CN106202668B - Complex equipment quality risk appraisal procedure based on quality problems data and reverse conduction neural network - Google Patents
Complex equipment quality risk appraisal procedure based on quality problems data and reverse conduction neural network Download PDFInfo
- Publication number
- CN106202668B CN106202668B CN201610512085.0A CN201610512085A CN106202668B CN 106202668 B CN106202668 B CN 106202668B CN 201610512085 A CN201610512085 A CN 201610512085A CN 106202668 B CN106202668 B CN 106202668B
- Authority
- CN
- China
- Prior art keywords
- quality
- equipment
- data
- quality risk
- risk
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/30—Circuit design
- G06F30/36—Circuit design at the analogue level
- G06F30/367—Design verification, e.g. using simulation, simulation program with integrated circuit emphasis [SPICE], direct methods or relaxation methods
-
- 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
Abstract
The present invention relates to the complex equipment quality risk appraisal procedures based on quality problems data and reverse conduction neural network, and steps are as follows: step 1: decomposing complex equipment structure.Step 2: clear quality risk is constituted.Step 3: collecting and arrange history quality problem data.Step 4: building and training BpNN equipment quality risk evaluation model.Step 5: collecting the data to be analyzed with arrangement equipment to be assessed.Step 6: assessment complex equipment quality risk.The method of the present invention uses input of the quality problems data as risk assessment, more can reflect to objective reality risk probability existing for current equipment quality, provide more scientific and effective foundation for administrative decision.Reverse conduction neural network substantially realizes one from the mapping function for being input to output, with self-learning capability, without artificially assigning index weights, it can greatly reduce artificial subjective factor by the automatic weight assignment for realizing indices of training and be interfered caused by equipment quality risk evaluation result.
Description
[technical field]
The present invention provides a kind of based on quality problems data (Data of Quality Problems, DoQP) and reversed biography
Lead the complex equipment quality risk appraisal procedure of neural network (Bp Neural Network, BpNN), i.e., it is a kind of to be based on DoQP-
The complex equipment quality risk appraisal procedure of BpNN ("-" indicates "and", by DoQP in conjunction with BpNN simultaneous), belongs to quality engineering
Technical field.
[background technique]
Modern weapons equipment has high complexity feature, is substantially to pass through " centre by various different function systems
The integrated architectonical system of part " (System of System, SoS), development, production process are also rather complicated.Meanwhile it is large-scale
Complex equipment cost is high, once quality risk, which occurs, will not only cause immeasurable economic loss, notably will lead to people
Body injures and deaths may threaten national security.Therefore, the clear of complex equipment quality risk is recognized to carry out equipment in order to realize
The work such as quality risk control, promotion, improvement just must carry out network analysis, measurement and assessment to complex equipment quality risk.
The present invention is having extensively studied various measurements with after decomposition method, is based on quality problems data (Data of
Quality Problems, DoQP) and reverse conduction neural network (Bp Neural Network, BpNN) method on the basis of
It is improved, has invented a kind of measurement and the method for decomposition for complex equipment mass property, the i.e. complicated dress of DoQP-BpNN
The measurement and decomposition method of standby mass property.
[summary of the invention]
(1) purpose
The object of the present invention is to provide a kind of based on quality problems data and reverse conduction neural network (DoQP-BpNN)
Complex equipment quality risk appraisal procedure, it is to divide for the deficiency of complex equipment quality risk assessment qualitative analysis summarizing
On the basis of analysing history quality problem information, complex equipment is decomposed, low-level mass of system problem is occurred not true
Determine degree to be differentiated, a kind of versatility, the stronger complex equipment quality risk appraisal procedure of operability are provided, for complicated dress
Standby quality risk assessment provides solution.
(2) technical solution
The present invention is the complex equipment quality wind based on quality problems data and reverse conduction neural network (DoQP-BpNN)
Dangerous appraisal procedure, the specific steps of which are as follows:
Step 1: decomposing complex equipment structure.According to the degree of integration of complex equipment, being decomposed can be measured most
It is low system-level, determine more detailed system structure composition.
Step 2: clear quality risk is constituted.Based on cost and quality problems probability of happening, the intension of quality risk is specified
And it constitutes.
Step 3: collecting and arrange history quality problem data.Based on equipment technology principle, with fault mode and influence
The mass analysis methods such as (FMEA), failure tree analysis (FTA) (FTA) are analyzed, identify the Key Performance Indicator and quality wind of low-level system
Dangerous source, the quality problems set of data samples of formation.
Step 4: building and training BpNN equipment quality risk evaluation model.According to the Key Performance Indicator of low-level system
Quantity and quality risk source quantity determine the input node quantity and output node number of BpNN equipment quality risk evaluation model
Amount constructs BpNN equipment quality risk evaluation model；Based on quality problems set of data samples data, programmed with Matlab, it is right
BpNN equipment quality risk evaluation model is trained.
Step 5: collecting the data to be analyzed with arrangement equipment to be assessed.According to identified Key Performance Indicator, treat
Each low-level system of assessment equipment is tested, and low-level system core performance achievement data is compiled.
Step 6: assessment complex equipment quality risk.With the BpNN equipment quality risk assessment mould that step 4 training is mature
Type, the Key Performance Indicator data to be assessed that step 5 is compiled are programmed as mode input data with Matlab, right
Equipment quality risk is assessed.
Wherein, the complex equipment STRUCTURE DECOMPOSITION in step 1 is carried out from structure to complex equipment by sufficiently investigating
It decomposes, determines the system-level of the lowest hierarchical level of equipment, i.e., can not estimate or lose structural function after lowest hierarchical level system decomposition.
Wherein, clear quality risk described in step 2 is constituted, the specific implementation process is as follows:
(1) quality risk T is made of expected loss C and quality problems probability of happening P, and relationship is as follows:
(number that i is low-level system)
(2) according to equipment cost, share each low-level system and quality problems bring expected loss C occursi；
(3) according to performance indicator, the probability P that quality problems occur for low-level system is assessedi；
Wherein, collection described in step 3 and arrangement history quality problem data, the specific implementation process is as follows:
(1) history quality problem information is collected, with failure mode and effect analysis (FMEA), failure tree analysis (FTA) (FTA),
Analytical equipment quality fault mode and mechanism；
(2) when collecting correspondence problem generation, the performance indicator and parameter of each low-level system；
(3) expert graded is used, in conjunction with equipment quality risk source, " people (Man), machine (Machine), material
(Material), method (Method), ring (Environment) " (4M1E) is sorted out the reason of induction quality problems, is beaten
Point；
(4) by the performance indicator data of each low-level system in conjunction with expert estimation result, quality problems data sample is formed
This collection.
Wherein, building described in step 4 and training BpNN equipment quality risk evaluation model, specific implementation process is such as
Under:
(1) according to the Key Performance Indicator quantity of low-level system and quality risk source quantity, BpNN equipment quality is determined
The input node quantity and output node quantity of risk evaluation model；
(2) formulaDetermine that BpNN model node in hidden layer, r are node in hidden layer, n
For input layer number, m is output layer number of nodes, and a is that experience differentiates constant, forms BpNN equipment quality risk evaluation model；
(3) quality problems set of data samples data are based on, are programmed with Matlab, to BpNN equipment quality risk assessment mould
Type is trained, and when model training error drops to 0.001, model training is mature, can be used for equipment quality risk assessment.
(3) beneficial effect
Of the invention is a kind of based on the assessment of the complex equipment quality risk of quality problems data and reverse conduction neural network
Method, advantage and effect are:
(1) more previous methods of risk assessment, the present invention use input of the quality problems data as risk assessment, can
More reflect to objective reality risk probability existing for current equipment quality, for administrative decision provide it is more scientific and effective according to
According to.
(2) reverse conduction neural network substantially realizes one from the mapping function for being input to output, has self study
Ability does not have to artificial imparting index weights, can greatly reduce people by the automatic weight assignment for realizing indices of training
It is interfered caused by equipment quality risk evaluation result for subjective factor.
[Detailed description of the invention]
Complex equipment quality risk estimation flow schematic diagram of the Fig. 1 based on DoQP-BpNN.
Fig. 2 complete equipment quality problem probability of happening frame diagram.
Fig. 3 BpNN equipment quality risk evaluation model.
[specific embodiment]
The present invention is a kind of complex equipment quality risk appraisal procedure based on DoQP-BpNN, the assessment provided according to Fig. 1
Specific implementation step is further described in flow diagram:
Step 1: decomposing complex equipment structure.It is adequately investigated, is started with from complex equipment design level, from knot
Complex equipment is successively decomposed on structure, determines that equipment module system is constituted, until lowest hierarchical level system can be estimated and tie
Structure sexual function is not lost.
Step 2: clear quality risk is constituted.The supply chain low side supplied along raw material, components traces lowest hierarchical level system
The manufacturing cost of system, meanwhile, the quality problems probability of happening that risk sources such as " people, machine, material, method, rings " are led to is compiled, is used
Above-mentioned formula T=Σ Ci·Pi, calculate the specific value of quality risk.
Step 3: collecting and arrange history quality problem data.Based on equipment technology principle, with fault mode and influence
The mass analysis methods such as (FMEA), failure tree analysis (FTA) (FTA) are analyzed, identify the Key Performance Indicator and quality wind of low-level system
Dangerous source, the quality problems set of data samples of formation.
(1) identification of lowest hierarchical level system core functional performance index value, acquisition.
According to the Functional Design schematic diagram of equipment, characteristic design framework figure or the form of quality process frame diagram by elder's top layer
Function, performance carry out STRUCTURE DECOMPOSITION, in conjunction with design parameter, identification, the key function performance parameter for obtaining lowest hierarchical level system
Index value obtains the input data for training BpNN equipment quality risk evaluation model.
(2) risk source data obtains.
With mass analysis methods such as failure mode and effect analysis (FMEA), failure tree analysis (FTA)s (FTA), explores quality and ask
The risk source occurred is inscribed, is divided with expert graded according to complete equipment quality problem probability of happening frame diagram (such as Fig. 2)
A situation arises gives a mark for the other quality problems to risk sources such as " people, machine, material, method, rings ", obtains equipment quality risk source number
According to, obtain train BpNN equipment quality risk evaluation model output data.
Wherein, the descriptive definition of complete quality safety risk class frame (Fig. 2) is: by it is limited, determine event structure
At event set state (whether quality security problem occurring) only have certainty feature, i.e., equipment quality problem occur probability
It is 0, quality risk rank is 0；Have single order not true under known event space and with the event set of stable probabilistic relation
Qualitative features, quality risk rank are 1,2,3 respectively from low to high；It is closed under known event space and with unstable probability
The event set state of system has second order uncertainty feature, and quality risk rank is 4,5,6 respectively from low to high；In unknown event
Event set state under space has three rank uncertainty features, and quality risk rank is 7,8,9 respectively from low to high.
(3) the quality problems set of data samples of BpNN equipment quality risk evaluation model obtains.
Lowest hierarchical level system core functional performance index value and risk source data are corresponded by certain logic rules,
Form independent sample data；In view of different channels obtain data dimension disunity, and then with Spss to these data into
Row normalized, the dimension of unified lowest hierarchical level system core functional performance index value and risk source data, exports text
Formatted file forms the quality problems set of data samples of BpNN equipment quality risk evaluation model.
Step 4: building and training BpNN equipment quality risk evaluation model.According to the Key Performance Indicator of low-level system
Quantity and quality risk source quantity determine the input node quantity and output node number of BpNN equipment quality risk evaluation model
Amount constructs BpNN equipment quality risk evaluation model；Based on quality problems set of data samples data, programmed with Matlab, it is right
BpNN equipment quality risk evaluation model is trained.
(1) the input node quantity of BpNN equipment quality risk evaluation model is determined.
According to lowest hierarchical level system core functional performance index quantity, BpNN equipment quality risk evaluation model can be determined
Input layer number, as shown in figure 3, using each lowest hierarchical level systematic survey index number in index quantity as model
Input node.
(2) the output node quantity of BpNN equipment quality risk evaluation model is determined.
Output node is generally determined by the desired output of assessment result.In previous risk assessment, generally only it is arranged one
A desired output.But the present invention not only needs to assess complex equipment quality risk, but also further to analyze risk source.
Therefore, the present invention compares 5 risk sources such as " people, machine, material, method, rings ", determines the output of BpNN equipment quality risk evaluation model
Number of nodes is 5.
(3) the hidden layer node quantity of BpNN equipment quality risk evaluation model is determined.
The hidden layer neuron number of BpNN equipment quality risk evaluation model selects the not only section with input and output layer
It counts related, more with the characteristic of form and sample data of the complexity of problem to be solved and transfer function etc. because being known as
It closes, is a sufficiently complex problem.If hidden layer neuron is very few, model cannot may train at all or model performance very
Difference；If hidden layer neuron is too many, although the systematic error of model can be made to reduce, the model training time is increased, is also easy to
Training is caused to fall into local minimum point, and the immanent cause of " over-fitting " occurs in training.Appearance when to avoid training as far as possible
" over-fitting " phenomenon, while guaranteeing sufficiently high network performance and generalization ability, determine the substantially former of hidden layer neuron number
It is then: takes structure as compact as possible under the premise of meeting required precision, i.e., under the premise of being able to solve problem, in addition 1
Accelerate the decrease speed of error to 2 neurons.Therefore, the present invention calculates public with hidden layer nodeDetermine that BpNN model node in hidden layer, r are node in hidden layer, n is input layer number, and m is
Output layer number of nodes, a 2.
Finally, pass through calculating, it is determined that BpNN equipment quality risk evaluation model, as shown in Figure 3.
(3) the BpNN equipment quality risk evaluation model training based on Matlab
Invention selects training sample of 2/3rds samples of quality problems set of data samples as model, remaining sample herein
This test samples as model；Then, model is trained with Matlab programming, initializes connection weight and threshold value,
With newff function auto-initiation BpNN equipment quality risk evaluation model, the order of traingdx training pattern is called such as
Under:
Training interval net.trainParam.show=25；
E-learning rate net.trainParam.lr=0.05；
Momentum coefficient net.trainParam.mc=25；
The maximum frequency of training net.trainParam.epochs=1000 of network；
Network training precision target value net.trainParam.goal=0.001.
When " TRAINGDX, Performance goal met. " printed words, show that model training is complete to Matlab interface display
At BpNN equipment quality risk evaluation model can be used for actual complex equipment quality risk assessment.
Step 5: collecting the data to be analyzed with arrangement equipment to be assessed.According to identified Key Performance Indicator, control
The data structure of quality problems set of data samples tests each lowest hierarchical level system of equipment to be assessed, collects low-level
System core performance achievement data, does normalized, and the key performance of each lowest hierarchical level system of unified equipment to be assessed refers to
Numerical value dimension is marked, equipment data to be assessed are formed.
Wherein, the method for normalized is operated by following algorithm.
When index value is bigger, when quality risk is higher, x*Numerical value according to formula (1) calculate:
When index value is smaller, when quality risk is higher, x*Numerical value according to formula (2) calculate:
Wherein, x*Indicate index xiStandardized value after normalized, ximinIndicate the minimum value of i-th of index,
ximaxRefer to the maximum value of i-th of index.
Step 6: assessment complex equipment quality risk.It is commented with the mature BpNN equipment quality risk of training in step (4)
Model is estimated, using the Key Performance Indicator data to be assessed compiled in step (5) as mode input data, with Matlab
Programming, assesses equipment quality risk, and model output result is the assessed value of equipment quality risk.
Wherein, Matlab programming specific steps are as follows:
(1) input vector for defining training sample is P and target vector T；
(2) a new BpNN equipment quality risk evaluation model is created:
Net=newff (minmax (P), [3,1], ' tansig', ' purelin'}, ' traingdx')
(3) input layer weight and threshold value are set:
InputWeights=net.IW { 1,1 }；
Inputbias=net.b { 1 }
(4) network layer weight and threshold value are set:
LayerWeights=net.LW { 2,1 }；
Layerbias=net.b { 2 }
(5) training parameter is set:
Net.trainParam.show=50；
Net.trainParam.lr=0.05；
Net.trainParam.mc=0.9；
Net.trainParam.epochs=1000；
Net.trainParam.goal=1e-3；
(6) traingdx algorithm training BpNN equipment quality risk evaluation model is called:
[net, tr]=train (net, P, T)
(7) BpNN equipment quality risk evaluation model is emulated:
A=sim (net, P).
Claims (4)
1. a kind of complex equipment quality risk appraisal procedure based on quality problems data and reverse conduction neural network, specific
Steps are as follows:
Step 1 decomposes complex equipment structure: according to the degree of integration of complex equipment, being decomposed the minimum system that can be measured
Irrespective of size determines that system structure forms；
Step 2, clear quality risk are constituted: being based on cost and quality problems probability of happening, specified the intension and structure of quality risk
At；
Step 3, collect with arrange history quality problem data: be based on equipment technology principle, with failure mode and effect analysis,
The mass analysis method of failure tree analysis (FTA) identifies Key Performance Indicator and the quality risk source of low-level system, the quality of formation
Problem data sample set；
Step 4, building are with training BpNN equipment quality risk evaluation model: according to the Key Performance Indicator quantity of low-level system
And quality risk source quantity, determine the input node quantity and output node quantity of BpNN equipment quality risk evaluation model, structure
Build BpNN equipment quality risk evaluation model；It based on quality problems set of data samples data, is programmed with Matlab, BpNN is filled
Standby quality risk assessment models are trained；
Step 5, the data to be analyzed for collecting and arranging equipment to be assessed: according to identified Key Performance Indicator, to be assessed
Each low-level system of equipment is tested, and low-level system core performance achievement data is compiled；
Step 6, assessment complex equipment quality risk:, will with the BpNN equipment quality risk evaluation model that step 4 training is mature
The Key Performance Indicator data to be assessed that step 5 is compiled are programmed as mode input data with Matlab, to equipment matter
Amount risk is assessed.
2. the complex equipment quality risk according to claim 1 based on quality problems data and reverse conduction neural network
Appraisal procedure, it is characterised in that: clear quality risk described in step 2 is constituted, the specific implementation process is as follows:
(1) quality risk T is made of expected loss C and quality problems probability of happening P, and relationship is as follows:
Wherein i is the number of low-level system；
(2) according to equipment cost, share each low-level system and quality problems bring expected loss C occursi；
(3) according to performance indicator, the probability P that quality problems occur for low-level system is assessedi。
3. the complex equipment quality risk according to claim 1 based on quality problems data and reverse conduction neural network
Appraisal procedure, it is characterised in that: collection described in step 3 and arrangement history quality problem data, specific implementation process is such as
Under:
(1) history quality problem information is collected, with failure mode and effect analysis, failure tree analysis (FTA), the event of analytical equipment quality
Barrier mode and mechanism；
(2) when collecting correspondence problem generation, the performance indicator and parameter of each low-level system；
(3) expert graded is used, in conjunction with equipment quality risk source, the reason of induction quality problems is sorted out, is given a mark；
(4) by the performance indicator data of each low-level system in conjunction with expert estimation result, quality problems set of data samples is formed.
4. the complex equipment quality risk according to claim 1 based on quality problems data and reverse conduction neural network
Appraisal procedure, it is characterised in that: building described in step 4 and training BpNN equipment quality risk evaluation model, it is specific real
Existing process is as follows:
(1) according to the Key Performance Indicator quantity of low-level system and quality risk source quantity, BpNN equipment quality risk is determined
The input node quantity and output node quantity of assessment models；
(2) formulaDetermine that BpNN model node in hidden layer, r are node in hidden layer, n is input
Node layer number, m are output layer number of nodes, and a is that experience differentiates constant, form BpNN equipment quality risk evaluation model；
(3) be based on quality problems set of data samples data, programmed with Matlab, to BpNN equipment quality risk evaluation model into
Row training, when model training error drops to 0.001, model training is mature, can be used for equipment quality risk assessment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610512085.0A CN106202668B (en) | 2016-07-01 | 2016-07-01 | Complex equipment quality risk appraisal procedure based on quality problems data and reverse conduction neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610512085.0A CN106202668B (en) | 2016-07-01 | 2016-07-01 | Complex equipment quality risk appraisal procedure based on quality problems data and reverse conduction neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106202668A CN106202668A (en) | 2016-12-07 |
| CN106202668B true CN106202668B (en) | 2019-02-19 |
Family
ID=57462877
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610512085.0A Expired - Fee Related CN106202668B (en) | 2016-07-01 | 2016-07-01 | Complex equipment quality risk appraisal procedure based on quality problems data and reverse conduction neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106202668B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108334907B (en) * | 2018-02-09 | 2019-05-17 | 哈尔滨工业大学（威海） | A kind of complex equipment point method for detecting abnormality and system based on deep learning |
| CN109255389B (en) * | 2018-09-28 | 2022-03-25 | 中国科学院长春光学精密机械与物理研究所 | Equipment evaluation method, device, equipment and readable storage medium |
| CN113516325A (en) * | 2020-04-10 | 2021-10-19 | 中国农业机械化科学研究院 | Information fusion-based combine harvester manufacturing quality analysis decision method and system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102129521A (en) * | 2011-03-16 | 2011-07-20 | 北京航空航天大学 | Improved strategy to task-quality function deployment (ISTT-QFD)-based complicated equipment quality characteristic measurement and decomposition method |
| CN103678952A (en) * | 2013-11-14 | 2014-03-26 | 昆明理工大学 | Elevator risk evaluation method |
| WO2016160539A1 (en) * | 2015-03-27 | 2016-10-06 | Equifax, Inc. | Optimizing neural networks for risk assessment |
-
2016
- 2016-07-01 CN CN201610512085.0A patent/CN106202668B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102129521A (en) * | 2011-03-16 | 2011-07-20 | 北京航空航天大学 | Improved strategy to task-quality function deployment (ISTT-QFD)-based complicated equipment quality characteristic measurement and decomposition method |
| CN103678952A (en) * | 2013-11-14 | 2014-03-26 | 昆明理工大学 | Elevator risk evaluation method |
| WO2016160539A1 (en) * | 2015-03-27 | 2016-10-06 | Equifax, Inc. | Optimizing neural networks for risk assessment |
Non-Patent Citations (4)
| Title |
|---|
| Risk Management Game Method of the Weapons Project Based on BP Neural Network;Y Yu;《International Conference on Information Technology》;20111231(第1期);第113-117页 |
| 关于我国武器装备质量监督的思考;高洋 ， 姚洪涛 ， 王国松;《质量与可靠性》;20160303(第6期);第26-28，43页 |
| 基于BP神经网络的武器装备研制项目风险评估模型;宋彦学，张志峰，霍亮;《先进制造与管理》;20070709(第5期);第40-42页 |
| 武器装备研制项目的技术风险评估;徐哲，冯允成，鲁大伟;《系统工程与电子技术》;20050728(第5期);第1123-1127页 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106202668A (en) | 2016-12-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106769052B (en) | A kind of mechanical system rolling bearing intelligent failure diagnosis method based on clustering | |
| CN106895975B (en) | Method for Bearing Fault Diagnosis based on Stacked SAE deep neural network | |
| CN109857835A (en) | A kind of adaptive network security knowledge assessment method based on cognitive diagnosis theory | |
| CN110262463B (en) | Rail transit platform door fault diagnosis system based on deep learning | |
| CN104712542B (en) | A kind of reciprocating compressor sensitive features based on Internet of Things are extracted and method for diagnosing faults | |
| CN105117602B (en) | A kind of metering device running status method for early warning | |
| CN108038300B (en) | Optical fiber state evaluation method based on improved membership function combined with neural network | |
| CN101587155B (en) | Oil soaked transformer fault diagnosis method | |
| CN106779069A (en) | A kind of abnormal electricity consumption detection method based on neutral net | |
| CN110455537A (en) | A kind of Method for Bearing Fault Diagnosis and system | |
| CN108197648A (en) | A kind of Fault Diagnosis Method of Hydro-generating Unit and system based on LSTM deep learning models | |
| CN108363690A (en) | Dialog semantics Intention Anticipation method based on neural network and learning training method | |
| CN107272644B (en) | The DBN network fault diagnosis method of latent oil reciprocating oil pumping unit | |
| CN107274011A (en) | The equipment state recognition methods of comprehensive Markov model and probability net | |
| CN108664690A (en) | Long-life electron device reliability lifetime estimation method under more stress based on depth belief network | |
| CN109242147A (en) | Signal fused fan condition prediction technique based on Bp neural network | |
| CN110929918A (en) | 10kV feeder line fault prediction method based on CNN and LightGBM | |
| CN105095963A (en) | Method for accurately diagnosing and predicting fault of wind tunnel equipment | |
| CN106202668B (en) | Complex equipment quality risk appraisal procedure based on quality problems data and reverse conduction neural network | |
| CN109597401A (en) | A kind of equipment fault diagnosis method based on data-driven | |
| CN108537259A (en) | Train control on board equipment failure modes and recognition methods based on Rough Sets Neural Networks model | |
| CN109407654A (en) | A kind of non-linear causality analysis method of industrial data based on sparse depth neural network | |
| CN111680875B (en) | Unmanned aerial vehicle state risk fuzzy comprehensive evaluation method based on probability baseline model | |
| CN107991876A (en) | Aero-engine condition monitoring data creation method based on production confrontation network | |
| CN107545307A (en) | Predicting model for dissolved gas in transformer oil method and system based on depth belief network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190219 Termination date: 20200701 |