CN109029699A - A kind of online method for detecting abnormality of transformer vibration - Google Patents
A kind of online method for detecting abnormality of transformer vibration Download PDFInfo
- Publication number
- CN109029699A CN109029699A CN201810598845.3A CN201810598845A CN109029699A CN 109029699 A CN109029699 A CN 109029699A CN 201810598845 A CN201810598845 A CN 201810598845A CN 109029699 A CN109029699 A CN 109029699A
- Authority
- CN
- China
- Prior art keywords
- sample
- newly
- training
- increased data
- increased
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
- G01H17/00—Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves, not provided for in the preceding groups
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
For how based on vibratory drilling method analysis running state of transformer and carrying out fault diagnosis at present, the embodiment of the invention discloses a kind of transformers to vibrate online method for detecting abnormality, this method is the newly-increased data sample obtained in preset time period about running state of transformer vibration signal, the characteristic parameter of newly-increased data sample is extracted based on wavelet packet analysis, the study of newly-increased data sample is completed based on quick algorithm of convex hull, training updates single class anomaly detector model, updated single class anomaly detector model is called, online abnormality detection is carried out to current transformer vibration.Since the application only learns the newly-increased data sample in preset time period using Incremental Learning Algorithm, so newly-increased data can be handled timely and effectively, newly-increased training sample can be learnt in real time online, it realizes the quick upgrading of detection model, while demand of the model modification to time and space can be reduced.
Description
Technical field
The present invention relates to power domains more particularly to a kind of transformer to vibrate online method for detecting abnormality.
Background technique
Power transformer is the important equipment in electric system, safety and economic benefit of the operating status to electric system
There is important influence.Transformer surface vibration is derived mainly from winding and core vibration under electric current and voltage drive, theory point
Analysis and practical experience show that winding and iron core working condition can be analyzed by transformer vibration signal and carry out fault diagnosis.
Vibratory drilling method analysis running state of transformer and fault diagnosis, domestic and international researcher have done a lot of research work, achieve quite
Achievement.Due to transformer vibration signal feature extraction be vibratory drilling method analysis running state of transformer and fault diagnosis premise and
Basis.But the data of existing research achievement are derived from laboratory or testing transformer emulation mostly, due to transformer surface vibration
By multiple factors complex effects, so transformer surface vibration signals and theory analysis and laboratory test item in actual motion
The transformer surface vibration signals difference obtained under part is obvious.
In actual production, transformer vibration generates flow data, and in these flow datas can include new sample knowledge,
How flow data is timely and effectively handled, is one of the difficult point of transformer exception detection.
Summary of the invention
In order to solve the above technical problems, the invention proposes a kind of transformers to vibrate online method for detecting abnormality, this method
It is the newly-increased data sample obtained in preset time period about running state of transformer vibration signal, is extracted based on wavelet packet analysis
The characteristic parameter of newly-increased data sample completes the study of newly-increased data sample based on quick algorithm of convex hull, and it is different that training updates single class
Normal detector model, calls updated single class anomaly detector model, carries out online abnormality detection to current transformer vibration.
Since the application only learns the newly-increased data sample in preset time period using Incremental Learning Algorithm, so can be timely
It is effective to handle newly-increased data, newly-increased training sample can be learnt in real time online, realize the quick liter of detection model
Grade, while demand of the model modification to time and space can be reduced.
The embodiment of the invention provides following technical solutions:
A kind of online method for detecting abnormality of transformer vibration, comprising:
Obtain the newly-increased data sample in preset time period about running state of transformer vibration signal；
The characteristic parameter of newly-increased data sample is extracted based on wavelet packet analysis；
The study of newly-increased data sample is completed based on quick algorithm of convex hull, training updates single class anomaly detector model；
Updated single class anomaly detector model is called, online abnormality detection is carried out to current transformer vibration.
Wherein, the study of newly-increased data sample is completed based on quick algorithm of convex hull, training updates single class anomaly detector mould
Type specifically includes:
If initial training sample A0, increasing sample newly is B={ B1,B2,…,Bn, andNew instruction
Practice sample Ai(i ∈ n),
1) initial training sample A0Training obtains Support Vector data description model Ω0, supporting vector SV0；
2) newly-increased sample B is addedi(i ∈ n), finds out BiThe middle sample for violating KKT condition, is denoted asIfThen return
Return Ωi-1；Otherwise sample A is calculatedi-1Except supporting vector SVi-1Sample A afterwardsi-1, i.e. Ai'-1=Ai-1-SVi-1, and utilize quickly convex
Packet algorithm calculates Ai'-1Shell vector Ci-1；
3) willNew Support Vector data description model Ω is obtained as the training of new training samplei,
I=i+1；As i > n, algorithm is terminated, when 2) i≤n is gone to step.
Wherein, KKT condition is expressed as follows:
Wherein, αi(i ∈ n) is Lagrange multiplier, and v is used to balance volume of hypersphere and training error, referred to as balance parameters,
R is that the radius of the minimal hyper-sphere of training sample is the centre of sphere, and a is that the radius of the minimal hyper-sphere of training sample is the centre of sphere, and z is to survey
Sample sheet, d2=| | z-a | |2。
Compared with prior art, above-mentioned technical proposal has the advantage that
Method provided by the embodiment of the present invention, the invention proposes a kind of transformers to vibrate online method for detecting abnormality,
This method is the newly-increased data sample obtained in preset time period about running state of transformer vibration signal, based on wavelet packet point
The characteristic parameter of newly-increased data sample is extracted in analysis, and the study of newly-increased data sample is completed based on quick algorithm of convex hull, and training updates
Single class anomaly detector model calls updated single class anomaly detector model, carries out to current transformer vibration online different
Often detection.Since the application only learns the newly-increased data sample in preset time period using Incremental Learning Algorithm, so
Newly-increased data can be timely and effectively handled, newly-increased training sample can be learnt in real time online, realize detection model
Quick upgrading, while demand of the model modification to time and space can be reduced.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is the present invention
Some embodiments for those of ordinary skill in the art without creative efforts, can also basis
These attached drawings obtain other attached drawings.
Fig. 1 is the schematic diagram that a kind of transformer vibrates online method for detecting abnormality.
Specific embodiment
As described in the background art, how running state of transformer is analyzed based on vibratory drilling method and carries out fault diagnosis, be
Those skilled in the art's technical problem urgently to be solved.
In order to solve the above-mentioned technical problem, the invention proposes a kind of transformers to vibrate online method for detecting abnormality, the party
Method is the newly-increased data sample obtained in preset time period about running state of transformer vibration signal, is mentioned based on wavelet packet analysis
The characteristic parameter for taking newly-increased data sample, the study of newly-increased data sample is completed based on quick algorithm of convex hull, and training updates single class
Anomaly detector model calls updated single class anomaly detector model, carries out online abnormal inspection to current transformer vibration
It surveys.Since the application only learns the newly-increased data sample in preset time period using Incremental Learning Algorithm, so can
Newly-increased data are timely and effectively handled, newly-increased training sample can be learnt in real time online, realize the fast of detection model
Speed upgrading, while demand of the model modification to time and space can be reduced.
Fig. 1 is the schematic diagram that a kind of transformer vibrates online method for detecting abnormality, which comprises
Step 101: obtaining the newly-increased data sample in preset time period about running state of transformer vibration signal.
In transformer actual moving process, the acquisition about running state of transformer vibration signal data sample is one
The process constantly accumulated forms vibration flow data, using the newly-increased data sample in preset time period as ephemeral data sample,
Then ephemeral data sample can be constantly updated.The training of anomaly detector model will also be related to large sample problem concerning study simultaneously.
If anomaly detector model upgrades and will learn again on the basis of original training sample and new training sample in each update
It practises, then can consume a large amount of computing resource and time.Therefore, on the basis of learning outcome before reservation, incremental learning
(Incremental Learning) algorithm can learn the knowledge of newly-increased sample, can be online in real time to newly-increased training sample
Learnt, realizes that anomaly detector model quickly upgrades, while demand of the model modification to time and space can be reduced.Cause
This, the application only learns the newly-increased data sample in preset time period using Incremental Learning Algorithm.
It should be noted that preset time period can be configured according to actual needs, such as can be set to one week,
Perhaps it can be set to one month or can be set to a season, the application does not do any restriction to this.
Step 102: the characteristic parameter of newly-increased data sample is extracted based on wavelet packet analysis.
Wavelet packet analysis can decompose running state of transformer original vibration signal on different frequency bands, and signal exists
Local energy on each frequency band can reflect the time-varying feature of signal frequency feature, have good time-frequency locating features and right
The adaptive ability of signal.The application is based on Wavelet Packet Analysis and establishes Faults by Vibrating extraction model, extracts characterization and becomes
The vibrational state characteristic parameter of depressor operating status.
Step 103: the study of newly-increased data sample is completed based on quick algorithm of convex hull, training updates single class anomaly detector
Model.
The application chooses Support Vector data description (Support Vector Domain Description, SVDD) mould
Type carries out online abnormality detection as single class anomaly detector, to transformer actual measurement vibration data.
The main thought of SVDD is one minimum sphere interface of searching in nucleus lesion, which should be as far as possible
All training samples are surrounded, and sample is classified and described with the interface, and the boundary of its description, it can be with
For reflecting the characteristic distributions of training sample.
Assuming that training sample X includes n target class sample, i.e. X={ x1,x2…,xn, in the higher dimensional space of mapping, ask
Out can be comprising the centre of sphere a and radius R of the minimal hyper-sphere of the training sample, then its optimization problem can indicate are as follows:
Wherein v is used to balance volume of hypersphere and training error, referred to as balance parameters, generally takes 0.1；ξiFor relaxation factor.
Using method of Lagrange multipliers, and introduce kernel function K (xi,xj) solve nonlinear problem, then willFormula is converted into dual problem:
Wherein αi(i ∈ n) is Lagrange multiplier.
Gaussian kernel function is chosen in the application:
Wherein s is Gaussian Bandwidth coefficient.
Pass throughFormula, available centre of sphere a and radius R:
Then αiThe sample of > 0 is known as supporting vector SV.Referred to as boundary supporting vector BSV, xk∈BSV。
For test sample z, if
Then think that test sample z is fallen into hypersphere, otherwise it is foreign peoples that receiving sample, which is target class,.
As all αiMeet Karush-Kuhn-Tucker (KKT) condition of objective function, so that it may be considered original equationOne solution.
KKT condition can be expressed as follows:
Transformer historical data sample collection is constructed SVDD model by the application.According to formulaNewly-increased data sample is detected, is judged whether
For exceptional sample, to realize abnormality detection.
Gaussian kernel is all made of in subsequent embodiment.By formulaKnow parameter s and training
Relative position between sample is related, therefore, the initial value of s can be arranged in the minimum value and maximum value of training sample spacing
Between, optimal value is determined by cross validation method later.
Theoretical by SVDD is introduced it is found that SVDD model found during training is played a decisive role to classification
The sample of supporting vector.Class near border sample is likely to become supporting vector during to sample training, therefore can be with
Non- boundary sample is carried out it is superseded, to further reduce training sample set scale.
After newly-increased data sample, the variation of supporting vector collection has following theorem:
If theorem 1, which increases newly to exist in sample, meets the sample of KKT condition, this part sample be not present certainly it is new support to
Amount；The sample of KKT condition is violated if it exists, then new supporting vector must be had by violating in the sample of KKT condition.
If theorem 2 increases the sample for existing in sample and violating KKT condition newly, non-supporting vector may switch in original sample
Supporting vector.
It, therefore, will by theorem 1 it is found that there is new supporting vector information in the sample for violating KKT condition in newly-increased sample
Whether newly-increased sample, which violates KKT condition, is screened, and the newly-increased sample for participating in training is reduced.By theorem 2 it is found that in incremental learning
The non-supporting vector in original sample can be converted into new supporting vector again later.
These may be converted to new supporting vector sample and be usually located at sample boundary.Shell vector is all positioned at training sample
The sample on class boundary in this, i.e. shell vector is training sample concave vertex.Quick algorithm of convex hull (Quickhull algorithm) is by instruction
Practice the point deleted inside convex closure of the screening at sample midpoint gradually, to quickly calculate shell vector.Therefore, in order not to losing
Training sample set scale is reduced while supporting vector information, the application is calculated using Quickhull algorithm except supporting vector
The shell vector of training sample guarantee the complete of supporting vector information.
The application is when learning newly-increased data sample, i.e., in incremental learning, the group of new training sample becomes former and supports
Vector violates the sample of KKT condition in the shell vector and newly-increased sample of former non-supporting vector sample.
The application be based on quick algorithm of convex hull newly-increased data sample is learnt, update abnormal detector model it is specific
Process are as follows:
Initial training sample A0, increasing sample newly is B={ B1,B2,…,Bn, andNew training
Sample Ai(i∈n)。
1) initial training sample A0Training obtains Support Vector data description model Ω0, supporting vector SV0；
2) newly-increased sample B is addedi(i ∈ n), finds out BiThe middle sample for violating KKT condition, is denoted as BiF, ifThen return
Return Ωi-1；Otherwise sample A is calculatedi-1Except supporting vector SVi-1Sample A afterwardsi-1, i.e. Ai'-1=Ai-1-SVi-1, and utilize
Quickhull algorithm calculates Ai'-1Shell vector Ci-1；
3) willNew Support Vector data description model Ω is obtained as the training of new training samplei,
Obtained the new centre of sphere and and radius:I=i+1；Work as i
When > n, algorithm is terminated, when 2) i≤n is gone to step.
In SVDD modeling process, need to solve standard quadratic programming problem, the complexity of algorithm is O ((n+
m)3), wherein n is former training sample number, and m is newly-increased sample.Using the application increment
Learning algorithm, calculating newly-increased sample m and violating the training sample m time complexity of KKT condition is O (m × l × k), l
For former supporting vector number, k is training set dimension.Known by computational geometry theory, shell to
Measure the number that number n is much smaller than former training sample.Then the time complexity of the application is O (m × l × k)+O ((m'
+n')3).Since m', n', l, k are much smaller than n, m is less than (n+m), therefore the application algorithm can effectively improve training effectiveness.
Step 104: calling updated single class anomaly detector model, online abnormal inspection is carried out to current transformer vibration
It surveys.
That is, updated list class anomaly detector model obtained the new centre of sphere and and radius:Then for needing data z to be tested, if meeting formulaThen think that test data z is fallen into hypersphere, receives
Sample is target class, is otherwise foreign peoples.
In summary, a kind of transformer described herein vibrates online method for detecting abnormality, and this method is to obtain to preset
Newly-increased data sample in period about running state of transformer vibration signal extracts newly-increased data sample based on wavelet packet analysis
This characteristic parameter, the study of newly-increased data sample is completed based on quick algorithm of convex hull, and training updates single class anomaly detector mould
Type calls updated anomaly detector model, carries out online abnormality detection to current transformer vibration.Since the application utilizes
Incremental Learning Algorithm only learns the newly-increased data sample in preset time period, so can timely and effectively handle newly-increased
Data can online in real time learn newly-increased training sample, realize the quick upgrading of detection model, while can reduce
Demand of the model modification to time and space.
Various pieces are described in a progressive manner in this specification, and what each some importance illustrated is and other parts
Difference, same and similar part may refer to each other between various pieces.
The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present invention.
Various modifications to these embodiments will be readily apparent to those skilled in the art, defined herein
General Principle can realize in other embodiments without departing from the spirit or scope of the present invention.Therefore, this hair
It is bright to be not intended to be limited to embodiment shown in the application, and be to fit to and principle disclosed in the present application and features of novelty phase
Consistent widest scope.
Claims (1)
1. a kind of transformer vibrates online method for detecting abnormality characterized by comprising
Obtain the newly-increased data sample in preset time period about running state of transformer vibration signal；
The characteristic parameter of newly-increased data sample is extracted based on wavelet packet analysis；
The study of newly-increased data sample is completed based on quick algorithm of convex hull, training updates single class anomaly detector model；
Updated single class anomaly detector model is called, online abnormality detection is carried out to current transformer vibration.
Wherein, the study of newly-increased data sample is completed based on quick algorithm of convex hull, training updates single class anomaly detector model, tool
Body includes:
If initial training sample A0, increasing sample newly is B={ B1,B2,…,Bn, andNew training sample
Ai(i ∈ n),
1) initial training sample A0Training obtains Support Vector data description model Ω0, supporting vector SV0；
2) newly-increased sample B is addedi(i ∈ n), finds out BiThe middle sample for violating KKT condition, is denoted asIfThen return
Ωi-1；Otherwise sample A is calculatedi-1Except supporting vector SVi-1Sample A afterwardsi-1, i.e. A 'i-1=Ai-1-SVi-1, and utilize quick convex closure
Algorithm calculates A 'i-1Shell vector Ci-1；
3) willNew Support Vector data description model Ω is obtained as the training of new training samplei, i=i
+1；As i > n, algorithm is terminated, when 2) i≤n is gone to step.
Wherein, KKT condition is expressed as follows:
Wherein, αi(i ∈ n) is Lagrange multiplier, and v is used to balance volume of hypersphere and training error, referred to as balance parameters, and R is instruction
The radius for practicing the minimal hyper-sphere of sample is the centre of sphere, and a is that the radius of the minimal hyper-sphere of training sample is the centre of sphere, and z is test specimens
This, d2=| | z-a | |2。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810598845.3A CN109029699B (en) | 2018-06-12 | 2018-06-12 | Transformer vibration online anomaly detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810598845.3A CN109029699B (en) | 2018-06-12 | 2018-06-12 | Transformer vibration online anomaly detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109029699A true CN109029699A (en) | 2018-12-18 |
| CN109029699B CN109029699B (en) | 2020-09-11 |
Family
ID=64612631
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810598845.3A Active CN109029699B (en) | 2018-06-12 | 2018-06-12 | Transformer vibration online anomaly detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109029699B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110749793A (en) * | 2019-10-31 | 2020-02-04 | 杭州中恒云能源互联网技术有限公司 | Dry-type transformer health management method and system based on width learning and storage medium |
| CN112257737A (en) * | 2020-06-29 | 2021-01-22 | 北京沃东天骏信息技术有限公司 | Abnormal commodity object identification method, equipment and storage medium |
| CN113670428A (en) * | 2021-07-19 | 2021-11-19 | 国网四川省电力公司电力科学研究院 | Transformer vibration online anomaly detection method |
| CN115905974A (en) * | 2022-11-04 | 2023-04-04 | 北京科技大学 | Method for detecting abnormal furnace condition of blast furnace |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103294716A (en) * | 2012-02-29 | 2013-09-11 | 佳能株式会社 | On-line semi-supervised learning method and device for classifier, and processing equipment |
| CN103645249A (en) * | 2013-11-27 | 2014-03-19 | 国网黑龙江省电力有限公司 | Online fault detection method for reduced set-based downsampling unbalance SVM (Support Vector Machine) transformer |
| CN107144430A (en) * | 2017-06-27 | 2017-09-08 | 电子科技大学 | A kind of Method for Bearing Fault Diagnosis based on incremental learning |
| WO2018010028A1 (en) * | 2016-07-14 | 2018-01-18 | HYDRO-QUéBEC | Detection of anomalies in an electrical network |
-
2018
- 2018-06-12 CN CN201810598845.3A patent/CN109029699B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103294716A (en) * | 2012-02-29 | 2013-09-11 | 佳能株式会社 | On-line semi-supervised learning method and device for classifier, and processing equipment |
| CN103645249A (en) * | 2013-11-27 | 2014-03-19 | 国网黑龙江省电力有限公司 | Online fault detection method for reduced set-based downsampling unbalance SVM (Support Vector Machine) transformer |
| WO2018010028A1 (en) * | 2016-07-14 | 2018-01-18 | HYDRO-QUéBEC | Detection of anomalies in an electrical network |
| CN107144430A (en) * | 2017-06-27 | 2017-09-08 | 电子科技大学 | A kind of Method for Bearing Fault Diagnosis based on incremental learning |
Non-Patent Citations (2)
| Title |
|---|
| 刘福荣等: "基于支持向量数据描述的变压器故障检测方法", 《黑龙江电力》 * |
| 陈利等: "支持向量数据描述法在变压器故障预警中的应用", 《西南民族大学学报(自然科学版)》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110749793A (en) * | 2019-10-31 | 2020-02-04 | 杭州中恒云能源互联网技术有限公司 | Dry-type transformer health management method and system based on width learning and storage medium |
| CN112257737A (en) * | 2020-06-29 | 2021-01-22 | 北京沃东天骏信息技术有限公司 | Abnormal commodity object identification method, equipment and storage medium |
| CN113670428A (en) * | 2021-07-19 | 2021-11-19 | 国网四川省电力公司电力科学研究院 | Transformer vibration online anomaly detection method |
| CN113670428B (en) * | 2021-07-19 | 2023-12-22 | 国网四川省电力公司电力科学研究院 | Transformer vibration online abnormality detection method |
| CN115905974A (en) * | 2022-11-04 | 2023-04-04 | 北京科技大学 | Method for detecting abnormal furnace condition of blast furnace |
| CN115905974B (en) * | 2022-11-04 | 2024-02-27 | 北京科技大学 | Method for detecting abnormal furnace condition of blast furnace |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109029699B (en) | 2020-09-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109029699A (en) | A kind of online method for detecting abnormality of transformer vibration | |
| CN109074037A (en) | For in quick predict pipeline, pressure vessels and pipes system hydrogen induced cracking (HIC) (HIC) and the system and method for taking relative action | |
| CN106897717A (en) | Bayesian model modification method under multiple test based on environmental excitation data | |
| CN105956709B (en) | A kind of modularization support vector machines tide prediction method based on GUI | |
| WO2019235614A1 (en) | Relationship analysis device, relationship analysis method, and recording medium | |
| CN106240839A (en) | For analyzing the system and method for flutter test data | |
| CN110211119A (en) | Image quality measure method, apparatus, electronic equipment and readable storage medium storing program for executing | |
| JP2019021186A (en) | Data processor, control system, data processing method, and program | |
| Crawford et al. | A hyperheuristic approach for dynamic enumeration strategy selection in constraint satisfaction | |
| CN113901977A (en) | Deep learning-based power consumer electricity stealing identification method and system | |
| CN110682159A (en) | Cutter wear state identification method and device | |
| US20180052441A1 (en) | Simulation system, simulation method, and simulation program | |
| CN110308658A (en) | A kind of pid parameter setting method, device, system and readable storage medium storing program for executing | |
| CN108984896A (en) | The airtight sampler turnover panel sealing valve optimum design method of the complete deep deposit in sea | |
| JP2022510667A (en) | Methods and equipment for predicting energy consumption | |
| CN109444957B (en) | Hydrate reservoir earthquake prediction method and processing terminal based on decision tree evaluation | |
| Colby et al. | Counterfactual Exploration for Improving Multiagent Learning. | |
| JPWO2019235603A1 (en) | Relationship analyzers, relationship analysis methods and programs | |
| CN109960146A (en) | The method for improving soft measuring instrument model prediction accuracy | |
| JP2017100159A (en) | State estimation method, molten metal surface level control method, program, and state estimation device | |
| CN109961160A (en) | A kind of power grid future operation trend predictor method and system based on trend parameter | |
| CN112597687A (en) | Turbine disk structure mixed reliability analysis method based on few-sample learning | |
| JP7058386B2 (en) | Analytical equipment, analytical methods and programs | |
| CN111066562A (en) | Grape downy mildew prediction method and system | |
| Zonzini et al. | Enhancing vibration-based structural health monitoring via edge computing: A tiny machine learning perspective |
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 |