118:, wavelet transform, short term Fourier transform, Gabor Expansion, Wigner-Ville distribution (WVD), cepstrum, bispectrum, correlation method, high resolution spectral analysis, waveform analysis (in the time domain, because spectral analysis usually concerns only frequency distribution and not phase information) and others. The results of this analysis are used in a root cause failure analysis in order to determine the original cause of the fault. For example, if a bearing fault is diagnosed, then it is likely that the bearing was not itself damaged at installation, but rather as the consequence of another installation error (e.g., misalignment) which then led to bearing damage. Diagnosing the bearing's damaged state is not enough for precision maintenance purposes. The root cause needs to be identified and remedied. If this is not done, the replacement bearing will soon wear out for the same reason and the machine will suffer more damage, remaining dangerous. Of course, the cause may also be visible as a result of the spectral analysis undertaken at the data-collection stage, but this may not always be the case.
353:
125:), is not a constant, especially not during the start-up and shutdown stages of the machine. Even if the machine is running in the steady state, the rotational speed will vary around a steady-state mean value, and this variation depends on load and other factors. Since sound and vibration signals obtained from a rotating machine are strongly related to its rotational speed, it can be said that they are time-variant signals in nature. These time-variant features carry the machine fault signatures. Consequently, how these features are extracted and interpreted is important to research and industrial applications.
181:, then quadratic time frequency analysis would be the power spectrum counterpart. Quadratic algorithms include the Gabor spectrogram, Cohen's class and the adaptive spectrogram. The main advantage of time frequency analysis is discovering the patterns of frequency changes, which usually represent the nature of the signal. As long as this pattern is identified the machine fault associated with this pattern can be identified. Another important use of time frequency analysis is the ability to filter out a particular frequency component using a time-varying filter.
72:
switches between the different modes of operation (passive, active, standby, off, and isolated) of each actuator. For example, if a fault is detected in hydraulic system 1, then the truth table sends an event to the state chart that the left inner actuator should be turned off. One of the benefits of this model-based FDI technique is that this reactive controller can also be connected to a continuous-time model of the actuator hydraulics, allowing the study of switching transients.
56:
422:
166:
reduction and time-varying filtering. Although the quadratic method describes the energy distribution of a signal in the joint time frequency domain, which is useful for analysis, classification, and detection of signal features, phase information is lost in the quadratic time-frequency representation; also, the time histories cannot be reconstructed with this method.
71:
The example shown in the figure on the right illustrates a model-based FDI technique for an aircraft elevator reactive controller through the use of a truth table and a state chart. The truth table defines how the controller reacts to detected faults, and the state chart defines how the controller
88:
where a signal is sent down a cable or electrical line and the reflected signal is compared mathematically to original signal to identify faults. Spread
Spectrum Time Domain Reflectometry, for instance, involves sending down a spread spectrum signal down a wire line to detect wire faults. Several
34:
has occurred, and pinpointing the type of fault and its location. Two approaches can be distinguished: A direct pattern recognition of sensor readings that indicate a fault and an analysis of the discrepancy between the sensor readings and expected values, derived from some model. In the latter
67:
or knowledge based. Some of the model-based FDI techniques include observer-based approach, parity-space approach, and parameter identification based methods. There is another trend of model-based FDI schemes, which is called set-membership methods. These methods guarantee the detection of fault
367:
in fault detection and diagnosis. ANNs are well-known for their efficient self-learning capabilities of the complex relations (which generally exist inherently in fault detection and diagnosis problems) and are easy to operate. Another advantage of ANNs is that they perform automatic feature
165:
The time frequency approach for machine fault diagnosis can be divided into two broad categories: linear methods and the quadratic methods. The difference is that linear transforms can be inverted to construct the time signal, thus, they are more suitable for signal processing, such as noise
101:
concerned with finding faults arising in machines. A particularly well developed part of it applies specifically to rotating machinery, one of the most common types encountered. To identify the most probable faults leading to failure, many methods are used for data collection, including
192:
As a result, using fault diagnostics to meet industrial needs in a cost-effective way, and to reduce maintenance costs without requiring more investments than the cost of what is to be avoided in the first place, requires an effective scheme of applying them. This is the subject of
648:"Pieter J. Mosterman and Jason Ghidella, "Model Reuse for the Training of Fault Scenarios in Aerospace," in Proceedings of the AIAA Modeling and Simulation Technologies Conference, CD-ROM, paper 2004-4931, August 16 - 19, Rhode Island Convention Center, Providence, RI, 2004"
161:
of a machine is increasing or decreasing during its startup or shutdown period, its bandwidth in the FFT spectrum will become much wider than it would be simply for the steady state. Hence, in such a case, the harmonics are not so distinguishable in the spectrum.
466:
can be directly classified to normal and faulty classes. Such a technique avoids omitting any important fault message and results in a better performance of fault detection and diagnosis. In addition, by transforming signals to image constructions, 2D
152:
is only suitable for signals whose frequency contents do not change over time; however, as mentioned above, the frequency contents of the sound and vibration signals obtained from a rotating machine are very much time-dependent. For this reason,
582:
Jason R. Ghidella and Pieter J. Mosterman, "Requirements-Based
Testing in Aircraft Control Design," Paper ID AIAA 2005-5886 in AIAA Modeling and Simulations Technologies Conference and Exhibit 2005, August 15-18, San Francisco, California,
376:
and achieve higher performance. Moreover, properly determining the size of the hidden layer needs an exhaustive parameter tuning, to avoid poor approximation and generalization capabilities. In general, different SVMs and ANNs models (i.e.
865:
Tian, Jing; Morillo, Carlos; Azarian, Michael H.; Pecht, Michael (March 2016). "Motor
Bearing Fault Detection Using Spectral Kurtosis-Based Feature Extraction Coupled With K-Nearest Neighbor Distance Analysis".
80:
In signal processing based FDI, some mathematical or statistical operations are performed on the measurements, or some neural network is trained using measurements to extract the information about the fault.
723:
Ahmadimanesh, Alireza, and Seyyed
Mohammad Shahrtash. "Time–time-transform-based fault location algorithm for three-terminal transmission lines." IET Generation, Transmission & Distribution 7.5 (2013):
733:
Ahmadimanesh, A., and S. M. Shahrtash. "Employing S-transform for fault location in three terminal lines." Environment and
Electrical Engineering (EEEIC), 2011 10th International Conference on. IEEE, 2011.
279:
NN) is one of the oldest techniques which has been used to solve fault detection and diagnosis problems. Despite the simple logic that this instance-based algorithm has, there are some problems with large
1515:
Jia, Feng; Lei, Yaguo; Lin, Jing; Zhou, Xin; Lu, Na (May 2016). "Deep neural networks: A promising tool for fault characteristic mining and intelligent diagnosis of rotating machinery with massive data".
1093:
Saravanan, N.; Siddabattuni, V.N.S. Kumar; Ramachandran, K.I. (January 2010). "Fault diagnosis of spur bevel gear box using artificial neural network (ANN), and proximal support vector machine (PSVM)".
349:, a parameter tuning process is required to be conducted first. Therefore, the low speed of the training phase is a limitation of SVMs when it comes to its usage in fault detection and diagnosis cases.
372:
the training set, which will have consequences of having poor validation accuracy on the validation set. Hence, often, some regularization terms and prior knowledge are added to the ANN model to avoid
714:
Ahmadimanesh, Alireza, and S. Mohammad
Shahrtash. "Transient-based fault-location method for multiterminal lines employing S-transform." IEEE transactions on power delivery 28.3 (2013): 1373-1380.
136:-based spectrum of a time signal shows us the existence of its frequency contents. By studying these and their magnitude or phase relations, we can obtain various types of information, such as
1437:
Lei, Yaguo; Jia, Feng; Lin, Jing; Xing, Saibo; Ding, Steven X. (May 2016). "An
Intelligent Fault Diagnosis Method Using Unsupervised Feature Learning Towards Mechanical Big Data".
330:, SVMs have an impressive performance in generalization, even with small training data. However, general SVMs do not have automatic feature extraction themselves and just like
1154:
Qi, Guanqiu; Zhu, Zhiqin; Erqinhu, Ke; Chen, Yinong; Chai, Yi; Sun, Jian (January 2018). "Fault-diagnosis for reciprocating compressors using big data and machine learning".
749:
830:
Verdier, Ghislain; Ferreira, Ariane (February 2011). "Adaptive
Mahalanobis Distance and $ k$ -Nearest Neighbor Rule for Fault Detection in Semiconductor Manufacturing".
509:
Fault
Recovery in FDIR is the action taken after a failure has been detected and isolated to return the system to a stable state. Some examples of fault recoveries are:
132:, or Fourier transform. The Fourier transform and its inverse counterpart offer two perspectives to study a signal: via the time domain or via the frequency domain. The
121:
The most common technique for detecting faults is the time-frequency analysis technique. For a rotating machine, the rotational speed of the machine (often known as the
1248:
Wong, Pak Kin; Yang, Zhixin; Vong, Chi Man; Zhong, Jianhua (March 2014). "Real-time fault diagnosis for gas turbine generator systems using extreme learning machine".
901:
Safizadeh, M.S.; Latifi, S.K. (July 2014). "Using multi-sensor data fusion for vibration fault diagnosis of rolling element bearings by accelerometer and load cell".
68:
under certain conditions. The main difference is that instead of finding the most likely model, these techniques omit the models, which are not compatible with data.
368:
extraction by allocating negligible weights to the irrelevant features, helping the system to avoid dealing with another feature extractor. However, ANNs tend to
605:"Model Invalidation for Switched Affine Systems with Applications to Fault and Anomaly Detection**This work is supported in part by DARPA grant N66001-14-1-4045"
769:
Bahrampour, Soheil; Moshiri, Behzad; Salahshour, Karim. "Weighted and constrained possibilistic C-means clustering for online fault detection and isolation
177:
are two algorithms commonly used as linear time-frequency methods. If we consider linear time-frequency analysis to be the evolution of the conventional
145:
1472:
Shao, Haidong; Jiang, Hongkai; Zhang, Xun; Niu, Maogui (1 November 2015). "Rolling bearing fault diagnosis using an optimization deep belief network".
960:
Liu, Ruonan; Yang, Boyuan; Zio, Enrico; Chen, Xuefeng (August 2018). "Artificial intelligence for fault diagnosis of rotating machinery: A review".
39:
goes above a certain threshold. It is then the task of fault isolation to categorize the type of fault and its location in the machinery.
636:
Farshad
Harirchi and Necmiye Ozay, "Guaranteed Model-Based Fault Detection in Cyber-Physical Systems: A Model Invalidation Approach", arXiv
1402:
Hoang, Duy-Tang; Kang, Hee-Jun (2019). "Rolling element bearing fault diagnosis using convolutional neural network and vibration image".
928:
Liu, Jie; Zio, Enrico (December 2016). "Feature vector regression with efficient hyperparameters tuning and geometric interpretation".
746:
557:
1033:
Kotsiantis, S.B.; Zaharakis, I.D.; Pintelas, P.E. (2006). "Machine learning: a review of classification and combining techniques".
89:
clustering methods have also been proposed to identify the novel fault and segment a given signal into normal and faulty segments.
647:
341:
technique. Another drawback of SVMs is that their performance is highly sensitive to the initial parameters, particularly to the
63:
In model-based FDI techniques some model of the system is used to decide about the occurrence of fault. The system model may be
1319:
1077:
47:) techniques can be broadly classified into two categories. These include model-based FDI and signal processing based FDI.
672:
Liu, Jie (2012). "Shannon wavelet spectrum analysis on truncated vibration signals for machine incipient fault detection".
222:
1275:
Tian, Yang; Fu, Mengyu; Wu, Fang (March 2015). "Steel plates fault diagnosis on the basis of support vector machines".
799:"Fault detection, classification and location for transmission lines and distribution systems: a review on the methods"
194:
542:
260:
to accurately identify the redundancies, faults and anomalous samples. During the past decades, there are different
352:
157:-based spectra are unable to detect how the frequency contents develop over time. To be more specific, if the
331:
316:
289:
269:
1558:
478:
468:
460:
456:
308:
304:
170:
1302:
Lv, Feiya; Wen, Chenglin; Bao, Zejing; Liu, Meiqin (July 2016). "Fault diagnosis based on deep learning".
201:
111:
440:
364:
360:
261:
245:
1123:"A hybrid artificial neural network with Dempster-Shafer theory for automated bearing fault diagnosis"
189:
In practice, model uncertainties and measurement noise can complicate fault detection and isolation.
85:
517:
1553:
532:
315:(CCA) accompany it to reach a better performance. In many industrial cases, the effectiveness of
296:
385:) have shown successful performances in the fault detection and diagnosis in industries such as
326:(SVMs), which is widely used in this field. Thanks to their appropriate nonlinear mapping using
322:
has been compared with other methods, specially with more complex classification models such as
1345:"A Novel Fault Diagnosis Method for Rotating Machinery Based on a Convolutional Neural Network"
552:
323:
216:
211:
178:
158:
154:
149:
133:
129:
122:
98:
1189:
Santos, Pedro; Villa, Luisa; Reñones, Aníbal; Bustillo, Andres; Maudes, Jesús (9 March 2015).
1003:
Genton, Marc G. (2001). "Classes of Kernels for Machine Learning: A Statistics Perspective".
562:
547:
382:
312:
651:
581:
489:
architectures which have been successfully used in this field of research. In comparison to
1521:
1481:
1356:
1202:
969:
681:
501:, however, they need larger samples and longer processing time to achieve higher accuracy.
486:
8:
1493:
785:
474:
338:
300:
265:
249:
206:
27:
1525:
1485:
1360:
1206:
973:
693:
685:
1497:
1454:
1419:
1379:
1344:
1325:
1225:
1190:
1171:
1121:
Hui, Kar Hoou; Ooi, Ching Sheng; Lim, Meng Hee; Leong, Mohd Salman (15 November 2016).
1050:
985:
883:
847:
697:
394:
64:
59:
Example of model-based FDI logic for an actuator in an aircraft elevator control system
31:
1501:
1384:
1315:
1230:
1073:
989:
701:
1423:
851:
522:
Change of state of the complete system into a Safe Mode with limited functionalities
1529:
1489:
1458:
1446:
1411:
1374:
1364:
1329:
1307:
1284:
1257:
1220:
1210:
1175:
1163:
1134:
1103:
1054:
1042:
1012:
977:
937:
910:
875:
839:
810:
689:
616:
490:
444:
239:
115:
110:, oil particle analysis, etc. Then these data are processed utilizing methods like
887:
1415:
1288:
1261:
1167:
941:
914:
621:
604:
378:
174:
107:
1533:
981:
233:
1107:
537:
448:
327:
281:
1046:
1016:
770:
228:
35:
case, it is typical that a fault is said to be detected if the discrepancy or
1547:
1450:
1311:
1070:
Business intelligence : data mining and optimization for decision making
879:
843:
753:
742:
494:
471:
can be implemented to identify faulty signals from vibration image features.
436:
430:
416:
342:
1139:
1122:
443:
have been developed to cope with fault detection and diagnosis. Most of the
1388:
1234:
402:
253:
55:
815:
798:
482:
410:
406:
373:
369:
1215:
398:
1369:
463:
390:
141:
137:
103:
268:
models that have been developed and proposed in this research area.
635:
30:
which concerns itself with monitoring a system, identifying when a
295:
is not able to automatically extract the features to overcome the
1092:
593:
Ding, S.X., Model-based fault diagnosis techniques, Springer 2008
498:
386:
346:
285:
257:
745:; Smith, Paul; Lo, Chet. "Spread Spectrum Sensors for Critical
452:
356:
Time domain waveform (top) and CWTS (bottom) of a normal signal
421:
234:
Machine learning techniques for fault detection and diagnosis
1191:"An SVM-Based Solution for Fault Detection in Wind Turbines"
447:
models extract a few feature values from signals, causing a
1343:
Guo, Sheng; Yang, Tao; Gao, Wei; Zhang, Chen (4 May 2018).
1032:
229:
Fault detection and diagnosis using artificial intelligence
1188:
797:
Chen, Kunjin; Huang, Caowei; He, Jinliang (1 April 2016).
417:
Deep learning techniques for fault detection and diagnosis
864:
786:"Robust residual selection for fault detection", 2014.
497:
models are able to learn more complex structures from
425:
Typical architecture of a Convolutional Neural Network
128:
The most common method used in signal analysis is the
435:
With the research advances in ANNs and the advent of
148:, bearing fault frequency and so on. However, the
1471:
1247:
1072:(. ed.). Hoboken, N.J.: Wiley. p. 436.
363:(ANNs) are among the most mature and widely used
84:A good example of signal processing based FDI is
1545:
1153:
832:IEEE Transactions on Semiconductor Manufacturing
439:algorithms using deep and complex layers, novel
1342:
900:
829:
774:Applied Intelligence, Vol 35, pp. 269-284, 2011
1120:
959:
75:
1514:
1436:
284:and processing time when it is used on large
602:
1439:IEEE Transactions on Industrial Electronics
1301:
868:IEEE Transactions on Industrial Electronics
796:
92:
1401:
1378:
1368:
1224:
1214:
1138:
1067:
955:
953:
951:
814:
620:
603:Harirchi, Farshad; Ozay, Necmiye (2015).
558:Spread-spectrum time-domain reflectometry
516:Switch-over from a faulty equipment to a
184:
1518:Mechanical Systems and Signal Processing
1274:
1156:Simulation Modelling Practice and Theory
962:Mechanical Systems and Signal Processing
758:Structural Control and Health Monitoring
420:
351:
54:
20:Fault detection, isolation, and recovery
1546:
1304:2016 American Control Conference (ACC)
1028:
1026:
1002:
948:
927:
365:mathematical classification algorithms
97:Machine fault diagnosis is a field of
1005:Journal of Machine Learning Research
223:Integrated vehicle health management
197:; the different strategies include:
1023:
671:
13:
1474:Measurement Science and Technology
674:Measurement Science and Technology
493:, due to their deep architecture,
246:mathematical classification models
244:In fault detection and diagnosis,
195:maintenance, repair and operations
169:The short-term Fourier transform (
50:
14:
1570:
543:Failure mode and effects analysis
504:
513:Switch-off of a faulty equipment
379:Back-Propagation Neural Networks
1508:
1465:
1430:
1395:
1336:
1295:
1268:
1241:
1182:
1147:
1114:
1086:
1061:
996:
921:
894:
858:
823:
790:
779:
763:
736:
16:Subfield of control engineering
1494:10.1088/0957-0233/26/11/115002
1035:Artificial Intelligence Review
727:
717:
708:
665:
640:
629:
596:
587:
575:
313:Canonical correlation analysis
207:Planned preventive maintenance
1:
694:10.1088/0957-0233/23/5/055604
568:
479:Restricted Boltzmann machines
469:Convolutional neural networks
457:Convolutional neural networks
41:Fault detection and isolation
1416:10.1016/j.cogsys.2018.03.002
1289:10.1016/j.neucom.2014.09.036
1262:10.1016/j.neucom.2013.03.059
1168:10.1016/j.simpat.2017.10.005
942:10.1016/j.neucom.2016.08.093
915:10.1016/j.inffus.2013.10.002
622:10.1016/j.ifacol.2015.11.185
491:traditional machine learning
461:continuous wavelet transform
451:reduction from the original
309:Linear discriminant analysis
305:Principal component analysis
273:-nearest-neighbors algorithm
252:methods, are trained on the
7:
1534:10.1016/j.ymssp.2015.10.025
1127:Journal of Vibroengineering
982:10.1016/j.ymssp.2018.02.016
526:
337:, are often coupled with a
202:Condition-based maintenance
76:Signal processing based FDI
10:
1575:
1404:Cognitive Systems Research
1108:10.1016/j.asoc.2009.08.006
428:
361:Artificial Neural Networks
237:
219:(does not use diagnostics)
1068:Vercellis, Carlo (2008).
1047:10.1007/s10462-007-9052-3
1017:10.1162/15324430260185646
86:time domain reflectometry
1451:10.1109/TIE.2016.2519325
1312:10.1109/ACC.2016.7526751
880:10.1109/TIE.2015.2509913
844:10.1109/TSM.2010.2065531
248:which in fact belong to
1140:10.21595/jve.2016.17024
756:on Live Wire Networks"
533:Control reconfiguration
324:Support Vector Machines
297:curse of dimensionality
93:Machine fault diagnosis
1306:. pp. 6851–6856.
1096:Applied Soft Computing
553:Predictive maintenance
426:
383:Multi-Layer Perceptron
357:
217:Corrective maintenance
212:Preventive maintenance
185:Robust fault diagnosis
99:mechanical engineering
60:
816:10.1049/hve.2016.0005
563:System identification
548:Fault-tolerant system
441:classification models
424:
355:
58:
487:deep neural networks
475:Deep belief networks
345:, so in each signal
1559:Systems engineering
1526:2016MSSP...72..303J
1486:2015MeScT..26k5002S
1361:2018Senso..18.1429G
1207:2015Senso..15.5627S
974:2018MSSP..108...33L
686:2012MeScT..23e5604L
518:redundant equipment
395:mechanical bearings
339:data pre-processing
250:supervised learning
28:control engineering
26:) is a subfield of
1520:. 72–73: 303–315.
1216:10.3390/s150305627
903:Information Fusion
427:
358:
301:data preprocessing
61:
1370:10.3390/s18051429
1321:978-1-4673-8682-1
1079:978-0-470-51138-1
609:IFAC-PapersOnLine
112:spectral analysis
1566:
1538:
1537:
1512:
1506:
1505:
1469:
1463:
1462:
1445:(5): 3137–3147.
1434:
1428:
1427:
1399:
1393:
1392:
1382:
1372:
1340:
1334:
1333:
1299:
1293:
1292:
1272:
1266:
1265:
1245:
1239:
1238:
1228:
1218:
1201:(3): 5627–5648.
1186:
1180:
1179:
1151:
1145:
1144:
1142:
1133:(7): 4409–4418.
1118:
1112:
1111:
1090:
1084:
1083:
1065:
1059:
1058:
1030:
1021:
1020:
1000:
994:
993:
957:
946:
945:
925:
919:
918:
898:
892:
891:
874:(3): 1793–1803.
862:
856:
855:
827:
821:
820:
818:
794:
788:
783:
777:
767:
761:
740:
734:
731:
725:
721:
715:
712:
706:
705:
669:
663:
662:
660:
659:
650:. Archived from
644:
638:
633:
627:
626:
624:
600:
594:
591:
585:
579:
445:shallow learning
303:techniques like
299:, so often some
240:Machine learning
116:wavelet analysis
1574:
1573:
1569:
1568:
1567:
1565:
1564:
1563:
1544:
1543:
1542:
1541:
1513:
1509:
1470:
1466:
1435:
1431:
1400:
1396:
1341:
1337:
1322:
1300:
1296:
1273:
1269:
1246:
1242:
1187:
1183:
1152:
1148:
1119:
1115:
1091:
1087:
1080:
1066:
1062:
1031:
1024:
1001:
997:
958:
949:
926:
922:
899:
895:
863:
859:
828:
824:
795:
791:
784:
780:
776:June 6th, 2005.
768:
764:
741:
737:
732:
728:
722:
718:
713:
709:
670:
666:
657:
655:
646:
645:
641:
634:
630:
615:(27): 260–266.
601:
597:
592:
588:
580:
576:
571:
529:
507:
433:
419:
242:
236:
231:
187:
175:Gabor transform
108:thermal imaging
95:
78:
53:
51:Model-based FDI
17:
12:
11:
5:
1572:
1562:
1561:
1556:
1554:Control theory
1540:
1539:
1507:
1480:(11): 115002.
1464:
1429:
1394:
1335:
1320:
1294:
1277:Neurocomputing
1267:
1250:Neurocomputing
1240:
1181:
1146:
1113:
1102:(1): 344–360.
1085:
1078:
1060:
1041:(3): 159–190.
1022:
995:
947:
930:Neurocomputing
920:
893:
857:
822:
789:
778:
762:
752:2010-05-01 at
747:Fault Location
743:Furse, Cynthia
735:
726:
716:
707:
664:
639:
628:
595:
586:
573:
572:
570:
567:
566:
565:
560:
555:
550:
545:
540:
538:Control theory
535:
528:
525:
524:
523:
520:
514:
506:
505:Fault recovery
503:
449:dimensionality
418:
415:
343:kernel methods
328:kernel methods
282:dimensionality
262:classification
235:
232:
230:
227:
226:
225:
220:
214:
209:
204:
186:
183:
146:beat frequency
94:
91:
77:
74:
52:
49:
15:
9:
6:
4:
3:
2:
1571:
1560:
1557:
1555:
1552:
1551:
1549:
1535:
1531:
1527:
1523:
1519:
1511:
1503:
1499:
1495:
1491:
1487:
1483:
1479:
1475:
1468:
1460:
1456:
1452:
1448:
1444:
1440:
1433:
1425:
1421:
1417:
1413:
1409:
1405:
1398:
1390:
1386:
1381:
1376:
1371:
1366:
1362:
1358:
1354:
1350:
1346:
1339:
1331:
1327:
1323:
1317:
1313:
1309:
1305:
1298:
1290:
1286:
1282:
1278:
1271:
1263:
1259:
1255:
1251:
1244:
1236:
1232:
1227:
1222:
1217:
1212:
1208:
1204:
1200:
1196:
1192:
1185:
1177:
1173:
1169:
1165:
1161:
1157:
1150:
1141:
1136:
1132:
1128:
1124:
1117:
1109:
1105:
1101:
1097:
1089:
1081:
1075:
1071:
1064:
1056:
1052:
1048:
1044:
1040:
1036:
1029:
1027:
1018:
1014:
1010:
1006:
999:
991:
987:
983:
979:
975:
971:
967:
963:
956:
954:
952:
943:
939:
935:
931:
924:
916:
912:
908:
904:
897:
889:
885:
881:
877:
873:
869:
861:
853:
849:
845:
841:
837:
833:
826:
817:
812:
808:
804:
800:
793:
787:
782:
775:
771:
766:
760:June 6, 2005.
759:
755:
754:archive.today
751:
748:
744:
739:
730:
720:
711:
703:
699:
695:
691:
687:
683:
679:
675:
668:
654:on 2013-01-28
653:
649:
643:
637:
632:
623:
618:
614:
610:
606:
599:
590:
584:
578:
574:
564:
561:
559:
556:
554:
551:
549:
546:
544:
541:
539:
536:
534:
531:
530:
521:
519:
515:
512:
511:
510:
502:
500:
496:
495:deep learning
492:
488:
484:
480:
476:
472:
470:
465:
462:
458:
454:
450:
446:
442:
438:
437:deep learning
432:
431:Deep learning
423:
414:
412:
408:
404:
400:
396:
392:
388:
384:
380:
375:
371:
366:
362:
354:
350:
348:
344:
340:
336:
334:
329:
325:
321:
319:
314:
310:
306:
302:
298:
294:
292:
287:
283:
278:
274:
272:
267:
266:preprocessing
263:
259:
256:of a labeled
255:
251:
247:
241:
224:
221:
218:
215:
213:
210:
208:
205:
203:
200:
199:
198:
196:
190:
182:
180:
176:
172:
167:
163:
160:
156:
151:
147:
143:
139:
135:
131:
126:
124:
119:
117:
113:
109:
105:
100:
90:
87:
82:
73:
69:
66:
57:
48:
46:
42:
38:
33:
29:
25:
21:
1517:
1510:
1477:
1473:
1467:
1442:
1438:
1432:
1407:
1403:
1397:
1352:
1348:
1338:
1303:
1297:
1280:
1276:
1270:
1253:
1249:
1243:
1198:
1194:
1184:
1159:
1155:
1149:
1130:
1126:
1116:
1099:
1095:
1088:
1069:
1063:
1038:
1034:
1008:
1004:
998:
965:
961:
933:
929:
923:
906:
902:
896:
871:
867:
860:
838:(1): 59–68.
835:
831:
825:
809:(1): 25–33.
806:
803:High Voltage
802:
792:
781:
773:
765:
757:
738:
729:
719:
710:
677:
673:
667:
656:. Retrieved
652:the original
642:
631:
612:
608:
598:
589:
577:
508:
483:Autoencoders
473:
434:
411:steel plates
407:gas turbines
393:parts (i.e.
374:over-fitting
359:
332:
317:
290:
276:
270:
254:training set
243:
191:
188:
168:
164:
127:
120:
106:monitoring,
96:
83:
79:
70:
65:mathematical
62:
44:
40:
36:
23:
19:
18:
1355:(5): 1429.
1283:: 296–303.
1256:: 249–257.
1162:: 104–127.
1011:: 299–312.
936:: 411–422.
680:(5): 1–11.
455:. By using
399:compressors
1548:Categories
658:2011-06-07
569:References
485:are other
429:See also:
238:See also:
173:) and the
1502:123774474
1410:: 42–50.
990:125976702
968:: 33–47.
702:121684952
464:scalogram
391:machinery
311:(LDA) or
142:sidebands
138:harmonics
104:vibration
1424:53265827
1389:29734704
1235:25760051
852:23707431
750:Archived
724:464-473.
527:See also
499:datasets
370:over-fit
288:. Since
286:datasets
37:residual
1522:Bibcode
1482:Bibcode
1459:2380770
1380:5982639
1357:Bibcode
1349:Sensors
1330:6019009
1226:4435112
1203:Bibcode
1195:Sensors
1176:5850817
1055:1721126
970:Bibcode
909:: 1–8.
682:Bibcode
387:gearbox
347:dataset
307:(PCA),
258:dataset
1500:
1457:
1422:
1387:
1377:
1328:
1318:
1233:
1223:
1174:
1076:
1053:
988:
888:265207
886:
850:
700:
459:, the
453:signal
1498:S2CID
1455:S2CID
1420:S2CID
1326:S2CID
1172:S2CID
1051:S2CID
986:S2CID
884:S2CID
848:S2CID
698:S2CID
583:2005.
32:fault
1385:PMID
1316:ISBN
1231:PMID
1074:ISBN
481:and
409:and
405:and
403:wind
381:and
264:and
171:STFT
24:FDIR
1530:doi
1490:doi
1447:doi
1412:doi
1375:PMC
1365:doi
1308:doi
1285:doi
1281:151
1258:doi
1254:128
1221:PMC
1211:doi
1164:doi
1135:doi
1104:doi
1043:doi
1013:doi
978:doi
966:108
938:doi
934:218
911:doi
876:doi
840:doi
811:doi
690:doi
617:doi
397:),
179:FFT
159:RPM
155:FFT
150:FFT
134:FFT
130:FFT
123:RPM
45:FDI
1550::
1528:.
1496:.
1488:.
1478:26
1476:.
1453:.
1443:63
1441:.
1418:.
1408:53
1406:.
1383:.
1373:.
1363:.
1353:18
1351:.
1347:.
1324:.
1314:.
1279:.
1252:.
1229:.
1219:.
1209:.
1199:15
1197:.
1193:.
1170:.
1160:80
1158:.
1131:18
1129:.
1125:.
1100:10
1098:.
1049:.
1039:26
1037:.
1025:^
1007:.
984:.
976:.
964:.
950:^
932:.
907:18
905:.
882:.
872:63
870:.
846:.
836:24
834:.
805:.
801:.
772:"
696:.
688:.
678:23
676:.
613:48
611:.
607:.
477:,
413:.
401:,
389:,
335:NN
320:NN
293:NN
144:,
140:,
114:,
1536:.
1532::
1524::
1504:.
1492::
1484::
1461:.
1449::
1426:.
1414::
1391:.
1367::
1359::
1332:.
1310::
1291:.
1287::
1264:.
1260::
1237:.
1213::
1205::
1178:.
1166::
1143:.
1137::
1110:.
1106::
1082:.
1057:.
1045::
1019:.
1015::
1009:2
992:.
980::
972::
944:.
940::
917:.
913::
890:.
878::
854:.
842::
819:.
813::
807:1
704:.
692::
684::
661:.
625:.
619::
333:k
318:k
291:k
277:k
275:(
271:K
43:(
22:(
Text is available under the Creative Commons Attribution-ShareAlike License. Additional terms may apply.
