CN114781450B - Train rolling bearing state identification method based on parameter optimization MOMEDA-MIA-CNN - Google Patents
Train rolling bearing state identification method based on parameter optimization MOMEDA-MIA-CNN Download PDFInfo
- Publication number
- CN114781450B CN114781450B CN202210433462.7A CN202210433462A CN114781450B CN 114781450 B CN114781450 B CN 114781450B CN 202210433462 A CN202210433462 A CN 202210433462A CN 114781450 B CN114781450 B CN 114781450B
- Authority
- CN
- China
- Prior art keywords
- rolling bearing
- train
- mia
- cnn
- momeda
- 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
- 238000005096 rolling process Methods 0.000 title claims abstract description 110
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000005457 optimization Methods 0.000 title claims abstract description 26
- 230000009467 reduction Effects 0.000 claims abstract description 13
- 238000007781 pre-processing Methods 0.000 claims abstract description 5
- 238000001914 filtration Methods 0.000 claims description 25
- 230000001133 acceleration Effects 0.000 claims description 13
- 230000006870 function Effects 0.000 claims description 13
- 238000012549 training Methods 0.000 claims description 12
- 238000012360 testing method Methods 0.000 claims description 9
- 239000002245 particle Substances 0.000 claims description 6
- 230000003595 spectral effect Effects 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 5
- 238000002474 experimental method Methods 0.000 claims description 4
- 238000001228 spectrum Methods 0.000 claims description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000005540 biological transmission Effects 0.000 abstract description 7
- 238000000605 extraction Methods 0.000 abstract description 4
- 238000013527 convolutional neural network Methods 0.000 description 17
- 230000008569 process Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 241001391944 Commicarpus scandens Species 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- 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
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
- Train Traffic Observation, Control, And Security (AREA)
Abstract
The invention relates to the technical field of running state identification of key parts of a train, and discloses a method for identifying the state of a rolling bearing of the train based on parameter optimization MOMEDA-MIA-CNN, which comprises the following steps: step (1), acquiring rolling bearing data of a train; step (2), optimizing MOMEDA parameters; step (3), data preprocessing; step (4), train rolling bearing operation data modal compartmentalization; step 5, establishing an initial model of the rolling bearing MIA-CNN of the train; step (6), optimizing a MIA-CNN model of the train rolling bearing; and (7) identifying the running state of the rolling bearing of the train. The MIA-CNN model is used for carrying out self-adaptive feature extraction and state identification on the section signals after noise reduction, so that end-to-end transmission of data is realized, and the artificial influence is reduced, and therefore, the train rolling bearing running state identification result has good reliability.
Description
Technical Field
The invention relates to the technical field of running state identification of key parts of a train, in particular to a method for identifying the state of a rolling bearing of the train based on parameter optimization MOMEDA-MIA-CNN.
Background
The rapid development of rail transit brings infinite convenience to the life of masses, and meanwhile, the safety of trains is receiving more and more attention. The rolling bearing is used as a key part of a bogie component and a wheel shaft, is one of the most important parts in a transmission system, and has the important significance in timely identifying the running state of the rolling bearing and ensuring the safe and stable running of a train and the life safety of train personnel.
In the actual running process of the train, the rolling bearing of the train bears huge load for a long time, and the high-speed and high-temperature complex and severe running environment where the bearing is located makes the rolling bearing of the train extremely easy to break down, and the rolling bearing is usually integrated in other mechanical equipment, so that the collected vibration signals contain a large amount of impact noise and interference of environmental noise besides the vibration signals of the bearing, and great difficulty is brought to accurately identifying the running state of the rolling bearing. The minimum entropy deconvolution algorithm (MOMEDA) with optimal adaptation of multiple points is a novel non-iterative blind deconvolution enhancement technology, the MOMEDA does not need to set iteration times in advance, the optimal deconvolution filter is obtained by using the maximized multi-point kurtosis as an objective function through a matrix operation method, fault pulse signals are extracted from actually measured rolling bearing vibration signals rapidly and efficiently, and therefore noise reduction of the actually measured signals is achieved. Modal Interval Analysis (MIA) is one of the effective tools to solve the uncertainty problem, which converts the uncertainty into modal interval form and quantifies the uncertainty by the width of the modal interval. Converting data into a modal interval form can contain richer information, and the credibility is higher. Convolutional Neural Networks (CNNs) are one of the typical deep learning algorithms, generally comprising convolutional layers, pooled layers, fully connected layers, and the like. The convolution layer of CNN can learn and extract key features in signals, the pooling layer screens fault features extracted by the convolution layer, filters useless features to realize dimension reduction of the fault features, the full-connection layer synthesizes the extracted features, and a softmax classifier identifies the running state of the equipment. The MIA-based CNN (MIA-CNN) has the same structure and working principle as the CNN, and in the processes of model establishment, optimization and identification, a mode interval theory of a data characteristic and the like from an original determined value is converted into a mode interval form, the mode interval contains all possibilities of a true value of the data, and a final output interval also contains all possible values of similarity with a corresponding model, so that some uncertainty factors caused by noise interference, a transmission path and the like can be eliminated. The MOMEDA and MIA-CNN are combined to identify the running state of the train rolling bearing, so that the reliability of the identification result of the train rolling bearing is greatly improved.
Disclosure of Invention
The invention aims to solve the problem that the running state recognition rate is low due to weak fault characteristics caused by the influence of a transmission path and background noise on a train rolling bearing vibration signal, and provides a train rolling bearing state recognition method based on the combination of parameter optimization MOMEDA-MIA-CNN. And determining an optimal filter period T and an optimal filter length L of Multipoint Optimal Minimum Entropy Deconvolution (MOMEDA) by utilizing a particle swarm optimization algorithm, performing MOMEDA noise reduction on the acquired train rolling bearing vibration signals by utilizing the obtained MOMEDA optimal parameters, converting the noise-reduced signals into a modal interval form, inputting a convolutional neural network model (MIA-CNN) based on modal interval analysis, and completing the extraction of train rolling bearing fault characteristics and the identification of running states.
In order to achieve the above purpose, the present invention provides the following technical solutions: a train rolling bearing state identification method based on parameter optimization MOMEDA-MIA-CNN comprises the following steps:
step (1), acquiring rolling bearing data of a train;
step (2), optimizing MOMEDA parameters;
optimizing a filtering period T and a filter length L of a MOMEDA algorithm by using a kurtosis spectral product as an optimization index through a particle swarm optimization algorithm, selecting the filtering parameter with the optimal filtering period and filter length when the kurtosis spectral product is maximum, and marking the filtering period and the filter length as Tu and Lu respectively;
step (3), data preprocessing;
performing MOMEDA noise reduction on the acquired train rolling bearing vibration signal data set { x (t) } by utilizing the optimal filtering parameters Tu and Lu obtained in the step (2), reducing noise interference, and forming a train rolling bearing operation data set { y (t) } after noise reduction;
step (4), train rolling bearing operation data modal compartmentalization;
converting the preprocessed train rolling bearing operation data set { y (t) } into a modal interval form y (t) according to an error theory and a modal interval theory,to increase the reliability of the data set to be analyzed; wherein the method comprises the steps ofy(t) represents the lower bound of the rolling bearing operating dataset,>representing the rolling bearing operational dataset upper bound.
Step 5, establishing an initial model of the rolling bearing MIA-CNN of the train;
dividing rolling bearing types according to the running state of the rolling bearing of the train, constructing an initial MIA-CNN model by combining a noise-reduced modal compartmentalized running data set of the rolling bearing of the train, and encoding the running state of the rolling bearing of the train to form an ideal output target of the MIA-CNN model;
step (6), optimizing a MIA-CNN model of the train rolling bearing;
according to the CNN model training method, taking a part of each state of the train rolling bearing operation data set with the mode interval in the step (4) as a training sample, and adopting an Adam iterative updating algorithm to iteratively update the weights and the bias of MIA-CNN key points; obtaining a loss function value and accuracy of training sample data of each iteration through forward propagation calculation, updating model parameters through reverse propagation, and completing train rolling bearing MIA-CNN model optimization when the training sample data loss function value is low and the accuracy is high and the expected requirement is met;
step (7), identifying the running state of the rolling bearing of the train;
and (3) taking the rest part of signals of the train rolling bearing signals which are subjected to the mode interval in the step (4) as test samples, inputting the test samples into an optimized train rolling bearing MIA-CNN model, calculating an output result of the MIA-CNN model, and comparing the obtained result by using a mode interval size comparison method, wherein an operation state corresponding to the mode interval maximum value corresponding to the code is the train rolling bearing operation state.
Preferably, in step (1), a train running part experiment platform is built aiming at the train rolling bearing, and the platform comprises two unidirectional acceleration sensors, a data acquisition card, a three-phase asynchronous motor and a PC (personal computer), wherein the unidirectional acceleration sensors acquire running data of the train running part rolling bearing, the two unidirectional acceleration sensors are respectively arranged in the horizontal direction and the vertical direction of a rolling bearing end cover of the test bed, time domain analysis is carried out on vibration signals in the horizontal direction and the vertical direction acquired by the two unidirectional acceleration sensors, and the vibration signals with large amplitude are selected as a train rolling bearing running original signal data set and recorded as { x (t) }, wherein t is time.
Preferably, in step (2), the kurtosis product is the product of the kurtosis and the envelope spectrum peak factor.
Preferably, in step (4), the mathematical definition of the modal interval z is:
i.e. the modal interval z passes through a pair of real numbersz，Is defined in whichzRepresenting the lower bound->Represents an upper bound, kR represents a real set of modal closed-form intervals, where z is not +.>The upper and lower bounds of the modal interval are not limited, and the theoretical basis of the mathematical operation of the modal interval is the Kaucher algorithm.
Preferably, in step (5), the initial parameters of MIA-CNN are initially set empirically, including: activation functions, loss functions, learning rate, batch size, optimizers, iteration number, and Dropout layers.
The beneficial effects of the invention are as follows:
compared with the existing train rolling bearing state identification method, the method provided by the invention has the advantages that interference generated by a transmission path, background noise and the like is weakened by using a MOMEDA algorithm, the filtering period and the filtering length of the MOMEDA algorithm are optimized by using a particle swarm optimization algorithm, the vibration signal of the train rolling bearing is recovered from a noise signal to the greatest extent, meanwhile, the filtered signal is subjected to modal interval by using a modal interval method in consideration of the incompleteness of filtering, the uncertainty problem in the vibration signal of the train rolling bearing is further processed, and the extraction of the fault characteristics and the identification of the running state of the train rolling bearing are automatically completed by using an MIA-CNN model; the method not only carries out filtering denoising on the original signal, but also secondarily processes the problem of uncertainty in the train rolling bearing operation data set by a mode interval method, so that the reliability of a state identification result is improved, and the method has the following advantages:
1. two acceleration sensors are adopted to acquire the running data of the rolling bearing of the train, and the data with larger amplitude is selected as the running data of the rolling bearing, so that the data source is more reliable;
2. the particle swarm optimization algorithm is adopted to optimize the parameters of the MOMEDA, so that the defect that the main influence parameters of the MOMEDA depend on artificial experience selection is overcome, and the MOMEDA filtering effect is better;
3. the filtered train rolling bearing signals are converted into a modal interval form, and the problem of numerical uncertainty in the input of a train rolling bearing state identification model can be effectively solved through the problem of uncertainty in the secondary processing signal acquisition process of modal interval theory;
in summary, the MIA-CNN model is used to perform adaptive feature extraction and state recognition on the section signal after noise reduction, end-to-end transmission of data is achieved, and human influence is reduced, so that the train rolling bearing running state recognition result has good reliability.
Drawings
FIG. 1 is a block flow diagram of an embodiment of a method for identifying the state of a rolling bearing of a train based on parameter optimization MOMEDA-MIA-CNN.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, the present invention provides a technical solution: the train rolling bearing state identification method based on the parameter optimization MOMEDA-MIA-CNN specifically comprises the following steps:
step (1), acquiring rolling bearing data of a train;
aiming at a train rolling bearing, a train running part experiment platform is built, and comprises two unidirectional acceleration sensors PCB 352C33, a data acquisition card NI PXI-1042, a three-phase asynchronous motor, a PC (personal computer) and the like, wherein the unidirectional acceleration sensors PCB 352C33 acquire running data of the train running part rolling bearing, and the two unidirectional acceleration sensors PCB 352C33 are respectively arranged in the horizontal direction and the vertical direction of a rolling bearing end cover of a test bed; performing time domain analysis on vibration signals in the horizontal and vertical directions acquired by the two unidirectional acceleration sensors, selecting the vibration signals with large amplitude as an original signal data set for running the rolling bearing of the train, and recording the original signal data set as { x (t) }, wherein t is time;
step (2), optimizing MOMEDA parameters;
parameters affecting the MOMEDA filtering effect are mainly: a filtering period T and a filter length L; when the set filtering period T is inconsistent with the actual filtering period, the obtained filtering result is very likely to be completely wrong, and the mistake is often difficult to be perceived; the filter length L is in direct proportion to the filtering effect of the MOMEDA algorithm, but when the filter length is too large, part of effective information of an original signal can be filtered;
therefore, the reasonable selection of the filtering period T and the filter length L has decisive influence on the noise reduction effect of the MOMEDA;
the kurtosis is a time domain index capable of representing the impact magnitude of a signal, the envelope spectrum peak factor is a frequency domain index capable of representing the periodicity of the signal and the impact magnitude of the signal, the kurtosis is the product of the kurtosis and the envelope spectrum peak factor, and when the kurtosis is larger, the periodicity of the signal is more obvious and the impact is larger; using kurtosis spectral products as optimization indexes, optimizing a filtering period T and a filter length L of a MOMEDA algorithm by using a particle swarm optimization algorithm, taking T and L when the kurtosis spectral products are maximum as optimal filtering parameters, and respectively marking the T and L as Tu and Lu;
step (3), data preprocessing;
performing MOMEDA noise reduction on the acquired train rolling bearing vibration signal data set { x (t) } by utilizing the optimal filtering parameters Tu and Lu obtained in the step (2), reducing noise interference, and forming a train rolling bearing operation data set { y (t) } after noise reduction;
step (4), train rolling bearing operation data modal compartmentalization;
considering the problem of uncertainty in the transmission path of the train rolling bearing vibration signal and the data preprocessing process, converting the preprocessed train rolling bearing operation data set { y (t) } into a modal interval form y (t) according to the modal interval theory and the error theory,to increase the reliability of the signal to be analyzed; wherein the method comprises the steps ofy(t) represents the lower bound of the rolling bearing operating dataset,>representing the rolling bearing operational dataset upper bound;
the mathematical definition of the modal interval z is:
i.e. the modal interval z passes through a pair of real numbersz，Is defined in whichzRepresenting the lower bound->Represents an upper bound, kR represents a real set of modal closed-form intervals, where z is not +.>The upper and lower bounds of the modal interval are not limited, and the theoretical basis of the mathematical operation of the modal interval is the Kaucher algorithm.
Step 5, establishing an initial model of the rolling bearing MIA-CNN of the train;
constructing an initial MIA-CNN model according to the train running state expression type and the train rolling bearing signals subjected to the modal interval after input noise reduction, and encoding the train running state expression type to form an ideal output target of the MIA-CNN model;
in the embodiment, the running states of the rolling bearings of the train are preferably 5, namely normal, inner ring fault, outer ring fault, cage fault and ball fault; the sensitive characteristic data of each running state is processed into 1024 x 1 vector form to be input into a convolutional neural network model, and simultaneously, 5 train rolling bearing running states are encoded to form an ideal output target of the convolutional neural network model,
wherein the normal code is ([ 1,1], [0,0 ]);
the inner ring fault code is ([ 0,0], [1,1], [0,0 ]);
the outer ring fault code is ([ 0,0], [1,1], [0,0 ]);
cage failure codes are ([ 0,0], [1,1], [0,0 ]);
the ball fault code is ([ 0,0], [1,1 ]);
initial parameters of the convolutional neural network are initially set according to experience, including: activation functions, loss functions, learning rate, batch size, optimizers, iteration number, dropout layers, etc.;
step (6), optimizing a MIA-CNN model of the train rolling bearing;
according to the CNN model training method, taking a part of each state of the train rolling bearing operation data set with the mode interval in the step (4) as a training sample, and adopting an Adam iterative updating algorithm to iteratively update the weights and the bias of MIA-CNN key points; obtaining a loss function value and accuracy of training sample data of each iteration through forward propagation calculation, updating model parameters through reverse propagation, and completing train rolling bearing MIA-CNN model optimization when the training sample data loss function value is low and the accuracy is high and the expected requirement is met;
step (7), identifying the running state of the rolling bearing of the train;
and (3) taking the rest part of signals of the train rolling bearing signals with the mode interval in the step (4) as test samples, inputting the test samples into an optimized train rolling bearing MIA-CNN model, calculating an output result of the MIA-CNN model, and comparing the obtained result by using a mode interval size comparison method, wherein an operation state corresponding to the mode interval maximum value corresponding to the code is the train rolling bearing operation state.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (5)
1. The train rolling bearing state identification method based on the parameter optimization MOMEDA-MIA-CNN is characterized by comprising the following steps of:
step (1), acquiring rolling bearing data of a train;
step (2), optimizing MOMEDA parameters;
optimizing a filtering period T and a filter length L of a MOMEDA algorithm by using a kurtosis spectral product as an optimization index through a particle swarm optimization algorithm, selecting the filtering parameter with the optimal filtering period and filter length when the kurtosis spectral product is maximum, and marking the filtering period and the filter length as Tu and Lu respectively;
step (3), data preprocessing;
performing MOMEDA noise reduction on the acquired train rolling bearing vibration signal data set { x (t) } by utilizing the optimal filtering parameters Tu and Lu obtained in the step (2), reducing noise interference, and forming a train rolling bearing operation data set { y (t) } after noise reduction;
step (4), train rolling bearing operation data modal compartmentalization;
converting the preprocessed train rolling bearing operation data set { y (t) } into a modal interval form y (t) according to an error theory and a modal interval theory,to increase the reliability of the data set to be analyzed; wherein the method comprises the steps ofy(t) represents the lower bound of the rolling bearing operating dataset,>representing the rolling bearing operational dataset upper bound;
step 5, establishing an initial model of the rolling bearing MIA-CNN of the train;
dividing rolling bearing types according to the running state of the rolling bearing of the train, constructing an initial MIA-CNN model by combining a noise-reduced modal compartmentalized running data set of the rolling bearing of the train, and encoding the running state of the rolling bearing of the train to form an ideal output target of the MIA-CNN model;
step (6), optimizing a MIA-CNN model of the train rolling bearing;
according to the CNN model training method, taking a part of each state of the train rolling bearing operation data set with the mode interval in the step (4) as a training sample, and adopting an Adam iterative updating algorithm to iteratively update the weights and the bias of MIA-CNN key points; obtaining a loss function value and accuracy of training sample data of each iteration through forward propagation calculation, updating model parameters through reverse propagation, and completing train rolling bearing MIA-CNN model optimization when the training sample data loss function value is low and the accuracy is high and the expected requirement is met;
step (7), identifying the running state of the rolling bearing of the train;
and (3) taking the rest part of signals of the train rolling bearing signals which are subjected to the mode interval in the step (4) as test samples, inputting the test samples into an optimized train rolling bearing MIA-CNN model, calculating an output result of the MIA-CNN model, and comparing the obtained result by using a mode interval size comparison method, wherein an operation state corresponding to the mode interval maximum value corresponding to the code is the train rolling bearing operation state.
2. The method for identifying the state of the rolling bearing of the train based on the parameter optimization MOMEDA-MIA-CNN according to claim 1, wherein the method comprises the following steps of: in the step (1), a train running part experiment platform is built aiming at a train rolling bearing, the experiment platform comprises two unidirectional acceleration sensors, a data acquisition card, a three-phase asynchronous motor and a PC machine, wherein the unidirectional acceleration sensors acquire running data of the train running part rolling bearing, the two unidirectional acceleration sensors are respectively arranged in the horizontal direction and the vertical direction of a rolling bearing end cover of a test bed, time domain analysis is carried out on vibration signals in the horizontal direction and the vertical direction acquired by the two unidirectional acceleration sensors, the vibration signals with large amplitude are selected to be used as a train rolling bearing running original signal data set, and the data set is recorded as { x (t) }, wherein t is time.
3. The method for identifying the state of the rolling bearing of the train based on the parameter optimization MOMEDA-MIA-CNN according to claim 1, wherein the method comprises the following steps of: in the step (2), the product of kurtosis and the peak factor of envelope spectrum is taken as the product of kurtosis and the kurtosis of kurtosis.
4. The method for identifying the state of the rolling bearing of the train based on the parameter optimization MOMEDA-MIA-CNN according to claim 1, wherein the method comprises the following steps of: in step (4), wherein the mathematical definition of the modal interval z is:
i.e. the modal interval z passes through a pair of real numbersz，Is defined in whichzRepresenting the lower bound->Represents an upper bound, kR represents a real set of modal closed-form intervals, where z is not +.>The upper and lower bounds of the modal interval are not limited, and the theoretical basis of the mathematical operation of the modal interval is the Kaucher algorithm.
5. The method for identifying the state of the rolling bearing of the train based on the parameter optimization MOMEDA-MIA-CNN according to claim 1, wherein the method comprises the following steps of: in step (5), initial parameters of MIA-CNN are initially set empirically, including: activation functions, loss functions, learning rate, batch size, optimizers, iteration number, and Dropout layers.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210433462.7A CN114781450B (en) | 2022-04-24 | 2022-04-24 | Train rolling bearing state identification method based on parameter optimization MOMEDA-MIA-CNN |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210433462.7A CN114781450B (en) | 2022-04-24 | 2022-04-24 | Train rolling bearing state identification method based on parameter optimization MOMEDA-MIA-CNN |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114781450A CN114781450A (en) | 2022-07-22 |
| CN114781450B true CN114781450B (en) | 2024-03-29 |
Family
ID=82433397
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210433462.7A Active CN114781450B (en) | 2022-04-24 | 2022-04-24 | Train rolling bearing state identification method based on parameter optimization MOMEDA-MIA-CNN |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114781450B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115859091B (en) * | 2022-11-01 | 2023-05-26 | 哈尔滨工业大学 | Bearing fault feature extraction method, electronic equipment and storage medium |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104502126A (en) * | 2014-12-28 | 2015-04-08 | 华东交通大学 | Modal intervals-based high-speed train bogie fault diagnosis method |
| CN108168886A (en) * | 2017-12-22 | 2018-06-15 | 合肥工业大学 | Rolling Bearing Fault Character extracts and method for monitoring operation states |
| CN109635334A (en) * | 2018-11-12 | 2019-04-16 | 武汉科技大学 | Fault Diagnosis of Roller Bearings, system and medium based on particle group optimizing |
| WO2021128510A1 (en) * | 2019-12-27 | 2021-07-01 | 江苏科技大学 | Bearing defect identification method based on sdae and improved gwo-svm |
| WO2021135630A1 (en) * | 2019-12-31 | 2021-07-08 | 福州大学 | Rolling bearing fault diagnosis method based on grcmse and manifold learning |
| WO2021143128A1 (en) * | 2020-01-16 | 2021-07-22 | 佛山科学技术学院 | Deep learning-based bearing state identification method and system |
| CN113743010A (en) * | 2021-08-31 | 2021-12-03 | 三峡大学 | Rolling bearing running state evaluation method based on EEMD energy entropy |
| CN113919220A (en) * | 2021-10-08 | 2022-01-11 | 西安科技大学 | Intelligent fault diagnosis method for rolling bearing driven by vibration image |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108388860B (en) * | 2018-02-12 | 2020-04-28 | 大连理工大学 | Aero-engine rolling bearing fault diagnosis method based on power entropy spectrum-random forest |
-
2022
- 2022-04-24 CN CN202210433462.7A patent/CN114781450B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104502126A (en) * | 2014-12-28 | 2015-04-08 | 华东交通大学 | Modal intervals-based high-speed train bogie fault diagnosis method |
| CN108168886A (en) * | 2017-12-22 | 2018-06-15 | 合肥工业大学 | Rolling Bearing Fault Character extracts and method for monitoring operation states |
| CN109635334A (en) * | 2018-11-12 | 2019-04-16 | 武汉科技大学 | Fault Diagnosis of Roller Bearings, system and medium based on particle group optimizing |
| WO2021128510A1 (en) * | 2019-12-27 | 2021-07-01 | 江苏科技大学 | Bearing defect identification method based on sdae and improved gwo-svm |
| WO2021135630A1 (en) * | 2019-12-31 | 2021-07-08 | 福州大学 | Rolling bearing fault diagnosis method based on grcmse and manifold learning |
| WO2021143128A1 (en) * | 2020-01-16 | 2021-07-22 | 佛山科学技术学院 | Deep learning-based bearing state identification method and system |
| CN113743010A (en) * | 2021-08-31 | 2021-12-03 | 三峡大学 | Rolling bearing running state evaluation method based on EEMD energy entropy |
| CN113919220A (en) * | 2021-10-08 | 2022-01-11 | 西安科技大学 | Intelligent fault diagnosis method for rolling bearing driven by vibration image |
Non-Patent Citations (2)
| Title |
|---|
| 优化参数VMD和MED在列车齿轮箱滚动轴承故障诊断中的应用;李长青;林建辉;胡永旭;;机车电传动;20200510(03);全文 * |
| 滚动轴承故障特征自适应增强的相关峭度联合降噪方法;张龙;蔡秉桓;吴佳敏;熊国良;;机械设计与研究;20200420(02);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114781450A (en) | 2022-07-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107657250B (en) | Bearing fault detection and positioning method and detection and positioning model implementation system and method | |
| CN111914883B (en) | Spindle bearing state evaluation method and device based on deep fusion network | |
| CN108398268B (en) | Bearing performance degradation evaluation method | |
| CN112906644B (en) | Mechanical fault intelligent diagnosis method based on deep migration learning | |
| CN111947928B (en) | Multi-source information fusion bearing fault prediction system and method | |
| CN111539152B (en) | Rolling bearing fault self-learning method based on two-stage twin convolutional neural network | |
| CN106124212A (en) | Based on sparse coding device and the Fault Diagnosis of Roller Bearings of support vector machine | |
| CN109946080B (en) | Mechanical equipment health state identification method based on embedded circulation network | |
| CN112926644A (en) | Method and system for predicting residual service life of rolling bearing | |
| CN112305388B (en) | On-line monitoring and diagnosing method for insulation partial discharge faults of generator stator winding | |
| CN110737976A (en) | mechanical equipment health assessment method based on multi-dimensional information fusion | |
| CN113485863B (en) | Method for generating heterogeneous imbalance fault samples based on improved generation of countermeasure network | |
| Son et al. | Deep learning-based anomaly detection to classify inaccurate data and damaged condition of a cable-stayed bridge | |
| CN114781450B (en) | Train rolling bearing state identification method based on parameter optimization MOMEDA-MIA-CNN | |
| CN115452381A (en) | Rolling bearing fault diagnosis method under strong noise interference based on multi-modal decomposition and long-short term memory network | |
| Liang et al. | Multibranch and multiscale dynamic convolutional network for small sample fault diagnosis of rotating machinery | |
| Wang et al. | Fault diagnosis of industrial robots based on multi-sensor information fusion and 1D convolutional neural network | |
| CN113705405A (en) | Nuclear pipeline fault diagnosis method | |
| CN110490218B (en) | Rolling bearing fault self-learning method based on two-stage DBN | |
| CN116910574A (en) | Mechanical equipment diagnosis method and device based on meta-learning and time convolution network | |
| CN116227172A (en) | Rolling bearing performance degradation evaluation method based on convolutional neural network | |
| CN114354185B (en) | Gear box fault diagnosis method based on decoupling network | |
| CN116186609A (en) | LSTM-based manned submersible sensor fault data enhancement method | |
| CN116007937A (en) | Intelligent fault diagnosis method and device for mechanical equipment transmission part | |
| CN115962944A (en) | VMD-based real-time diagnosis method for bearing fault of cutting part of intelligent coal mining machine group |
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 |