CN105182219A - Power converter fault classification method based on Hamming error correcting code support vector machine - Google Patents
Power converter fault classification method based on Hamming error correcting code support vector machine Download PDFInfo
- Publication number
- CN105182219A CN105182219A CN201510559236.3A CN201510559236A CN105182219A CN 105182219 A CN105182219 A CN 105182219A CN 201510559236 A CN201510559236 A CN 201510559236A CN 105182219 A CN105182219 A CN 105182219A
- Authority
- CN
- China
- Prior art keywords
- fault
- sub
- classifier
- power converter
- error correction
- 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
Landscapes
- Error Detection And Correction (AREA)
Abstract
The invention discloses a power converter fault classification method based on a Hamming error correcting code support vector machine. First of all, possible fault modes of a power converter are analyzed, testability analyzing is carried out, and measurable nodes are selected; then, output signals in various fault modes are acquired at the test nodes, and a certain quantity of fault samples are obtained through feature extraction; and finally, based on coding rules of Hamming codes, various faults are coded, after labels of the fault samples are determined according to code words, each sub classifier is trained, a model based on the Hamming error correcting code support vector machine is constructed, and in a test phase, after error detection and error correction are carried out on an output result of each sub classifier, a final classification result is obtained. According to the invention, when multiple faults existing in the power converter are classified, fewer sub classifiers are needed, the classification speed is fast, errors of the final classification result can be detected and corrected, and the fault diagnosis reliability of the power converter can be improved.
Description
Technical field
The present invention relates to a kind of Power Converter faults sorting technique based on Hamming error correction codes support vector machine, belong to Fault Diagnosis of Power Electronic Circuits field.
Background technology
Power Electronic Technique is 21 century one of most widely used technology, and along with the effect of Power Electronic Technique in national economy constantly strengthens, the development of Power Electronic Technique is also very fast.Power converter is as a kind of power electronic equipment that certain electric current can be converted to other types electric current, be widely used in the fields such as industry, military affairs and Aero-Space, be a part indispensable in electric system, the fault correctly diagnosing out power converter to produce has great importance to the safety and reliability improving electric system.
Containing device for power switching in power converter circuit, there is the feature of strong nonlinearity, be difficult to set up accurate mathematical model, therefore by the method for modeling, fault diagnosis is carried out to power converter and there is certain difficulty.At present, the method based on pattern-recognition does not rely on the mathematical model of object, carries out fault diagnosis have larger advantage to power converter, thus receives increasing concern and research.Support vector machine (supportvectormachine, SVM) be a kind of machine learning method that the Corpus--based Method theories of learning propose, to the pattern-recognition of Small Sample Database, there is outstanding study generalization ability, become the emphasis of the another research after various neural network, but support vector machine can only solve two classification problems, in order to use support vector machine method, the multiclass fault occurred in power converter is classified, multiple support vector machine classifier is needed to combine, conventional multi-class support vector machine method has one to one, one-to-many and directed acyclic graph.Consider the n kind diagnosing malfunction classification to existing in power converter, One-against-one needs the number of support vector machine classifier to be n, and the number of the support vector machine classifier of support vector machine and directed acyclic graph support vector machine needs is n (n+1)/1 one to one.Power converter is generally operational in the occasion of operating mode more complicated, due to the interference of noise, the data of each collection can also exist difference, erroneous judgement is produced sometimes when using support vector machine method to carry out failure modes, and conventional multi-class classification support vector machine method does not have the ability of EDC error detection and correction, and when the number of defects is larger, the sub-classifier number of needs is a lot, calculated amount is large, is unfavorable for the on-line fault diagnosis of power converter.
The present invention is in conjunction with power converter strong nonlinearity, feature that fault category is many, the Fault Classification based on Hamming error correction codes support vector machine is adopted to carry out fault diagnosis to power converter, without the need to setting up power converter model, use the least possible sub-classifier number, classification speed is fast; Use the error correction principles based on Hamming code coding to carry out EDC error detection and correction to the result exported simultaneously, avoid the shortcoming using very few sub-classifier number to bring model generalization indifferent, improve the reliability of power converter being carried out to failure modes.
Summary of the invention
The object of the present invention is to provide a kind of Power Converter faults sorting technique based on Hamming error correction codes support vector machine, solve traditional multi-class classification support vector machine method when classification error appears in sub-classifier on the one hand, not there is the ability of EDC error detection and correction, the problem judged by accident may be there is, the least possible sub-classifier number is used to solve the more problem of Power Converter faults classification on the other hand, owing to having carried out EDC error detection and correction to Output rusults, avoided the shortcoming using very few sub-classifier number to bring model generalization indifferent simultaneously.
The present invention for achieving the above object, adopts following technical scheme:
The present invention is a kind of Power Converter faults sorting technique based on Hamming error correction codes support vector machine, comprises the steps:
1) analyze fault mode common in power converter, and carry out Testability Analysis, after determining test node, gather the output signal under various fault mode, after feature extraction and pre-service, obtain fault training sample.
2) for the n class failure modes problem in power converter, according to binary coding principle, minimum needs
(symbol
represent and round up) individual sub-classifier, i-th (i=0,1,2 ..., n-1) the corresponding fault label of class fault is i, being converted into binary expression form is C
i, then the corresponding binary code word C of each fault
i, figure place is k.
3) code word C corresponding to each fault is obtained
iafter, according to the simplest parity check method, after each fault code word, increase a parity bit obtain new malfunction coding C
i', figure place is k+1; The code word C of parity bit will be added
i' as raw information code, according to the coding rule of Hamming code to C
i' encode, (r gets and meets formula 2 to need r check digit
rthe smallest positive integral value of>=r+k+2), obtain the malfunction coding C based on Hamming code coding that each fault is corresponding
i*, figure place is k+r+1.
4) fault sample is divided into two classes: training sample and test sample book.In the training stage, according to the malfunction coding C obtained
i* to each sub-classifier SVM
i(i=1,2 ..., k+r+1) corresponding fault category divides, then according to ready-portioned positive class and negative class training sample to SVM
itrain, obtain the support vector machine classifier model based on Hamming error correction codes trained.
5) at test phase, sample to be tested is first inputted k+1 sub-classifier SVM before in the model trained
j(j=1,2 ..., k+1) in, the result obtained is carried out parity checking, if verification is correct, then think that front k sub-classifier classification is correct, r remaining sub-classifier need not participate in test again, differentiates obtain final fault generic according to the result of calculation of a front k sub-classifier.If there is mistake in inspection, r then remaining sub-classifier participates in test, obtain the classification results of all k+r+1 sub-classifier, determine which sub-classifier is mistake occur on according to the error correction principles of Hamming code, if mistake occurs in kth+1 sub-classifier, then the direct classification results according to a front k sub-classifier carries out differentiation and obtains fault generic; If mistake occurs in a front k sub-classifier, then error correction is carried out to the classification results of this sub-classifier, differentiate last affiliated fault category according to the classification results of k sub-classifier before error correction.
The present invention is based on the Code And Decode method of Hamming code, Power Converter faults classification is encoded and EDC error detection and correction is carried out to Output rusults, the generalization ability of the support vector machine classifier model of structure is strengthened.In the training stage, the total number of subclassification needing training is k+r+1, when Power Converter faults classification number n is larger, sub-classifier number is less than the sub-classifier number of conventional multi-class classification method needs, such as, have 20 kinds of faults, conventional one-against-rest needs 20 sub-classifiers, need 190 sub-classifiers with directed acyclic drawing method one to one, and the method proposed needs 10 sub-classifiers.At test phase, in order to improve the test speed of algorithm, employ the simplest parity check method and error detection is carried out to result, if Output rusults is correct, then do not need to use the sub-classifier based on the check bit of Hamming code increase is corresponding to carry out computing error correction, the sub-classifier number participating in test is like this k+1; If there is mistake, need to carry out error correction according to Hamming code rule, the sub-classifier number participating in test is k+r+1.Like this by this two step of EDC error detection and correction separately, avoid when diagnostic result is correct, also need to call extra sub-classifier and go to calculate, improve the speed of failure modes.
Accompanying drawing explanation
Fig. 1: fault diagnosis block diagram of the present invention;
Fig. 2: method for diagnosing faults process flow diagram of the present invention.
Embodiment
Below in conjunction with accompanying drawing, technical scheme of the present invention is described in detail.
As shown in Figure 1, fault diagnosis principle of the present invention is: analyze fault mode common in power converter, and carry out Testability Analysis, after determining test node, gather the output signal under various fault mode, fault training sample is obtained after feature extraction and pre-service, then, according to Hamming code coding method, all kinds of fault is encoded, obtain the code word that every class fault is corresponding, after the label of training sample is determined according to these code words, train each sub-classifier (sub-classifier select be support vector machine), after training terminates, then construct the model based on Hamming error correction codes support vector machine, finally, use the same method the test sample book processing and collect, by each sub-classifier in test sample book input model, after the result obtained is carried out EDC error detection and correction, export last failure modes result.
As shown in Figure 2, a kind of Power Converter faults sorting technique based on Hamming error correction codes support vector machine of the present invention, concrete enforcement comprises the steps:
1) power converter breaks down and is mainly caused by power tube, with open a way and short trouble the most common.The fault mode of power converter is analyzed, after determining fault type and the number of defects, select test node and measured signal, choosing of test node will take into full account actual operability, be convenient to carry out data acquisition, the measured signal selected will comprise the information under various fault mode, and what generally select is output voltage or electric current.
2), when gathering fault-signal, if collection is current signal, then needs that current signal is converted into voltage signal and gather again, sample frequency f
ssize decide according to fault-signal, sampling number is N, and after collecting fault-signal, need to carry out feature extraction, conventional fault signature extracting method has Fourier transform, wavelet transformation etc., and the feature according to fault-signal decides.Such as, fault-signal is steady periodic signal, then can carry out feature extraction by Fourier transform, after Fourier transform, and spectral resolution size delta f=f
s/ N, m (m=1,2, ..., N) corresponding frequency size f=(m-1) Δ f, under selecting various fault, the amplitude that the Frequency point that otherness is larger is corresponding and phase angle are as eigenwert, these eigenwerts are formed a proper vector by the amplitude of general selection each harmonic and phase angle.After extracting the fault feature vector of all kinds of fault, then to carry out average be 0, and variance is the normalized of 1, obtains last fault sample; Change initial conditions or load, many group fault samples can be obtained like this for training and testing afterwards.
3), after fault sample gathers, suppose in power converter, there is n class fault, use the method based on support vector machine to classify to this n class fault, according to binary coding principle, minimum needs
individual sub-classifier, i-th (i=0,1,2 ..., n-1) the corresponding fault label of class fault is i, being converted into binary expression form is C
i, binary figure place is k, then the corresponding binary code word C of every class fault
i.
4) code word C corresponding to every class fault is obtained
iafter, according to the simplest parity check method, after each fault code word, increase a parity bit obtain new malfunction coding C
i', figure place is k+1; The code word C of parity bit will be added
i' as raw information code, according to the coding rule of Hamming code to C
i' encode, need r check digit, the formula of r demand fulfillment is as follows:
2
r≥r+k+2(1)
Wherein, r is the minimum positive integer meeting above-mentioned formula, and the malfunction coding based on Hamming code coding obtaining every class fault so corresponding is C
i*, figure place is k+r+1.Such as, power converter has 8 kinds of faults, fault category be respectively i (F0,1 ..., 7), 8 kinds of faults need k=3 sub-classifier, i.e. C
inumber of bits be 3, C
i' number of bits be 4, according to the check bit r=3 that formula (1) known Hamming code coding needs, then the malfunction coding C that every class fault is corresponding
i* number of bits is 7, each binary digit b
l(l=0,1 ..., 6) represent, then the malfunction coding C of the every class fault obtained of encoding based on Hamming code
i*=b
6b
5b
4b
3b
2b
1b
0.Carry out, in malfunction coding process, using S with Hamming code coding
1, S
2, S
3represent the syndrome in 3 supervision relational expressions, then S
1, S
2and S
3value is as shown in table 1 with the corresponding relation of error code position.Also can specify another kind of corresponding relation, this discusses the training of sub-classifier and Output rusults is carried out to the generality of EDC error detection and correction after not affecting.
Table 1
| S 1S 2S 3 | Error code position | S 1S 2S 3 | Error code position |
| 001 | b 0 | 101 | b 4 |
| 010 | b 1 | 110 | b 5 |
| 100 | b 2 | 111 | b 6 |
| 011 | b 3 | 000 | Without error code |
Three following even number supervision relations can be obtained according to table 1:
According to three supervision relations in formula (2), make S
1, S
2and S
3be 0, then can obtain b
0, b
1and b
2, formula is as follows:
Code word C
i=b
6b
5b
4the binary representation form of Power Converter faults classification i, b
3to code word C
ieven parity check, like this according to the fault category that power converter is known, just can obtain b
6, b
5, b
4and b
3, then can obtain binary digit b according to formula (3)
2, b
1and b
0, then press Hamming code rule encoding to 8 kinds of faults of power converter, the malfunction coding obtained is as shown in table 2.
Table 2
As shown in table 2, the malfunction coding C of the i-th class fault
i* each the binary digit b in
irepresent by the sub-classifier SVM of correspondence
j(j=0,1 ..., 6) classification that divides, be 1 and be divided into positive class, be 0 and be divided into negative class.Such as, SVM
0the 6th b that corresponding is in malfunction coding
6, then fault 0,1,2,3 is divided into negative class, and fault 4,5,6,7 is divided into positive class, and other sub-classifier division rule is identical.
5) fault sample is divided into two classes: training sample and test sample book.In the training stage, training sample is according to malfunction coding C
i* label is determined, then to each sub-classifier SVM
i(i=1,2 ..., k+r+1) corresponding fault category divides, according to ready-portioned positive class and negative class training sample to this sub-classifier SVM
itraining, after training terminates, by training the model parameter obtained to store, then constructing the supporting vector machine model based on Hamming error correction codes.
6) at test phase, k+1 sub-classifier SVM before in the model first test sample book input trained
j(j=0,1 ..., k) in, the result obtained is carried out parity checking, if verification is correct, then think that front k sub-classifier classification is correct, r remaining sub-classifier need not participate in test again, differentiates obtain final fault generic according to the result of calculation of a front k sub-classifier.If there is mistake in inspection, then need to carry out error correction, test sample book is continued in r remaining sub-classifier of input, obtain the classification results of all k+r+1 sub-classifier like this, first determine which sub-classifier is mistake occur on according to Hamming code error correction principles, if mistake occurs on kth+1 sub-classifier, expression is sub-classifier classification error corresponding to even parity bit, the classification results of a front k sub-classifier is correct, then the direct classification results according to a front k sub-classifier differentiates, obtains fault generic; If mistake occurs in a front k sub-classifier, then carry out error correction to the classification results of this sub-classifier, before good according to error correction, the Output rusults of k sub-classifier differentiates last affiliated fault category.Be described for 8 kinds of failure modess in power converter equally, if the test sample book of input is fault 2, in correct classification situation, front 4 sub-classifier Output rusults are 0101, represent that the 1st subclassification exports 0, second sub-classifier and export 1,3rd sub-classifier exports 0,4th sub-classifier exports 1, and according to even parity check, Output rusults is correct, then the result of front 3 subclassifications is 010, can discriminating fault types be fault 2; If the 1st sub-classifier classification error, Output rusults is 1, Output rusults 110 according to front 3 subclassifications directly carries out failure modes, fault 6 can be mistaken for, even parity check be carried out to the Output rusults 1101 of front 4 sub-classifiers, can find not meet rule, need to carry out error correction, by sub-classifier participation test corresponding for 3 check bit below, the result of output is 101, obtains S according to the supervision relation in formula (2)
1s
2s
3be 111, known b
6there is mistake in position, namely first sub-classifier classification error, carries out error correction, changes Output rusults 1 into 0, then the Output rusults of front 3 sub-classifiers after error correction is 010, and failure modes result is fault 2.
Claims (3)
1., based on a Power Converter faults sorting technique for Hamming error correction codes support vector machine, comprise the steps:
(1) analyze fault mode common in power converter, and carry out Testability Analysis, determine test node;
(2) gather the output signal of power converter under various fault mode at test node place, obtained the fault sample under various fault mode by feature extraction (Fourier analysis and wavelet transformation etc.) and pre-service (normalization);
(3) in the model training stage, encode to every class fault according to Hamming code coding principle, the corresponding code word of every class fault, after the label of fault sample is determined according to these code words, each sub-classifier is trained, obtains the model based on Hamming error correction codes support vector machine;
(4) in the class test stage, by test sample book input model, each sub-classifier all has an output, detect Output rusults whether to make mistakes, if there is mistake, then carry out error correction according to the subclassification result of error correction principles to mistake of Hamming code, then export last classification results.
2., according to a kind of Power Converter faults sorting technique based on Hamming error correction codes support vector machine described in claim 1, it is characterized in that the coding method of every class fault in described step (3) is as follows:
(1) fault mode to be sorted has n, i-th (i=0,1,2 ..., n-1) the corresponding fault category of class fault is i, being converted into binary expression form is C
i, wherein number of bits is minimum value
(2) according to parity check method, at each fault code word C
irear increase parity bit, obtains new malfunction coding C
i', figure place is k+1; The code word C of parity bit will be added
i' as raw information code, then according to the coding rule of Hamming code to C
i' encode, (r gets and meets formula 2 to need r check digit
rthe minimum positive integer value of>=r+k+2), obtain the malfunction coding C that final every class fault is corresponding
i *, figure place is k+r+1.
3., according to a kind of Power Converter faults sorting technique based on Hamming error correction codes support vector machine described in claim 1, it is characterized in that the test phase in described step (4), to classification results carry out error detection, error correction method as follows:
(1) first by k+1 sub-classifier before test sample book input, obtain the output of a front k+1 sub-classifier, result is carried out odd-even check, if correct, then directly can obtain classification results according to the output of a front k sub-classifier, otherwise, need the error correction carrying out next step;
(2) test sample book is continued in r last sub-classifier of input, obtain the output of all sub-classifiers like this, according to Hamming code error correction principles, error correction is carried out to a front k+1 sub-classifier, if mistake occurs in kth+1 sub-classifier, then the direct output according to a front k sub-classifier obtains classification results, if occur in a front k sub-classifier, first error correction is carried out to the Output rusults of this sub-classifier, and then obtain last result according to the output of a front k sub-classifier.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510559236.3A CN105182219A (en) | 2015-09-06 | 2015-09-06 | Power converter fault classification method based on Hamming error correcting code support vector machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510559236.3A CN105182219A (en) | 2015-09-06 | 2015-09-06 | Power converter fault classification method based on Hamming error correcting code support vector machine |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105182219A true CN105182219A (en) | 2015-12-23 |
Family
ID=54904430
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510559236.3A Pending CN105182219A (en) | 2015-09-06 | 2015-09-06 | Power converter fault classification method based on Hamming error correcting code support vector machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105182219A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107967688A (en) * | 2017-12-21 | 2018-04-27 | 联想(北京)有限公司 | The method and system split to the object in image |
| CN108254688A (en) * | 2018-04-02 | 2018-07-06 | 中国矿业大学 | Power converter of switch reluctance motor small echo no-load voltage ratio method for diagnosing faults |
| CN111339443A (en) * | 2020-03-09 | 2020-06-26 | 腾讯科技（深圳）有限公司 | User label determination method and device, computer equipment and storage medium |
| CN111382809A (en) * | 2020-06-01 | 2020-07-07 | 国网江西省电力有限公司电力科学研究院 | Isolating switch mechanical fault diagnosis method based on motor output power |
| CN111950692A (en) * | 2019-05-16 | 2020-11-17 | 国际商业机器公司 | Robust output coding based on hamming distance for improved generalization |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110004810A1 (en) * | 2009-07-06 | 2011-01-06 | Himax Media Solutions, Inc. | Method and System of Receiving Data with Enhanced Error Correction |
| CN103941143A (en) * | 2014-05-07 | 2014-07-23 | 中国矿业大学 | Switched reluctance motor power converter main switch short-circuit fault diagnosis method |
| CN103995528A (en) * | 2014-05-09 | 2014-08-20 | 南京航空航天大学 | Intelligent self-repairing technology for main circuit of power converter |
-
2015
- 2015-09-06 CN CN201510559236.3A patent/CN105182219A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110004810A1 (en) * | 2009-07-06 | 2011-01-06 | Himax Media Solutions, Inc. | Method and System of Receiving Data with Enhanced Error Correction |
| CN103941143A (en) * | 2014-05-07 | 2014-07-23 | 中国矿业大学 | Switched reluctance motor power converter main switch short-circuit fault diagnosis method |
| CN103995528A (en) * | 2014-05-09 | 2014-08-20 | 南京航空航天大学 | Intelligent self-repairing technology for main circuit of power converter |
Non-Patent Citations (3)
| Title |
|---|
| NIMA HATAMI 等: "ECOC-based training of neural networks for face recognition", 《CONFERENCE ON CYBERNETICS AND INTELLIGENT SYSTEMS》 * |
| 孙永奎 等: "基于可测性分析和支持向量机的模拟电路故障诊断", 《仪器仪表学报》 * |
| 胡莹岑: "基于支持向量机的模拟电路故障诊断技术研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107967688A (en) * | 2017-12-21 | 2018-04-27 | 联想(北京)有限公司 | The method and system split to the object in image |
| CN108254688A (en) * | 2018-04-02 | 2018-07-06 | 中国矿业大学 | Power converter of switch reluctance motor small echo no-load voltage ratio method for diagnosing faults |
| CN108254688B (en) * | 2018-04-02 | 2020-05-08 | 中国矿业大学 | Wavelet transformation ratio fault diagnosis method for power converter of switched reluctance motor |
| CN111950692A (en) * | 2019-05-16 | 2020-11-17 | 国际商业机器公司 | Robust output coding based on hamming distance for improved generalization |
| CN111950692B (en) * | 2019-05-16 | 2024-05-28 | 国际商业机器公司 | Robust output coding based on hamming distance for improved generalization |
| CN111339443A (en) * | 2020-03-09 | 2020-06-26 | 腾讯科技（深圳）有限公司 | User label determination method and device, computer equipment and storage medium |
| CN111339443B (en) * | 2020-03-09 | 2023-04-07 | 腾讯科技（深圳）有限公司 | User label determination method and device, computer equipment and storage medium |
| CN111382809A (en) * | 2020-06-01 | 2020-07-07 | 国网江西省电力有限公司电力科学研究院 | Isolating switch mechanical fault diagnosis method based on motor output power |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105182219A (en) | Power converter fault classification method based on Hamming error correcting code support vector machine | |
| CN102435910B (en) | Power electronic circuit health monitoring method based on support vector classification | |
| CN109489977B (en) | KNN-AdaBoost-based bearing fault diagnosis method | |
| CN113702895B (en) | Online quantitative evaluation method for error state of voltage transformer | |
| CN109389325B (en) | Method for evaluating state of electronic transformer of transformer substation based on wavelet neural network | |
| CN104460654B (en) | A kind of gear imperfection failure diagnosis Rules extraction method based on quantization characteristic relation | |
| CN101871994B (en) | Method for diagnosing faults of analog circuit of multi-fractional order information fusion | |
| CN114090396A (en) | Cloud environment multi-index unsupervised anomaly detection and root cause analysis method | |
| CN113111591B (en) | Automatic diagnosis method, device and equipment based on internal fault of modular power distribution terminal | |
| CN116304898A (en) | Sensor data intelligent storage system based on machine learning | |
| Zhang et al. | Diagnostic system based on support-vector machines for board-level functional diagnosis | |
| CN112416662A (en) | Multi-time series data anomaly detection method and device | |
| CN109726770A (en) | A kind of analog circuit fault testing and diagnosing method | |
| CN116295948A (en) | Abnormality detection method, system and storage medium of industrial temperature sensor in large temperature difference environment | |
| CN115712871A (en) | Power electronic system fault diagnosis method combining resampling and integrated learning | |
| CN115184054A (en) | Mechanical equipment semi-supervised fault detection and analysis method, device, terminal and medium | |
| CN110717602A (en) | Machine learning model robustness assessment method based on noise data | |
| CN114330486A (en) | Power system bad data identification method based on improved Wasserstein GAN | |
| CN117630800A (en) | Fault diagnosis method and system for automatic calibrating device of electric energy meter | |
| CN102697491A (en) | Identification method and system of characteristic waveform of electrocardiogram | |
| CN112949714A (en) | Fault possibility estimation method based on random forest | |
| Pan et al. | Unsupervised two-stage root-cause analysis for integrated systems | |
| Guoming et al. | Analog circuit fault diagnosis using wavelet feature optimization approach | |
| CN114841262A (en) | Rolling bearing fault diagnosis method based on DS evidence theory | |
| Zuiev et al. | Questions of radioelectronic equipment diagnostics programs efficiency analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20151223 |
|
| WD01 | Invention patent application deemed withdrawn after publication |