional knowledge, effectively reduces the data volume of the input long-time memory cyclic neural network, simplifies the operation, and comprehensively and effectively reflects the degradation state of the bearing by the prediction model.
Description
Technical Field
The invention relates to the technical field of life prediction of rolling bearings, in particular to a residual life prediction method of a rolling bearing based on a convolution length-time memory cyclic neural network.
Background
As an important component of modern manufacturing, bearings play a crucial role in the reliable operation of most rotating machines, directly affecting the health of the machine system. In the operation process of mechanical equipment, the rolling bearing is damaged by the change and uncertainty of complex working conditions, and the precision and reliability of the whole mechanical equipment system are directly influenced. By effectively monitoring the working state of the rolling bearing and accurately predicting the degradation trend of the rolling bearing, the safe and reliable operation of mechanical equipment can be ensured, and the major economic loss is avoided. Therefore, it is necessary to monitor the health of the bearings and predict their degradation tendency.
In the existing bearing degradation trend prediction model, the bearing degradation trend is reflected by characteristic indexes such as root mean square values, kurtosis values, peak values, fault characteristic frequencies, wavelet entropies, empirical mode decomposition entropies and the like, and the characteristic indexes can be used as degradation index quantities of the degradation characteristics of the bearing. However, these characteristic quantities are sensitive only to a certain stage of the bearing degradation process, and the predicted result error is large when a certain characteristic quantity is used alone as the degradation characteristic quantity of the bearing. A comprehensive index of various characteristics can describe the bearing degradation process relatively accurately, but adds computational complexity. In addition, in most of the existing prediction models, the extraction of features still requires expert experience and manual extraction.
Therefore, it is necessary to provide a new method for predicting the remaining life of the rolling bearing.
Disclosure of Invention
In view of this, the present invention provides a method for predicting the remaining life of a rolling bearing based on a convolution long-term and short-term memory cyclic neural network, which adopts a convolution automatic encoder to extract the characteristics of original data, combines the long-term and short-term memory cyclic neural network with the encoding process of the convolution automatic encoder, reduces the dependence on human experience and professional knowledge, simultaneously effectively reduces the data amount of the input long-term and short-term memory cyclic neural network, and simplifies the operation. In addition, a fuzzy C-means algorithm is adopted, a sectional type prediction model is realized by presetting a health assessment index threshold value, the running characteristics of the bearing are fully considered, and the prediction model comprehensively and effectively reflects the recession state of the bearing.
The invention provides a method for predicting the residual life of a rolling bearing based on a convolution length time memory cyclic neural network, which comprises the following steps:
acquiring a vibration acceleration signal of a rolling bearing as a data sample;
establishing a convolution self-encoder, taking a data sample as input, and training to obtain a multilayer convolution self-encoder;
acquiring characteristic data of the data sample according to the data sample and the multilayer convolution self-encoder;
determining health evaluation indexes of the rolling bearing according to the characteristic data of the data sample by using a fuzzy C-means algorithm;
and establishing a convolution long-time and short-time memory cyclic neural network model, taking a bearing original data sample in a decay period and a corresponding health evaluation index as the input of the convolution long-time and short-time memory cyclic neural network, and training to obtain a prediction model.
Further, the convolutional self-encoder comprises an input layer, a convolutional layer, a pooling layer, a full-link layer, a reconstruction layer and an output layer;
the convolutional autocoder is a one-dimensional convolution.
Further, the determining the health assessment index of the rolling bearing according to the characteristic data of the data sample by using the fuzzy C-means algorithm specifically comprises:
and taking the characteristic data as the input of a fuzzy C-means algorithm, taking the characteristic data of the normal operation of the bearing as a clustering center, and calculating the membership value of each section of the characteristic data to the clustering center to obtain the health assessment index.
Further, before the convolutional length time is established and the cyclic neural network model is memorized, the method further comprises the following steps: and determining a threshold value of the health evaluation index for dividing the normal period and the decline period of the rolling bearing.
Further, the determining the threshold of the health assessment index for dividing the normal period and the decline period of the rolling bearing specifically includes: the minimum health evaluation index of the rolling bearing in a period of normal operation is used as a threshold value.
Further, the convolution layer of the recurrent neural network model and the coding process of the multilayer convolution self-coder are memorized in the same structure and parameters in the long and short convolution periods.
Further, the reconstruction method adopted in the process of establishing the convolution self-encoder and training the data sample as input to obtain the multilayer convolution self-encoder is bilinear interpolation.
Further, the activation function of the convolutional auto-encoder is an LRule function.
The invention has the beneficial effects that:
1. the invention adopts a convolution automatic encoder to compress the bearing original data and extract effective characteristics without expert experience and manual extraction methods;
2. according to the method, a fuzzy C-means algorithm is adopted, the membership degree of each section of characteristics to a normal-period clustering center is used as a health evaluation index, and the recession state of the bearing can be fully, comprehensively and effectively reflected;
3. according to the method, a sectional type prediction model is adopted, a health assessment index threshold value is preset, the whole life cycle of the bearing is divided into a normal period and a decline period, and the running characteristics of the bearing are fully considered, so that the prediction is more accurate;
4. the invention adopts the convolution long-time and short-time memory cyclic neural network, the parameter structure of the convolution layer is consistent with the parameter structure of the encoding process of the convolution automatic encoder, the operation is simplified, the characteristics can be effectively extracted, the bearing data sample and the corresponding health evaluation index are used as the input of the convolution long-time and short-time memory cyclic neural network for bearing degradation trend prediction, and the advantage of the long-time and short-time memory cyclic neural network for processing the sequence data is fully utilized.
Drawings
The invention is further described below with reference to the following figures and examples:
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a graph illustrating health assessment indicator trends for life cycle data in an example;
FIG. 3 is a comparison graph of the actual health assessment indicator and the prediction result of the convolutional long-and-short memory recurrent neural network in the embodiment.
Detailed Description
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
As shown in fig. 1, the method for predicting the remaining life of a rolling bearing based on a convolution duration memory cyclic neural network provided by the invention comprises the following steps:
and acquiring a vibration acceleration signal of the rolling bearing as a data sample.
Establishing a convolution self-encoder, taking a data sample as input, and training to obtain a multilayer convolution self-encoder; in this embodiment, the training iteration number of the convolutional auto-encoder is 20, the batch size is 50, the learning rate is 0.001, and batch normalization can be integrated into each layer of the auto-encoder, so that the model can be rapidly converged in the training process.
Acquiring characteristic data of the data sample according to the data sample and the multilayer convolution self-encoder;
determining health evaluation indexes of the rolling bearing according to the characteristic data of the data sample by using a fuzzy C-means algorithm;
establishing a convolution long-time memory cyclic neural network model, taking a decay period bearing original data sample and a corresponding health evaluation index as the input of the convolution long-time memory cyclic neural network, calculating through a forgetting gate, an input gate and an output gate, and training to obtain a prediction model. The calculation of the long-term and short-term memory recurrent neural network can adopt the method disclosed in the prior art, and is not described herein. By the method, the characteristics of the original data are extracted by the convolution automatic encoder, and the coding processes of the long-time memory cyclic neural network and the short-time memory cyclic neural network are combined, so that the dependence on manual experience and professional knowledge is reduced, meanwhile, the data volume of the input long-time memory cyclic neural network is effectively reduced, and the operation is simplified. In addition, a fuzzy C-means algorithm is adopted, a health assessment index threshold value is preset, a sectional type prediction model is realized, the running characteristics of the bearing are fully considered, and the prediction model comprehensively and effectively reflects the recession state of the bearing.
Further, the convolutional self-encoder comprises an input layer, a convolutional layer, a pooling layer, a full-link layer, a reconstruction layer and an output layer; in this embodiment, the convolution algorithm is integrated into the self-coding algorithm to extract local features of complex bearing signals, and a convolution self-encoder with 19 layers in total is established, including 1 input layer, 9 convolution layers, 4 pooling layers, 1 full-link layer, 4 reconstruction layers, 4 output layers, and 1 output layer.
The convolutional autocoder is a one-dimensional convolution. In this embodiment, the convolutional auto-encoder improves the conventional two-dimensional convolution into one-dimensional convolution to adapt to one-dimensional bearing signals, and the number of neurons in each layer is 4096 × 1, 4096 × 16, 1024 × 16, 1024 × 32, 256 × 32, 256 × 64, 64 × 64, 64 × 64, 16 × 64, 32 × 1, 16 × 64, 64 × 64, 256 × 64, 256 × 32, 1024 × 32, 1024 × 16, 4096 × 16, and 4096 × 1.
Further, the determining the health assessment index of the rolling bearing according to the characteristic data of the data sample by using the fuzzy C-means algorithm specifically comprises:
s101: and (3) taking the characteristic data as the input of a fuzzy C-means algorithm, taking the characteristic data of the normal operation of the bearing as a clustering center, and calculating the membership value of each section of characteristic data to the clustering center to obtain a health evaluation index. In this embodiment, the specific process is as follows:
s102: let xi(i 1, 2.. n.) is the set of feature samples generated from the encoder, c is a predetermined number of classes μj(xi) Is the membership function of the ith sample to the jth class. Writing with a clustering loss function defined by a membership function:
wherein, b>1 is a constant that can control the degree of blurring of the clustering results, mjIs the class j center.
S103: the number of clusters c and the weighting index b are set. In this example, c is 1, and empirically, b is most preferably 2.
S104: initializing each cluster center, selecting the early normal operation data of the rolling bearing as a normal-period cluster center, and selecting the early normal operation data as a decline-period cluster center when the experiment is finished. Considering the condition that the collected signals are not stable enough when the test is started and the test is ended, the data of the second sampling period after the start is selected as the initial normal period clustering center, and the last sampling period before the end is used as the initial decay period clustering center.
S105: the following operations were repeated until the membership of each sample was stable:
calculating a membership function in the current cluster according to:
and (3) updating and calculating various clustering centers by using the current membership function according to the following formula:
when the fuzzy C-means algorithm is converged, various clustering centers and membership values of various samples to various categories are obtained, so that fuzzy clustering division is completed, and the membership of each sample to a normal-period clustering center is used as a health assessment index.
Further, before the convolutional length time is established and the cyclic neural network model is memorized, the method further comprises the following steps: and determining a threshold value of the health evaluation index for dividing the normal period and the decline period of the rolling bearing. The determining of the threshold value of the health assessment index for dividing the normal period and the decline period of the rolling bearing specifically comprises the following steps: the minimum health evaluation index of the rolling bearing in a period of normal operation is used as a threshold value. Meanwhile, considering the instability at the end of the test, the membership value is set to be reduced to the minimum value, and the value is taken as the end point of the fading period when the value approaches 0. And considering that the bearing enters a degradation period when the continuous 10 values are lower than the threshold value, and starting to establish the operation of the memory cycle neural network model with long convolution time.
Further, the convolution layer of the recurrent neural network model and the coding process of the multilayer convolution self-coder are memorized in the same structure and parameters in the long and short convolution periods. The convolution layer has a uniform structure, the effectiveness of the characteristic vector in the prediction model is guaranteed, the deviation of the prediction result is avoided, and meanwhile, the operation is simplified.
Further, in the process of establishing the convolutional auto-encoder, training a data sample as input to obtain a multilayer convolutional auto-encoder, by combining the characteristics of the maximum pooling layer adopted in the encoding process and the continuous change of the bearing signals, the adopted reconstruction method is bilinear interpolation, and the remodeling capability of the decoding process on the bearing signals is enhanced.
Further, the activating function of the convolution self-encoder is an LRule function, the negative semiaxis slope of the LRule function is 0.1, the positive semiaxis slope of the LRule function is 0.5, and the activating function solves the problem that the negative axis feature of the vibration signal is completely eliminated by a traditional ReLU function.
Further, the method is detailed by measuring data through a deep groove ball bearing 6212 as a specific embodiment, wherein the axial load is zero, the radial load is 20kN, the rotation speed is 2000rad/min, the time interval of sample collection is 15 minutes, each time collection is 1 minute, and the sampling frequency is 10000 Hz.
Acquiring the full life cycle data of the rolling bearing, preprocessing the data to obtain sample data, and segmenting the data, wherein the length of each segment of data is 4096, and 50000 data samples are obtained.
Establishing a convolution self-encoder, taking all data samples as input, acquiring characteristic data of the sample data, wherein the reconstruction method of the self-encoder is bilinear interpolation, an activation function is an LRule function, the negative half-axis slope of the self-encoder is 0.1, the positive half-axis slope of the self-encoder is 0.5, the training iteration number is 20, the batch size is 50, and the learning rate is 0.001.
Considering the situation that the acquired signals are not stable enough when the test is started and ended, the data of the second adopted period after the start is selected as the initial normal period clustering center, the penultimate period before the end is used as the initial decay period clustering center, the characteristic data is used as the input of the fuzzy C mean value algorithm to obtain the clustering centers of the final normal period and decay period data, and the membership value of each section of characteristic to the normal period clustering center is used as the health assessment index, as shown in the following figure 2.
Setting a health evaluation index threshold, defining the minimum membership value of the data of the second sampling period after the start as a threshold, setting 10 continuous membership values smaller than the threshold as a decay period starting point, and setting the membership value to be reduced to the minimum value in consideration of instability at the end of the test, and taking the value as a decay period end point when the value approaches 0. In this embodiment, the decay period starting point is 0.99876, and the decay period ending point is 0.00966. And (4) after the bearing enters the decline period, entering the next step, namely, establishing a convolution length and memory cyclic neural network model.
And constructing a long-time and short-time convolution memory cyclic neural network, taking the bearing data sample in the decay period and the corresponding health evaluation index as the input of the long-time and short-time convolution memory cyclic neural network, training a prediction model, and finally realizing the prediction of the degradation trend of the bearing.
Finally, although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that various changes and modifications may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (4)
1. The method for predicting the residual life of the rolling bearing based on the convolution length time memory cyclic neural network is characterized by comprising the following steps: the method comprises the following steps:
acquiring a vibration acceleration signal of a rolling bearing as a data sample;
establishing a convolution self-encoder, taking a data sample as input, and training to obtain a multilayer convolution self-encoder;
acquiring characteristic data of the data sample according to the data sample and the multilayer convolution self-encoder;
determining health evaluation indexes of the rolling bearing according to characteristic data of the data sample by using a fuzzy C-means algorithm;
establishing a convolution long-time and short-time memory cyclic neural network model, taking a bearing data sample in a decay period and a corresponding health evaluation index as the input of the convolution long-time and short-time memory cyclic neural network, and training to obtain a prediction model;
the method for determining the health evaluation index of the rolling bearing according to the characteristic data of the data sample by using the fuzzy C-means algorithm specifically comprises the following steps:
taking the characteristic data as the input of a fuzzy C-means algorithm, taking the characteristic data of the normal operation of the bearing as a clustering center, and calculating the membership value of each section of the characteristic data to the clustering center to obtain a health assessment index;
before establishing a convolution duration memory cyclic neural network model, the method further comprises the following steps: determining a threshold value of a health evaluation index for dividing a normal period and a decline period of the rolling bearing;
the determining of the threshold value of the health assessment index for dividing the normal period and the decline period of the rolling bearing specifically comprises the following steps: taking the minimum health evaluation index of the rolling bearing in a period of normal operation as a threshold value;
the convolution layer of the convolutional neural network model and the coding process of the multilayer convolutional self-coder are memorized in the same structure and parameters in the long-time convolution period.
2. The method for predicting the residual life of the rolling bearing based on the convolution duration memory cyclic neural network as claimed in claim 1, wherein the method comprises the following steps: the convolution self-encoder comprises an input layer, a convolution layer, a pooling layer, a full-link layer, a reconstruction layer and an output layer;
the convolutional autocoder is a one-dimensional convolution.
3. The method for predicting the residual life of the rolling bearing based on the convolution duration memory cyclic neural network as claimed in claim 1, wherein the method comprises the following steps: the method comprises the steps of establishing a convolution self-encoder, taking a data sample as input, and training to obtain a multilayer convolution self-encoder, wherein a reconstruction method adopted is bilinear interpolation.
4. The method for predicting the residual life of the rolling bearing based on the convolution duration memory cyclic neural network as claimed in claim 1, wherein the method comprises the following steps: the activating function of the convolution self-encoder is an LRule function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910970482.6A CN110705181B (en) | 2019-10-13 | 2019-10-13 | Rolling bearing residual life prediction method based on convolution length-time memory cyclic neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910970482.6A CN110705181B (en) | 2019-10-13 | 2019-10-13 | Rolling bearing residual life prediction method based on convolution length-time memory cyclic neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110705181A CN110705181A (en) | 2020-01-17 |
| CN110705181B true CN110705181B (en) | 2022-06-24 |
Family
ID=69199574
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910970482.6A Active CN110705181B (en) | 2019-10-13 | 2019-10-13 | Rolling bearing residual life prediction method based on convolution length-time memory cyclic neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110705181B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111581892B (en) * | 2020-05-29 | 2024-02-13 | 重庆大学 | Bearing residual life prediction method based on GDAU neural network |
| CN111595583B (en) * | 2020-05-29 | 2022-02-01 | 重庆大学 | GAU neural network-based bearing residual life prediction method |
| CN112069724B (en) * | 2020-07-21 | 2023-06-20 | 上海宇航系统工程研究所 | Rocket health degree assessment method based on long-short-time memory self-encoder |
| CN111985155B (en) * | 2020-08-10 | 2023-08-04 | 武汉大学 | Circuit health state prediction method and system based on integrated deep neural network |
| CN112347898B (en) * | 2020-11-03 | 2024-04-09 | 重庆大学 | Rolling bearing health index construction method based on DCAE neural network |
| CN112729825A (en) * | 2020-12-18 | 2021-04-30 | 中车永济电机有限公司 | Method for constructing bearing fault diagnosis model based on convolution cyclic neural network |
| CN112766219A (en) * | 2021-01-31 | 2021-05-07 | 石家庄铁道大学 | Bearing health assessment method and device based on depth separable convolution self-encoder |
| CN113255209B (en) * | 2021-04-27 | 2022-06-07 | 燕山大学 | Method for predicting residual life of bearing of gearbox |
| CN113221277B (en) * | 2021-05-14 | 2022-12-09 | 西安交通大学 | Bearing performance degradation evaluation method and system based on digital twin model |
| CN115096592A (en) * | 2022-06-27 | 2022-09-23 | 重庆交通大学 | Rolling bearing fault trend prediction method based on PSO-LSTM |
| CN116257747B (en) * | 2022-12-22 | 2024-08-02 | 中国人民解放军海军工程大学 | Early abnormality detection method for rotor system of fusion kurtosis and convolution self-encoder |
| CN116990021B (en) * | 2023-09-22 | 2024-01-02 | 万向钱潮股份公司 | Fatigue life assessment method and device for hub bearing |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102902859A (en) * | 2012-10-09 | 2013-01-30 | 西安交通大学 | Hybrid sliding bearing fluid-solid coupling design method based on phase change model |
| CN104751229A (en) * | 2015-04-13 | 2015-07-01 | 辽宁大学 | Bearing fault diagnosis method capable of recovering missing data of back propagation neural network estimation values |
| CN106885697A (en) * | 2017-03-17 | 2017-06-23 | 华东交通大学 | The performance degradation assessment method of the rolling bearing based on FCM HMM |
| CN106991200A (en) * | 2016-01-20 | 2017-07-28 | 沪东重机有限公司 | A kind of sliding bearing end cap installs gap width design method |
| CN107036817A (en) * | 2017-04-05 | 2017-08-11 | 哈尔滨理工大学 | SVR rolling bearing performances decline Forecasting Methodology based on krill group's algorithm |
| CN109460618A (en) * | 2018-11-13 | 2019-03-12 | 华中科技大学 | A kind of rolling bearing remaining life on-line prediction method and system |
| CN109726524A (en) * | 2019-03-01 | 2019-05-07 | 哈尔滨理工大学 | A kind of rolling bearing remaining life prediction technique based on CNN and LSTM |
| CN109883699A (en) * | 2018-12-20 | 2019-06-14 | 上海理工大学 | A kind of rolling bearing method for predicting residual useful life based on long memory network in short-term |
| CN109902399A (en) * | 2019-03-01 | 2019-06-18 | 哈尔滨理工大学 | Rolling bearing fault recognition methods under a kind of variable working condition based on ATT-CNN |
| CN110232249A (en) * | 2019-06-17 | 2019-09-13 | 中国人民解放军陆军装甲兵学院 | A kind of rolling bearing method for predicting residual useful life |
| CN110261109A (en) * | 2019-04-28 | 2019-09-20 | 洛阳中科晶上智能装备科技有限公司 | A kind of Fault Diagnosis of Roller Bearings based on bidirectional memory Recognition with Recurrent Neural Network |
-
2019
- 2019-10-13 CN CN201910970482.6A patent/CN110705181B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102902859A (en) * | 2012-10-09 | 2013-01-30 | 西安交通大学 | Hybrid sliding bearing fluid-solid coupling design method based on phase change model |
| CN104751229A (en) * | 2015-04-13 | 2015-07-01 | 辽宁大学 | Bearing fault diagnosis method capable of recovering missing data of back propagation neural network estimation values |
| CN106991200A (en) * | 2016-01-20 | 2017-07-28 | 沪东重机有限公司 | A kind of sliding bearing end cap installs gap width design method |
| CN106885697A (en) * | 2017-03-17 | 2017-06-23 | 华东交通大学 | The performance degradation assessment method of the rolling bearing based on FCM HMM |
| CN107036817A (en) * | 2017-04-05 | 2017-08-11 | 哈尔滨理工大学 | SVR rolling bearing performances decline Forecasting Methodology based on krill group's algorithm |
| CN109460618A (en) * | 2018-11-13 | 2019-03-12 | 华中科技大学 | A kind of rolling bearing remaining life on-line prediction method and system |
| CN109883699A (en) * | 2018-12-20 | 2019-06-14 | 上海理工大学 | A kind of rolling bearing method for predicting residual useful life based on long memory network in short-term |
| CN109726524A (en) * | 2019-03-01 | 2019-05-07 | 哈尔滨理工大学 | A kind of rolling bearing remaining life prediction technique based on CNN and LSTM |
| CN109902399A (en) * | 2019-03-01 | 2019-06-18 | 哈尔滨理工大学 | Rolling bearing fault recognition methods under a kind of variable working condition based on ATT-CNN |
| CN110261109A (en) * | 2019-04-28 | 2019-09-20 | 洛阳中科晶上智能装备科技有限公司 | A kind of Fault Diagnosis of Roller Bearings based on bidirectional memory Recognition with Recurrent Neural Network |
| CN110232249A (en) * | 2019-06-17 | 2019-09-13 | 中国人民解放军陆军装甲兵学院 | A kind of rolling bearing method for predicting residual useful life |
Non-Patent Citations (8)
| Title |
|---|
| Learning to Monitor Machine Health with Convolutional Bi-Directional LSTM Networks;Rui Zhao等;《sensors》;20170130;第17卷;第1-18页 * |
| Predicting Remaining Useful Life of Rolling Bearings Based on Deep Feature Representation and Transfer Learning;Wentao Mao等;《IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT》;20190520;第1594-1608页 * |
| Remaining life prognostics of rolling bearing based on relative features and multivariable support vector machine;Xuefeng chen等;《Proceedings of the Institution of Mechanical Engineers， Part C: Journal of Mechanical Engineering Science》;20130111;第227卷(第12期);第2849-2860页 * |
| Remaining useful life estimation of engineered systems using vanilla LSTM neural networks;Yuting Wu等;《Neurocomputing》;20170531;第275卷;第167-179页 * |
| Rolling Bearing Performance Degradation Assessment Based on Convolutional Sparse Combination Learning;WENYI HUANG等;《IEEE ACCESS》;20190214;第7卷;第17834-17846页 * |
| 基于多健康状态评估的滚动轴承剩余寿命预测方法研究;张龙龙;《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅱ辑》;20160315(第3期);C029-96 * |
| 基于模糊C-均值的空间机构轴系健康评估方法;刘红星等;《振动、测试与诊断》;20180615;第38卷(第3期);第559-563页 * |
| 基于状态振动特征的空间滚动轴承可靠性评估方法研究;陈昌;《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅱ辑》;20150115(第1期);C029-107 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110705181A (en) | 2020-01-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110705181B (en) | Rolling bearing residual life prediction method based on convolution length-time memory cyclic neural network | |
| CN109978134B (en) | Fault prediction method based on fast integrated convolutional neural network | |
| CN112070322B (en) | High-voltage cable line running state prediction method based on long-short term memory network | |
| CN112414694B (en) | Equipment multistage abnormal state identification method and device based on multivariate state estimation technology | |
| CN115903741B (en) | Industrial control system data anomaly detection method | |
| CN110348468B (en) | Bearing sub-health identification method for optimized strong reconstruction edge noise reduction automatic encoder of correlation vector machine | |
| CN114839540B (en) | Lithium ion battery residual life online prediction method based on deep learning algorithm | |
| CN114741945B (en) | On-line fault diagnosis method for aero-engine | |
| CN113158814A (en) | Bearing health state monitoring method based on convolution self-encoder | |
| CN115017945A (en) | Mechanical fault diagnosis method and system based on enhanced convolutional neural network | |
| CN113627594B (en) | One-dimensional time sequence data augmentation method based on WGAN | |
| CN115034137A (en) | RVM and degradation model-based two-stage hybrid prediction method for residual life of bearing | |
| CN116595465A (en) | High-dimensional sparse data outlier detection method and system based on self-encoder and data enhancement | |
| CN111783242B (en) | RVM-KF-based rolling bearing residual life prediction method and device | |
| CN116401950B (en) | Rolling bearing performance degradation prediction method based on time chart convolution neural network | |
| CN113850161A (en) | Flywheel fault identification method based on LSTM deep noise reduction self-encoder | |
| CN114626625B (en) | Bearing life prediction method based on multi-module U-BiLSTM network | |
| CN117435873A (en) | Data management method based on intelligent spraying dust fall | |
| CN113435321A (en) | Method, system and equipment for evaluating state of main shaft bearing and readable storage medium | |
| CN116743180B (en) | Intelligent storage method for energy storage power supply data | |
| CN118819104A (en) | Power plant equipment fault diagnosis method, device and equipment | |
| CN114021275B (en) | Rolling bearing fault diagnosis method based on deep convolution fuzzy system | |
| CN116433667B (en) | Detection method for low-voltage intelligent switch of power distribution Internet of things | |
| CN118132934A (en) | Real-time state analysis method and system for machine tool spindle | |
| CN117851872A (en) | High-frequency vibration treatment diagnosis method and system for aero-engine |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20200117 Assignee: CHONGQING SURVEY INSTITUTE Assignor: CHONGQING JIAOTONG University Contract record no.: X2024980019287 Denomination of invention: Remaining life prediction method for rolling bearings based on convolutional long short-term memory recurrent neural network Granted publication date: 20220624 License type: Common License Record date: 20241030 |