CN106769030A - A kind of bearing state tracking and Forecasting Methodology based on MEA BP neural network algorithms - Google Patents
A kind of bearing state tracking and Forecasting Methodology based on MEA BP neural network algorithms Download PDFInfo
- Publication number
- CN106769030A CN106769030A CN201610988210.5A CN201610988210A CN106769030A CN 106769030 A CN106769030 A CN 106769030A CN 201610988210 A CN201610988210 A CN 201610988210A CN 106769030 A CN106769030 A CN 106769030A
- Authority
- CN
- China
- Prior art keywords
- bearing
- neural network
- group
- mea
- sub
- 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
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
Landscapes
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A kind of bearing state tracking and Forecasting Methodology based on MEA BP neural network algorithms, comprise the following steps：S1. life cycle management inner bearing vibration signal in the horizontal and vertical directions is gathered；S2. the virtual value of vibration signal in both direction is calculated respectively；S3. RRMS values are calculated and as the health index of bearing, it is smoothed using median filter method；S4. the initial weight and threshold value of BP neural network algorithm are optimized using MEA algorithms, using time index as input variable, the health index of bearing is trained as output variable to BP neural network；After the completion of S5.BP neural metwork trainings, any time index before input current time and current time, it is calculated the state corresponding to the time index, reach the effect of status tracking, input current time subsequent time time index, prediction obtains the state of subsequent time, reaches the effect of status predication.The present invention is good to the tracking effect of bearing health status, and precision of prediction is higher, takes less.
Description
Technical field
It is the invention belongs to bearing failure diagnosis and prediction field more particularly to a kind of based on MEA-BP neural network algorithms
Bearing state is tracked and Forecasting Methodology.
Background technology
Bearing is widely applied to precision machine tool, high-speed railway, wind-driven generator etc. as a parts for key
In great dynamoelectric equipment.However, bearing is also a weak element in these equipments.Related data shows, 40% or so
Motor failure is caused by bearing fault, wind-driven generator failure also be often by bearing fault caused by.Bearing is once sent out
Raw failure, gently then reduces or loses some functions of system, and unexpected shutdown that is heavy then causing system is failed and some great peaces
Full accident.Therefore a research weight of status monitoring, fault diagnosis and fault prediction always business circles and academia is carried out to bearing
Point, and the online tracking to bearing state is to evaluate the real-time health status of bearing with prediction, judges whether it can meet system
The important leverage of job requirement.Because neural network algorithm has stronger Nonlinear Learning and mapping ability, can be preferable
Reflect the development trend of distinct device (such as bearing) working condition, have in terms of status tracking and predicting residual useful life compared with
To be widely applied.But the characteristics of being limited to its own structure, neural network algorithm convergence rate is slower, and precision of prediction is relatively low,
And easily the problems such as there is local optimum and over-fitting, the algorithm receives certain limitation in the middle of actual application.
The content of the invention
In order to overcome existing bearing state tracking relatively low not compared with slow, precision of prediction with the convergence rate of Forecasting Methodology
Foot, it is higher, time-consuming shorter based on mind evolutionary (MEA)-BP neural network algorithm that the present invention provides a kind of precision of prediction
Bearing state tracking and Forecasting Methodology, MEA algorithms have stronger local optimal searching ability, optimization neural network can be used for
The initial weight and threshold value of algorithm, and then accelerate the convergence rate of neutral net and improve precision of prediction, preferably compensate for
The deficiency of neural network algorithm itself.
The technical solution adopted for the present invention to solve the technical problems is：
A kind of bearing state tracking and Forecasting Methodology based on MEA-BP neural network algorithms, methods described include following step
Suddenly：
S1. life cycle management inner bearing vibration signal in the horizontal and vertical directions is gathered；
S2. the virtual value of both direction vibration signal is calculated respectively；
S3. RRMS values are calculated and as the index of reflection bearing health status according to virtual value, using medium filtering
Method is smoothed to the health index；
S4. the initial weight and threshold value of BP neural network algorithm are optimized using MEA algorithms, using time index as
Input variable, the health index of bearing is trained as output variable to BP neural network；
After the completion of S5.BP neural metwork trainings, any time index before input current time and current time is calculated
The state value corresponding to the time index is obtained, the effect of status tracking is reached, current time subsequent time time index is input into,
Prediction obtains the state of subsequent time, reaches the effect of status predication.
Further, in the step S4, the BP neural network structure includes an input layer, a hidden layer and one
Output layer, input layer number determines that hidden layer neuron number refers to empirical equation according to the quantity of influence factorDetermination is gathered in examination, and n is node in hidden layer, niIt is input number of nodes, njIt is output node number, I is between 1~10
Constant；Output layer neuron number determines according to output quantity；Input layer is to hidden layer and hidden layer to the transmission of output layer
Function uses tangent S type transmission function tansig, training algorithm to use trainlm algorithms.
Further, in the step S4, the initial weight and threshold process of MEA algorithm optimization BP neural network algorithms are such as
Under：
Initial population, winning sub-group and interim sub-group are produced first；
Secondly, after winning sub-group and interim sub-group are produced, each sub-group will first carry out operation similartaxis, using population
Ripe discriminant function judges whether each sub-group operation similartaxis completes；
Again, treat after the completion of each winning sub-group and interim sub-group operation similartaxis, perform operation dissimilation, face when one
When sub-group score higher than certain maturation winning sub-group, then the winning sub-group substituted by interim sub-group, former winning son
Individuality in colony is released, if a score for the interim sub-group of maturation is less than the score of any one winning sub-group,
Then the interim sub-group is gone out of use, and individuality therein is released, and the individuality being released is re-searched for and formed in global scope
New interim colony；
Finally, when iteration stopping condition is met, MEA algorithms terminate optimization process, now, according to coding rule, to seeking
The optimum individual for finding is parsed, so as to obtain the initial weight and threshold value of BP neural network algorithm.
Further, in the step S4, BP neural network is trained firstly the need of the data to being input into and export
It is normalized and is processed with renormalization, process is as follows：
Data normalization treatment is carried out according to following linear function, data are mapped to Ymin~YmaxWithin the scope of：
In above formula, i=[1,2 ..., m], X are that length is the array of m, XmaxIt is the maximum in the array, XminIt is the number
Minimum value in group, Y is the array after normalization；
Then, maximum convergence number of times, display interval, convergence error, learning rate are set；After the completion of training, training is tied
Fruit carries out renormalization treatment：
In the step S2, to the bearing life cycle management vibration signal obtained by S1, construction feature parameter, process is as follows：
If kth moment vibration signal Zk, wherein comprising N number of sampled point, sample data set Zk=(z1,z2,…,zN), calculate
The virtual value of the sample set：
Due to acquiring the vibration signal of horizontally and vertically both direction upper bearing (metal) respectively, therefore to shaking in both direction
Dynamic signal is calculated after virtual value respectively, the virtual value in synchronization both direction is calculated into Euler's distance, so that two
The information integration in direction is on a virtual value for synthesis.
In the step S3, using comprehensive virtual value, RRMS values are calculated and as the finger for characterizing bearing health status
Number, process is as follows：
The comprehensive virtual value RMS of any instant s when taking bearing normal worksAs reference, the synthesis of kth moment bearing
Virtual value is RMSk, then the RRMS values at the moment be defined as：
RRMS to gained carries out medium filtering, health of the bearing in life cycle management is obtained after being smoothed and is referred to
Number HI.
Technology design of the invention is：By gathering bearing vibration signal, virtual value is calculated and obtained based on vibration signal,
Virtual value during reference bearing normal work calculates RRMS values, and as the health index of reflection bearing state, by the time
Index determines the topological structure of BP neural network as output variable as input variable, health index.In training BP nerve nets
Before network, the initial weight and threshold value of BP neural network algorithm are optimized using MEA algorithms, using learning sample to BP god
Be trained through network, after the completion of training, using the MEA-BP neural network algorithms for building bearing is carried out the tracking of state with
Prediction.
Beneficial effects of the present invention are：The health status of bearing can be preferably tracked, to the status predication of subsequent time
It is more accurate and time-consuming less.
Brief description of the drawings
Fig. 1 is bearing state tracking and Forecasting Methodology flow chart based on MEA-BP neural network algorithms.
Fig. 2 is bearing life cycle management health index schematic diagram.
Fig. 3 is MEA algorithmic system structural representations.
Fig. 4 is BP neural network topological structure schematic diagram.
Fig. 5 is bearing life cycle management status tracking result.
Fig. 6 is bearing subsequent time status predication result.
Specific embodiment
The invention will be further described below in conjunction with the accompanying drawings.
1~Fig. 6 of reference picture, a kind of bearing state tracking and Forecasting Methodology based on MEA-BP neural network algorithms are described
Method is comprised the following steps：
S1. life cycle management inner bearing vibration signal in the horizontal and vertical directions is gathered；
S2. the virtual value of both direction vibration signal is calculated respectively；
S3. RRMS values are calculated and as the index of reflection bearing health status according to virtual value, using medium filtering
Method is smoothed to the health index；
S4. the initial weight and threshold value of BP neural network algorithm are optimized using MEA algorithms, using time index as
Input variable, the health index of bearing is trained as output variable to BP neural network；
After the completion of S5.BP neural metwork trainings, any time index before input current time and current time is calculated
The state value corresponding to the time index is obtained, the effect of status tracking is reached, current time subsequent time time index is input into,
Prediction obtains the state of subsequent time, reaches the effect of status predication.
Further, in the step S2, to the bearing life cycle management vibration signal obtained by S1, construction feature parameter, mistake
Journey is as follows：
If kth moment vibration signal Zk, wherein comprising N number of sampled point, sample data set Zk=(z1,z2,…,zN), calculate
The virtual value of the sample set：
Due to acquiring the vibration signal of horizontally and vertically both direction upper bearing (metal) respectively, therefore to shaking in both direction
Dynamic signal is calculated after virtual value respectively, the virtual value in synchronization both direction is calculated into Euler's distance, so that two
The information integration in direction is on a virtual value for synthesis.
Further, in the step S3, using the comprehensive virtual value for obtaining, RRMS values are calculated and as sign axle
The index of health status is held, process is as follows：
The comprehensive virtual value RMS of any instant s when taking bearing normal worksAs reference, the synthesis of kth moment bearing
Virtual value is RMSk, then the RRMS values at the moment be defined as：
RRMS to gained carries out medium filtering, health of the bearing in life cycle management is obtained after being smoothed and is referred to
Number HI.Further, in the step S4, the BP neural network structure includes an input layer, a hidden layer and one
Output layer, input layer number determines that hidden layer neuron number refers to empirical equation according to the quantity of influence factorDetermination is gathered in examination, and n is node in hidden layer, niIt is input number of nodes, njIt is output node number, I is between 1~10
Constant；Output layer neuron number determines according to output quantity；Input layer is to hidden layer and hidden layer to the transmission of output layer
Function uses tangent S type transmission function tansig, training algorithm to use trainlm algorithms.
In the step S4, the initial weight and threshold process of MEA algorithm optimization BP neural network algorithms are as follows：
Initial population, winning sub-group and interim sub-group are produced first；
Secondly, after winning sub-group and interim sub-group are produced, each sub-group will first carry out operation similartaxis, using population
Ripe discriminant function judges whether each sub-group operation similartaxis completes；
Again, treat after the completion of each winning sub-group and interim sub-group operation similartaxis, perform operation dissimilation, face when one
When sub-group score higher than certain maturation winning sub-group, then the winning sub-group substituted by interim sub-group, former winning son
Individuality in colony is released, if a score for the interim sub-group of maturation is less than the score of any one winning sub-group,
Then the interim sub-group is gone out of use, and individuality therein is released, and the individuality being released is re-searched for and formed in global scope
New interim colony；
Finally, when iteration stopping condition is met, MEA algorithms terminate optimization process, now, according to coding rule, to seeking
The optimum individual for finding is parsed, so as to obtain the initial weight and threshold value of BP neural network algorithm.
In the step S4, BP neural network is trained and is normalized firstly the need of the data to being input into and export
With renormalization treatment, comprise the following steps that：
Data normalization treatment is carried out according to following linear function, data are mapped to Ymin~YmaxWithin the scope of：
In above formula, i=[1,2 ..., m], X are that length is the array of m, XmaxIt is the maximum in the array, XminIt is the number
Minimum value in group, Y is the data after normalization；
Then, maximum convergence number of times, display interval, convergence error, learning rate are set；After the completion of training, training is tied
Fruit carries out renormalization treatment：
The present embodiment is using the bearing complete period lifetime data of PRONOSTIA platforms to based on MEA-BP neural network algorithms
Bearing state tracking verified with Forecasting Methodology.Detailed process is as follows：
S1. life cycle management inner bearing vibration signal in the horizontal and vertical directions is gathered.By acceleration transducer
The vibration signal of as above both direction is gathered, vibration signal is gathered once every 10s, a length of 0.1s, data when gathering each time
Sample frequency is 25.6kHz；
S2. the virtual value of both direction vibration signal is calculated respectively；
S3. RRMS values are calculated and as the index of reflection bearing health status according to comprehensive virtual value, using intermediate value
Filtering method is smoothed to the health index, the health index for obtaining as shown in Figure 2, it can be seen that institute
The health index of structure has obvious tendency, can preferably reflect healthy shape of the bearing in whole life cycle
State；
S4. the initial weight and threshold value of BP neural network algorithm are optimized using MEA algorithms, MEA algorithmic system knots
Structure as shown in Figure 3, using time index as input variable, the health index of bearing as output variable, to BP neural network
It is trained, BP neural network topological structure is as shown in Figure 4；
After the completion of S5.BP neural metwork trainings, if any time index before input current time and current time, can
With the health status being calculated corresponding to the moment, so as to reach the effect of status tracking, the status tracking result such as institute of accompanying drawing 5
Show, it can be seen that by training, BP neural network can preferably track the health status of bearing, if input is current
The time index of moment subsequent time, the predictable health status for obtaining subsequent time, the prediction to subsequent time health status
Result as shown in Figure 6, it can be seen that can be carried out to the bearing health status of subsequent time using neural network algorithm
Prediction, and it is more accurate to predict the outcome.
Claims (6)
1. a kind of bearing state tracking and Forecasting Methodology based on MEA-BP neural network algorithms, it is characterised in that：
The described method comprises the following steps：
S1. life cycle management inner bearing vibration signal in the horizontal and vertical directions is gathered；
S2. the virtual value of both direction vibration signal is calculated respectively；
S3. RRMS values are calculated and as the index of reflection bearing health status according to virtual value, using median filter method
The health index is smoothed；
S4. the initial weight and threshold value of BP neural network algorithm are optimized using MEA algorithms, using time index as input
Variable, the health index of bearing is trained as output variable to BP neural network；
After the completion of S5.BP neural metwork trainings, any time index before input current time and current time is calculated
State value corresponding to the time index, reaches the effect of status tracking, is input into current time subsequent time time index, prediction
The state of subsequent time is obtained, the effect of status predication is reached.
2. bearing state tracking and Forecasting Methodology based on MEA-BP neural network algorithms as claimed in claim 1, its feature
It is：In the step S4, the BP neural network structure includes input layer, a hidden layer and an output layer, defeated
Enter layer neuron number to be determined according to the quantity of influence factor, hidden layer neuron number refers to empirical equationExamination
Gather determination, n is node in hidden layer, niIt is input number of nodes, njIt is output node number, I is the constant between 1~10；Output layer
Neuron number determines according to output quantity；Input layer uses tangent to hidden layer and hidden layer to the transmission function of output layer
S type transmission function tansig, training algorithm uses trainlm algorithms.
3. bearing state tracking and Forecasting Methodology based on MEA-BP neural network algorithms as claimed in claim 1 or 2, it is special
Levy and be：In the step S4, the initial weight and threshold process of MEA algorithm optimization BP neural network algorithms are as follows：
Initial population, winning sub-group and interim sub-group are produced first；
Secondly, after winning sub-group and interim sub-group are produced, each sub-group will first carry out operation similartaxis, ripe using population
Discriminant function judges whether each sub-group operation similartaxis completes；
Again, treat after the completion of each winning sub-group and interim sub-group operation similartaxis, perform operation dissimilation, when an interim son
Colony's score is higher than the winning sub-group of certain maturation, then the winning sub-group is substituted by interim sub-group, former winning sub-group
In individuality be released, if the score of score for the interim sub-group of maturation less than any one winning sub-group, should
Interim sub-group is gone out of use, and individuality therein is released, and the individuality being released is re-searched in global scope and forms new
Interim colony；
Finally, when iteration stopping condition is met, MEA algorithms terminate optimization process, now, according to coding rule, to searching out
Optimum individual parsed, so as to obtain the initial weight and threshold value of BP neural network algorithm.
4. bearing state tracking and Forecasting Methodology based on MEA-BP neural network algorithms as claimed in claim 1 or 2, it is special
Levy and be：In the step S4, BP neural network is trained and is normalized firstly the need of the data to being input into and export
Processed with renormalization, process is as follows：
Data normalization treatment is carried out according to following linear function, data are mapped to Ymin~YmaxWithin the scope of：
In above formula, i=[1,2 ..., m], X are that length is the array of m, XmaxIt is the maximum in the array, XminFor in the array
Minimum value, Y be normalization after array；
Then, maximum convergence number of times, display interval, convergence error, learning rate are set；After the completion of training, training result is entered
The treatment of row renormalization：
5. bearing state tracking and Forecasting Methodology based on MEA-BP neural network algorithms as claimed in claim 1 or 2, it is special
Levy and be：In the step S2, to the bearing life cycle management vibration signal obtained by S1, construction feature parameter, process is as follows：
If kth moment vibration signal Zk, wherein comprising N number of sampled point, sample data set Zk=(z1,z2,…,zN), calculate the sample
The virtual value of this collection：
Due to acquiring the vibration signal of horizontally and vertically both direction upper bearing (metal) respectively, therefore to the vibration letter in both direction
Calculate number respectively after virtual value, the virtual value in synchronization both direction is calculated into Euler's distance, so as to both direction
Information integration on a virtual value for synthesis.
6. bearing state tracking and Forecasting Methodology based on MEA-BP neural network algorithms as claimed in claim 5, its feature
It is：In the step S3, using the comprehensive virtual value for obtaining, RRMS values are calculated and as sign bearing health status
Index, process is as follows：
The comprehensive virtual value RMS of any instant s when taking bearing normal worksUsed as reference, the synthesis of kth moment bearing is effectively
It is RMS to be worthk, then the RRMS values at the moment be defined as：
RRMS to gained carries out medium filtering, and health index of the bearing in life cycle management is obtained after being smoothed
HI。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610988210.5A CN106769030A (en) | 2016-11-10 | 2016-11-10 | A kind of bearing state tracking and Forecasting Methodology based on MEA BP neural network algorithms |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610988210.5A CN106769030A (en) | 2016-11-10 | 2016-11-10 | A kind of bearing state tracking and Forecasting Methodology based on MEA BP neural network algorithms |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106769030A true CN106769030A (en) | 2017-05-31 |
Family
ID=58973399
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610988210.5A Pending CN106769030A (en) | 2016-11-10 | 2016-11-10 | A kind of bearing state tracking and Forecasting Methodology based on MEA BP neural network algorithms |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106769030A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107292873A (en) * | 2017-06-29 | 2017-10-24 | 西安工程大学 | A kind of close degree detecting method of porcelain insulator ash based on color character |
| CN107358021A (en) * | 2017-06-01 | 2017-11-17 | 华南理工大学 | DO prediction model establishment method based on BP neural network optimization |
| CN107741324A (en) * | 2017-10-13 | 2018-02-27 | 北京工业大学 | A kind of housing washer fault section diagnosis method |
| CN109297533A (en) * | 2018-09-20 | 2019-02-01 | 北京恩福健康科技有限公司 | A kind of method of precise measurement skin surface temperature and humidity |
| CN110047015A (en) * | 2019-04-22 | 2019-07-23 | 水利部信息中心 | A kind of water total amount prediction technique merging KPCA and thinking Optimized BP Neural Network |
| CN111598435A (en) * | 2020-05-14 | 2020-08-28 | 北京工业大学 | Quality trend prediction method based on adaptive feature selection and improved thought evolution algorithm |
| CN112232404A (en) * | 2020-10-13 | 2021-01-15 | 中国铁路上海局集团有限公司南京供电段 | Reliability calculation method and system based on historical abnormity and operation and maintenance information of railway power supply equipment |
| CN113011463A (en) * | 2021-02-25 | 2021-06-22 | 上海工程技术大学 | Chinese herbal medicine equipment bearing service life prediction method based on improved GA-BP |
| CN113092115A (en) * | 2021-04-09 | 2021-07-09 | 重庆大学 | Digital twin model construction method of digital-analog combined drive full-life rolling bearing |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5995910A (en) * | 1997-08-29 | 1999-11-30 | Reliance Electric Industrial Company | Method and system for synthesizing vibration data |
| CN101576443A (en) * | 2009-06-16 | 2009-11-11 | 北京航空航天大学 | Life prediction method of accelerated life test based on grey RBF neural network |
| CN106022549A (en) * | 2016-07-28 | 2016-10-12 | 兰州理工大学 | Short term load predication method based on neural network and thinking evolutionary search |
-
2016
- 2016-11-10 CN CN201610988210.5A patent/CN106769030A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5995910A (en) * | 1997-08-29 | 1999-11-30 | Reliance Electric Industrial Company | Method and system for synthesizing vibration data |
| CN101576443A (en) * | 2009-06-16 | 2009-11-11 | 北京航空航天大学 | Life prediction method of accelerated life test based on grey RBF neural network |
| CN106022549A (en) * | 2016-07-28 | 2016-10-12 | 兰州理工大学 | Short term load predication method based on neural network and thinking evolutionary search |
Non-Patent Citations (3)
| Title |
|---|
| 张以帅等: "基于MEA优化BP神经网络的天然气短期负荷预测", 《自动化与仪表》 * |
| 申中杰等: "基于相对特征和多变量支持向量机的滚动轴承剩余寿命预测", 《机械工程学报》 * |
| 阙子俊等: "基于UKF的轴承剩余寿命预测方法研究", 《仪器仪表学报》 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107358021B (en) * | 2017-06-01 | 2020-07-28 | 华南理工大学 | DO prediction model establishment method based on BP neural network optimization |
| CN107358021A (en) * | 2017-06-01 | 2017-11-17 | 华南理工大学 | DO prediction model establishment method based on BP neural network optimization |
| CN107292873A (en) * | 2017-06-29 | 2017-10-24 | 西安工程大学 | A kind of close degree detecting method of porcelain insulator ash based on color character |
| CN107741324A (en) * | 2017-10-13 | 2018-02-27 | 北京工业大学 | A kind of housing washer fault section diagnosis method |
| CN107741324B (en) * | 2017-10-13 | 2019-05-24 | 北京工业大学 | A kind of housing washer fault section diagnosis method |
| CN109297533A (en) * | 2018-09-20 | 2019-02-01 | 北京恩福健康科技有限公司 | A kind of method of precise measurement skin surface temperature and humidity |
| CN109297533B (en) * | 2018-09-20 | 2021-04-13 | 北京恩福健康科技有限公司 | Method for accurately measuring skin surface temperature and humidity |
| CN110047015A (en) * | 2019-04-22 | 2019-07-23 | 水利部信息中心 | A kind of water total amount prediction technique merging KPCA and thinking Optimized BP Neural Network |
| CN111598435A (en) * | 2020-05-14 | 2020-08-28 | 北京工业大学 | Quality trend prediction method based on adaptive feature selection and improved thought evolution algorithm |
| WO2021227406A1 (en) * | 2020-05-14 | 2021-11-18 | 北京工业大学 | Quality trend prediction method based on adaptive feature selection and improved mind evolutionary algorithim |
| CN111598435B (en) * | 2020-05-14 | 2023-08-04 | 北京工业大学 | Quality trend prediction method based on self-adaptive feature selection and improved thinking evolutionary algorithm |
| CN112232404A (en) * | 2020-10-13 | 2021-01-15 | 中国铁路上海局集团有限公司南京供电段 | Reliability calculation method and system based on historical abnormity and operation and maintenance information of railway power supply equipment |
| CN112232404B (en) * | 2020-10-13 | 2023-12-22 | 中国铁路上海局集团有限公司南京供电段 | Reliability calculation method and system based on historical abnormality and operation and maintenance information of railway power supply equipment |
| CN113011463A (en) * | 2021-02-25 | 2021-06-22 | 上海工程技术大学 | Chinese herbal medicine equipment bearing service life prediction method based on improved GA-BP |
| CN113092115A (en) * | 2021-04-09 | 2021-07-09 | 重庆大学 | Digital twin model construction method of digital-analog combined drive full-life rolling bearing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106769030A (en) | A kind of bearing state tracking and Forecasting Methodology based on MEA BP neural network algorithms | |
| CN105354587B (en) | A kind of method for diagnosing faults of wind-driven generator group wheel box | |
| CN104819846B (en) | Rolling bearing sound signal fault diagnosis method based on short-time Fourier transform and sparse laminated automatic encoder | |
| US8781982B1 (en) | System and method for estimating remaining useful life | |
| CN103995237A (en) | Satellite power supply system online fault diagnosis method | |
| CN113609955A (en) | Three-phase inverter parameter identification method and system based on deep learning and digital twinning | |
| CN104463359A (en) | Dredging operation yield prediction model analysis method based on BP neural network | |
| CN107392304A (en) | A kind of Wind turbines disorder data recognition method and device | |
| CN106934199A (en) | A kind of psychological pressure appraisal procedure and device | |
| CN105183841A (en) | Recommendation method in combination with frequent item set and deep learning under big data environment | |
| CN105615901A (en) | Emotion monitoring method and system | |
| CN106910078A (en) | Risk identification method and device | |
| DE112018005922T5 (en) | DEVICE, METHOD AND PROGRAM FOR THE PREDICTION OF A DISEASE OUTBACK RISK | |
| CN111292020A (en) | Power grid real-time operation risk assessment method and system based on random forest | |
| CN109917200B (en) | Train traction converter fault diagnosis method, system, equipment and storage medium | |
| CN107391385A (en) | A kind of construction method of Generation of software test case technology | |
| CN106618611A (en) | Sleeping multichannel physiological signal-based depression auxiliary diagnosis method and system | |
| CN108630228A (en) | Sound quality recognition methods, device, system and vehicle | |
| KR20210023631A (en) | System and method for improving development disorder using deep learning module | |
| CN102682180A (en) | Evaluation method for performance degradation of rotary mechanical equipment | |
| CN114492831A (en) | Method and device for generating federal learning model | |
| CN108399434A (en) | The analyzing and predicting method of the higher-dimension time series data of feature based extraction | |
| CN109873425A (en) | Electric system based on deep learning and user behavior adjusts trend method and system | |
| CN105701280A (en) | Southern America white-leg shrimp pond culture water quality prediction method | |
| CN102034111A (en) | Method for identifying and detecting aircraft structural damage conditions in diversified way |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170531 |
|
| RJ01 | Rejection of invention patent application after publication |