Sie sind auf Seite 1von 8

QUT Digital Repository:
Yang, Hongyu and Mathew, Joseph and Ma, Lin (2002) Intelligent diagnosis of rotating
machinery faults - A review. In: 3rd Asia-Pacific Conference on Systems Integrity and
Maintenance, ACSIM 2002, 25-27 September 2002, Cairns, Australia.

Copyright 2002 (please consult author)

Intelligent Diagnosis of Rotating Machinery Faults-A Review
Hongyu Yang, Joseph Mathew & Lin Ma
School of Mechanical Manufacturing and Medical Engineering, QUT, Brisbane, QLD 4001, Australia
The task of condition monitoring and fault diagnosis of rotating machinery faults is both significant
and important but is often cumbersome and labour intensive. Automating the procedure of feature
extraction, fault detection and identification has the advantage of reducing the reliance on experienced
personnel with expert knowledge. Various diagnostics methods have been proposed for different types
of rotating machinery. However, little research has been conducted on synthesizing and analysing
these techniques, resulting in apprehension when technicians need to choose a technique suitable for
application. This paper presents a review of a variety of diagnosis techniques that have had
demonstrated success when applied to rotating machinery and highlights fault detection and
identification techniques based mainly on artificial intelligence approaches. The literature is
categorised in the following diagnostic groups: neural networks, fuzzy sets, expert systems, and
hybrid AI techniques based fault diagnosis. The paper concludes with a brief description of a new
approach to diagnosis using a Wavelet based Coactive Artificial Neuro-Fuzzy Inference System
(CANFIS) which the authors plan to develop and implement for diagnosing machine faults.
1. Introduction
Todays industry uses increasingly complex machines, some with extremely demanding performance criteria.
Attempting to diagnose faults in these systems is often a difficult and daunting task for operators and plant
maintainers. Failed machines can lead to economic loss and safety problems due to unexpected and sudden
production stoppages. These machines need to be monitored during the production process to improve machine
operation reliability and reduce unavailability. Therefore, conducting effective condition monitoring brings
significant benefits to industry [1, 2]. However, condition monitoring requires effective fault diagnosis, which is a
labour oriented exercise to this day. Without efficient diagnosis, one is unable to make reliable prediction of lead-
time to failure. A natural progression is the automation of this labour oriented process of diagnosis by implementing
intelligent diagnosis strategies so that experts or technicians can be relieved of this relatively expensive task.
Fault diagnosis is conducted typically in the following phases: data collection, feature extraction, and fault
detection and identification. Fault detection and identification usually employs artificial intelligent (AI) approaches
for pattern classification. Numerous attempts have been made to improve the accuracy and efficiency of fault
diagnosis of rotating machinery by employing AI techniques. Few have attempted to summarise these techniques
comprehensively. Zhong [3] introduced new developments in the theory and application of intelligent condition
monitoring and diagnostics in China. He concluded that the trends in intelligent diagnosis are NN-based fault
classifiers, NN-based expert systems, NN-based prognosis, behaviour-based intelligent diagnosis, remote distributed
intelligent diagnosis networks and intelligent multi-agent architecture for fault diagnosis. He provided a good
overview of intelligent fault diagnosis of machinery but was somewhat general.
Pham [4] theoretically analysed the applicability of artificial intelligence in engineering problems and
predominantly looked at knowledge-based systems, fuzzy logic, inductive learning, neural networks and genetic
algorithms in different branches of engineering but not in machinery fault diagnosis. Tandon [5] mentioned that
automatic diagnosis was a trend in the fault diagnosis of rolling elements. Gao [6] provided an up-to-date review on
recent progresses of soft computing methods-based motor fault diagnosis systems. He summarized several motor
fault diagnosis techniques using neural networks, fuzzy logic, neural-fuzzy, and genetic algorithms (GAs) and
compared them with conventional techniques such as direct inspection. Chow [7] also gave a brief review, which
listed 14 papers from experts in the area of motor fault detection and diagnosis. He grouped those papers into five
main categories: survey papers, model-based approaches, signal processing approaches, emerging technology
approaches, and experimentation.
This paper presents a review of the application of Artificial Intelligence (AI) techniques in fault diagnosis of
rotating machinery. Rotating machinery comprises rolling bearings, gears, pumps, chains, belts, and electric motors.
Common faults of rotating machinery include spalling and crack defects in inner, outer race and rolling elements of
ball/roller rolling bearings, cracked shaft, and misalignment in gears. In the literature, it appears that diagnostics
techniques for different rotating machinery parts can have much in common. Previous researches have been
conducted predominantly on specific components and faults in rotating machinery. This paper provides a review of
techniques concerned with rotating elements and shows the potential of generic application of techniques over a
range of rotating elements.
The second section of the paper presents a summary of AI based intelligent fault techniques. These are grouped
into the following categories: neural networks, fuzzy sets, expert systems, and hybrid approaches. The limitations
and strengths of these different techniques are addressed respectively. The paper concludes with a brief description

of an intelligent fault diagnosis schema titled CANFIS, Coactive Neuro-Fuzzy Inference System. The authors are
currently in the process of implementing CANFIS in a laboratory environment.
2. Intelligent fault diagnosis- a schema
An overall schema for intelligent diagnostics is presented in Figure 1. Intelligent diagnosis begins with the act of
data collection which is followed by feature extraction usually employing the frequency spectra. Feature extraction
techniques are widespread and can range from statistical to model based techniques and comprises a variety of
signal processing algorithms which includes wavelet transforms. Fault detection and identification is a subsequent
step and is further classified in this review into the four categories shown in the figure - these will now be treated

Figure 1. Intelligent fault diagnosis- a schema
3. Neural networks-based techniques
In general intelligent diagnosis is carried out when known inputs are fed into black boxes which subsequently
produce outputs in accordance with machine faults. Neural networks (NN) are suitable for these tasks and have been
widely researched as an artificial intelligence tool for machinery fault diagnosis [8]. By employing such a tool,
maintenance personnel need not understand or operate the internal mechanisms of a neural network. He or she will
only be responsible for inputting the appropriate data to a neural network. The neural network will then be trained
on this data so that it can diagnose faults.
Neural networks have become widely adopted in rotating machinery for fault diagnosis due to their learning
ability. The learning ability of NN makes neural networks capable of tackling a new problem by making use of
existing information. In addition, neural networks also have the capacity to model complex nonlinear problems
which may approximate real world fault diagnosis.
In rotating machinery fault diagnosis, a neural network is usually trained on features extracted from obtained
signals. The neural network is usually viewed as a fault classifier. To date, several types of neural networks [9] have
been applied in machinery fault diagnosis including a feedforward neural network (FFNN), Back Propagation for
Feed Forward Networks (BPFF), Multi Layer Perceptrons (MLP), Back Propagation Multiplayer Perceptrons
(BPMP), Radial Basis Function networks (RBF), Self Organised Maps (SOM), Principal Component Analysis
(PCA), and a recurrent neural network (RNN).
Neural networks, which have been successfully applied in rotating machinery fault detection and identification,
are listed in Table 1. The table provides a preliminary summary of the type of NN used and its application in
diagnosing faults in common machine components.
Table 1
Neural networks applied in diagnosing rotating machinery faults
Neural networks pumps rolling bearings gears gearboxes shafts fans motor

BPFF [10] [11]
FFNN [12] [12]
RNN [12] [12]
RBF [8] [13]
BP [14, 15] [8] [16] [14] [17]
MLP [18, 19] [18, 19] [19] [13, 18] [20] [19]
Kohonen SOM [21]
LVQ [18] [18] [18]

detection and
Neural networks Fuzzy sets Expert systems
Hybrid AI

Baillie and Mathew [8]s results in diagnosing faults of rolling element bearings indicated that backpropagation
(BP) neural networks generally outperformed the radial basis functions (RBF). McCormick and Nandi [13] used
multi-layer perceptron and radial basis function neural networks to classify the condition of a small rotating
machine. It was found that both networks achieved similar success in fault classification. However RBF networks
could be trained in a significantly shorter length of time. Multilayer perceptrons (MLP) however required fewer
neurons and were faster in recall operation.
Meesad and Yen [18] applied MLP and Learning Vector Quantization (LVQ) classifiers to diagnosing faults in
gears, bearings, and shafts. It was proven that these two neural networks were successful while they required off-line
training and iterative presentation of the training data. They were cumbersome when applied to pattern classification
problems that needed fast, on-line, real-time, incremental learning.
A new diagnostic algorithm using Kohonens network proposed by Tanaka [21] can classify data obtained from
a motor in plant. The proposed diagnostic algorithm was compared with BP-networks for both fault detection and
fault identification. In Tanakas application, it was proven that the performance of the Kohonen SOM was better
than BP networks. It was also concluded that the map size of the Kohonens network was a factor which affected its
The performance of neural networks is sometimes subjective. The selection of network structure may affect the
network performance to a significant extent. Some neural networks can have very slow convergence speeds.
Training these neural networks take a substantial length of time. Even worse, training can fall victim to the law of
diminishing returns. In particular, the slow operating speed of some neural networks is unable to meet the
requirement that machinery be diagnosed on line or in certain timeframes. Furthermore, a neural network can output
local optimisation thereby not guaranteeing an optimal solution. This disadvantage of neural networks can lead
directly to errors in diagnosis. Yet another limitation of NN is the lack of semantics [22].
4. Fuzzy set-based fault diagnosis techniques
Fuzzy logic-based fault diagnosis methods have the advantages of embedded linguistic knowledge and
approximate reasoning capability. The Fuzzy logic proposed by Zadeh [23] performs well at qualitative description
of knowledge. However, the design of such a system depends heavily on the intuitive experience acquired from
practicing operators thus resulting in subjectivity of diagnosed faults. The fuzzy membership function and fuzzy
rules cannot be guaranteed to be optimal in any case. Furthermore, fuzzy logic systems lack the ability of self-
learning, which is compulsory in some highly demanding real-time fault diagnosis cases [6]. Rough set based
intelligence diagnostic systems have been constructed and used in diagnosing valves in three-cylinder reciprocating
pumps [24] and turbogenerators [25].
5. Expert system-based fault diagnosis techniques
Expert systems are computer programs embodying knowledge about a narrow domain for solving problems
related to that domain. An expert system usually comprises two main elements, a knowledge base and an inference
mechanism. The knowledge base contains domain knowledge which may be expressed as any combination of IF-
THEN rules, factual statements, frames, objects, procedures and cases [4].
A fault diagnosis expert system is a system with a knowledge base storing the accumulated experience of fault
diagnostics experts. However, expert systems can only diagnose faults stored in the knowledge base and are thus
unable to tackle new problems which have not been classfied, ie, a fault diagnosis expert system cannot diagnose
new machinery faults when compared to real human expertise. Furthermore, building knowledge bases is labour
oriented and time consuming. This often makes other AI techniques as the preferred ones due to their flexibility and
efficiency in diagnosing rotating machinery faults.
Available expert systems for rotating machinery fault diagnosis are limited. Amethyst [26] is an expert system
to assist with vibration based condition monitoring of rotating machinery. Vibration patterns are collected from
machinery such as pumps, fans, motors and generators during normal operating conditions. Another expert system
[27] developed by El Adawi performs preventive maintenance tasks and detects faults/failure during standard
operating cycles. Georgin et. al [28] carried out research based on two of EDF's diagnostic expert systems, DIVA
(for turbine-generators) and DIAPO (for primary cooling pumps), which interactively deals with users with different
knowledge levels. Shao [29] proposed a new concept on the degree of credibility of parameter value variations
(DCPV factor) and developed an expert system for online monitoring and diagnostics of rolling element bearings.
6. Hybrid AI techniques based fault diagnosis
Neurofuzzy computing recently emerged as an alternative technique to diagnosing rotating machinery faults
[18, 30-43]. Neurofuzzy techniques work in the following way: A fuzzy set interpretation is incorporated into the
network design to handle imprecise information. A neural network architecture is used to automatically deduce
fuzzy if-then rules based on a hybrid supervised learning scheme. For instance, Altmann and Mathew successfully
applied an adaptive neural fuzzy inference system (ANFIS), using a zero-order sugeno fuzzy model (shown in

Figure 2) in rolling element bearings fault diagnosis [43]. They adopted a Discrete Wavelet Packet Analysis
(DWPA) multiple band-pass to extract features, which provided input data sets, checking data sets, and testing data
sets for ANFISs for various types of rolling bearing faults. These included diagnosis of inner race faults, rolling
element faults, and outer race faults respectively. Given training and checking input/output data sets, the
membership function parameters were adjusted using a back-propagation algorithm in combination with a least
squares method. Checking data was used to cross-validate and test the generalisation capability of the fuzzy
inference system. According to Altmann and Mathews application of ANFISs, it was concluded that fault
classification through fusion of band-pass fuzzy confidence factors resulted in a successful classification rate of
100% for the signals examined, with 95% requiring no additional analysis for verification of the fault classification.
The 5% of signals requiring additional analysis were broken into, 3% due to an inadequate average confidence level
for a definitive diagnosis to be made (30%<ACL<50%), and 2% that consisted of three or fewer wavelet packets
indicating the presence of bearing fault related transients [43].

input input
rule output




x y
x y




normalisation node





Figure 2: A Zero-Order Sugeno Fuzzy Model
Fuzzy expert systems which fuse fuzzy logic and expert systems have been explored [44-46]. Nabeshima [47]
investigated a technique fusing neural networks, fuzzy logic, and expert systems. Jack [48] combined genetic
algorithms used for feature selection with neural networks in machinery fault diagnosis with limited success.
7. Concluding comments
Although much progress has been made on intelligent fault diagnosis of rotating machinery, there is no doubt
that current techniques still have limitations. A significant challenge is the application of these techniques to
industry in a reliable, efficient, precise, and real-time manner indicating that there is significant scope for
additional research into intelligent fault diagnosis techniques.
The authors have embarked on a project to improve the ANFIS technique for machinery fault diagnosis. ANFIS
was designed as a single output inference system (see Figure 2), thus each ANFIS system can only successfully
diagnose one type of machine fault. We now propose a significant enhancement through a Coactive ANFIS
(CANFIS) technique which is capable of multiple outputs, thus diagnosing multiple faults, at the one time. Certain
inference rules will be refined based on Altmanns work [1]. The CANFIS application we propose involves
diagnosing faults in bearings and will use advanced feature extraction techniques, such as the DWPA as its input. A
schema for the proposed work is shown in Figure 3.
Initially a demodulation process will be conducted including band pass filter and full wave rectification.
Subsequently a DWPA best basis [1] for the collected signal is calculated in the joint time-frequency domain.
Wavelet coefficients which are mainly affected by the bearing faults are determined and extracted. The DWPA
demodulation process will have a significant advantage over the normal demodulation process, as no knowledge of
the resonant frequency is required. Further, a basis that is well matched to the characteristics of the signal will be
chosen to improve the representation of bearing impact resonance. Features from the frequency range of interest can

then be extracted, while contributions from noise and other components outside the frequency range of interest can
be substantially reduced. To reduce the noise more efficiently, a hard threshold denoising method will be adopted.
An alternative feature extraction technique is the spectrogram which is defined as the square of STFT.

The spectrogram shows promise in extracting features [49] and can be used with the Kurtosis function to further
process data. The proposed schema will be able to automatically output diagnosed machinery faults according to
collected signals. The results of this work will be presented in due course.

Collect Data

Band Pass
Full Wave
matched basis

[1] Altmann, J. (1999) Application of Discrete Wavelet Packet Analysis for the Detection and Diagnosis of Low
Speed Rolling-Element Bearing Faults. PhD Thesis, Monash University: Melbourne.
[2] Baillie, D C and Mathew, J (1996), A comparison of autoregressive modelling techniques for fault diagnosis
of rolling element bearings, Journal of Mechanical Systems and Signal Processing, Vol 10(1), pp. 1-17.
[3] Zhong, B. (2000) Developments in intelligent condition monitoring and diagnostics. in System Integrity and
Maintenance , the 2nd Asia-Pacific Conference(ACSIM2000), pp 1-6, Brisbane, Australia.
[4] Pham, D.T. and P.T.N. Pham. (1999) Artificial intelligence in engineering. International Journal of Machine
Tools & Manufacture, 39, 937-949.
[5] Tandon, N. and A. Choudhury. (1999) A review of vibration and acoustic measurement methods for the
detection of defects in rolling element bearings. Tribology International, 32(8), 469-480.
[6] Gao, X.Z. and S.J. Ovaska. (2001) Soft computing methods in motor fault diagnosis. Applied Soft
Computing, 1(1), 73-81.
[7] Chow, M.-Y. (2000) Guest editorial special section on motor fault detection and diagnosis. Industrial
Electronics, IEEE Transactions on, 47(5), 982-983.
( ) ( ) ( ) ( )
, ,

= = t t t
t t
d e t w x f t STFT f t SPEC
f j
x x
Figure 3: A DWPA demodulation based neuro-fuzzy fault diagnosis approach

[8] Baillie, D. and J. Mathew. (1994) Diagnosing Rolling Element Bearing Faults with Artificial Neural
Networks. Acoustics Australia, 22(3), 79-84.
[9] Jang, J.-S.R., C.-T. Sun, and E. Mizutani. (1997) Neuro-fuzzy and soft computing: a computational approach
to learning and machine intelligence. MATLAB curriculum series. Upper Saddle River, NJ: Prentice Hall.
xxvi, 614.
[10] Wang, K. and B. Lei. (1998) Genetic algorithms for constructing feed forward multiple layered neural
network in a centrifugal pump condition monitoring. Intelligent Engineering Systems Through Artificial
Neural Networks, 8, 303-310.
[11] Lu, Q. and D. Li. (1998) Neural network method for diagnosing faults rolling bearing in electrical machines
with frequency signatures. Qinghua Daxue Xuebao/Journal of Tsinghua University, 38(4), 94-97.
[12] Parlos, A.G., S.K. Menon, and A.F. Atiya.(1999) Adaptive state estimation using dynamic recurrent neural
networks. in Neural Networks, 1999. IJCNN '99. International Joint Conference on. Dept. of Nucl. Pp: 3361-
3364, vol.5. Eng., Texas A&M Univ., College Station, TX, USA.
[13] McCormick, A.C. and A.K. Nandi.(1996) A comparison of artificial neural networks and other statistical
methods for rotating machine condition classification, IEE Colloquium on Modelling and Signal Processing
for Fault Diagnosis (Ref. Leicester, UK: Dept. of Electron. & Electr. Eng. Strathclyde Univ. Glasgow
UK,IEE Colloquium on Modelling and Signal Processing for Fault Diagnosis (Ref. No.1996/260). Pp. 100.
IEE London UK.
[14] Priddy, K.L., M.D. Lothers, and R.E. Saeks.(1993) Neural networks and fault diagnosis in rotating
machinery. in Systems, Man and Cybernetics, 1993. 'Systems Engineering in the Service of Humans',
Conference Proceedings. International Conference on. pp.640-644 vol.2. Accurate Autom. Corp.,
Chattanooga, TN, USA: Practical.
[15] In Soo, K. and K. Whan Woo. (2000) The development of reactor coolant pump vibration monitoring and a
diagnostic system in the nuclear power plant. ISA Transactions, 39(3), 309-16.
[16] Engin, S.N. and K. Gulez.(1999) A wavelet transform-artificial neural networks (WT-ANN) based rotating
machinery fault diagnostics methodology,Proceedings of the IEEE-EURASIP Workshop on Nonlinear Signal
and Image Processing (NSIP'99). pp 894, vol. 2. Antalaya, Turkey: Dept. of Electr. Eng. Yildiz Tech. Univ.
[17] Zhao, L. and Z. Sheng. (1996) Combination of discrete cosine transform with neural network in fault
diagnosis for rotating machinery. Proceedings of the IEEE International Conference on Industrial
Technology, 450-454.
[18] Meesad, P. and G.G. Yen. (2000) Pattern classification by a neurofuzzy network: application to vibration
monitoring. ISA Transactions, 39(3), 293-308.
[19] Paya, B.A., I.I. Esat, and M.N.M. Badi. (1997) Artificial neural network based fault diagnostics of rotating
machinery using wavelet transforms as a preprocessor. Mechanical Systems & Signal Processing, 11(5), 751-
[20] Lowes, S. and J.M. Shippen. (1997) A diagnostic system for industrial fans. Measurement and Control,
30(1), 9-13.
[21] Tanaka, M., et al. (1995) Application of Kohonen's self-organizing network to the diagnosis system for
rotating machinery: Fac. of Eng. Hiroshima Univ. Japan,1995 IEEE International Conference on Systems
Man and Cybernetics. Intelligent Systems for the 21st Century (Cat. No.95CH3576-7). Pp: 4711, vol. 5.
[22] Shulin, L. and S. Wengang.(2001) Rough set based intelligence diagnostic system for valves in reciprocating
pumps,Proceedings of IEEE International Conference on Systems, Man & Cybernetics. Tucson, AZ, USA:
Dept. of Mech. Eng. Daqing Pet. Inst. Heilongjiang China,2001 IEEE International Conference on Systems
Man and Cybernetics. e-Systems and e-Man for Cybernetics in Cyberspace (Cat.No.01CH37236). Pp. 3494,
vol 5. IEEE Piscataway NJ USA.
[23] Zadeh, L.A. (1965) Fuzzy sets. Information and Control, 8, 338-353.
[24] Liu, S. and W. Shi.(2001) Rough set based intelligence diagnostic system for valves in reciprocating pumps.
in Systems, Man, and Cybernetics, 2001 IEEE International Conference on. Dept. of Mech. Eng., Pp. 353-
358, vol.1. Daqing Pet. Inst., Heilongjiang, China: Practical.
[25] Hu, T., B.C. Lu, and G.J. Chen.(2000) A Rotary Machinery Fault Diagnosis Approach Based on Rough Set
Theory. in the 3rd World Congress on Intelligent Control and Automation. Pp. 589-685. Hefei, China.
[26] Milne, R. (1992) Amethyst: Intelligent Applications Ltd. Kirkton Bus. Centre Livingston UK,IEE
Colloquium on 'Intelligent Fault Diagnosis - Part 1: Classification-Based Techniques' (Digest No.045). Pp.
32. IEE London UK.
[27] El Adawi, S., et al. (1992) Computer based expert system for rotating machinery (preventive and predictive
maintenance): Dept. of Mech. Power Zagazig Univ. Egypt,Proceedings of the Second IASTED International
Conference. Computer Applications in Industry. pp. 585, vol. 2. ACTA Press Zurich Switzerland.
[28] Georgin, E., et al. (1994) The importance of cases and domain models in explanation, Proceedings of ISAP
'94. Montpellier, France: Centre for Electr. Power Eng. Strathclyde Univ. Glasgow UK, ISAP '94.

International Conference on Intelligent System Application to Power Systems. Pp: 894, vol.2. EC2 Nanterre
Cedex France.
[29] Shao, Y. and K. Nezu. (1996) An online monitoring and diagnostic method of rolling element bearing with
AI. Transactions of the Society of Instrument and Control Engineers, 32(8), 1287-93.
[30] Yen, G.G. and P. Meesad. An effective neuro-fuzzy paradigm for machinery condition health monitoring.
IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics, 31(4), 523-536.
[31] Vilim, R.B., H.E. Garcia, and F.W. Chen. (1997) Machine condition monitoring using neural networks and
the likelihood function. Intelligent Engineering Systems Through Artificial Neural Networks, 7, 653-659.
[32] Emmanouilidis, C., J. MacIntyre, and C. Cox.(1998) An integrated, soft computing approach for machine
condition diagnosis: Sch. of Comput. & Inf. Syst. Sunderland Univ. UK,6th European Congress on
Intelligent Techniques and Soft Computing. EUFIT '98. Pp. 2010, vol. 3. Verlag Mainz Aachen Germany.
[33] Hao, L. and X. Xu. (2001) The application of rough set neural network system in fault diagnosis. Control
Theory & Applications, 18(5), 681-5.
[34] Kesheng, W. and L. Bing. (2001) Using B-spline neural network to extract fuzzy rules for a centrifugal pump
monitoring. Journal of Intelligent Manufacturing, 12(1), 5-11.
[35] Melvin, D.G. and J. Penman. (1995) Fusing human knowledge with neural networks in machine condition
monitoring systems. Proceedings of the SPIE The International Society for Optical Engineering, 2492(1),
[36] Satoh, S., M.S. Shaikh, and Y. Dote.(2001) Fast fuzzy neural network for fault diagnosis of rotational
machine parts using general parameter learning and adaptation,SMCia/01. Blacksburg, VA, USA: Dept. of
Comput. Sci. & Syst. Eng. Muroran Inst. of Technol. Japan,SMCia/01. Proceedings of the 2001 IEEE
Mountain Workshop on Soft Computing in Industrial Applications (Cat. No.01EX504). Pp. 134. IEEE
Piscataway NJ USA.
[37] Satoh, S., M.S. Shaikh, and Y. Dote.(2001) Fault diagnosis for dynamical systems using soft
computing,Proceedings of IEEE International Conference on Systems, Man & Cybernetics. Pp. 585, vol. 2.
Tucson, AZ, USA: Dept. of Comput. Sci. & Syst. Eng. Muroran Inst. of Technol. Japan.
[38] Yakuwa, F., et al.(2002) Fault diagnosis for dynamical systems using soft computing. in Fuzzy Systems,
2002. FUZZ-IEEE'02. Proceedings of the 2002 IEEE International Conference on. Pp. 261-266, vol. 1.
Muroran Institute of Technology.
[39] Taniguchi, S., D. Akhmetov, and Y. Dote. (1999) Fault detection of rotating machine parts using novel fuzzy
neural network. Proceedings of the IEEE International Conference on Systems, Man and Cybernetics, 1, I-
365 - I-369.
[40] Taniguchi, S., D. Akhmetov, and Y. Dote.(1999) Fault detection of rotating machine parts using novel fuzzy
neural network, IEEE SMC'99 Conference Proceedings. Pp. 1179, vol. 6. Tokyo, Japan: Dept. of Comput.
Sci. & Syst. Eng.
[41] Taniguchi, S., et al.(2000) Nonlinear modeling and fault detection using fuzzy-neural network: Dept. of
Comput. Sci. & Syst. Eng. Muroran Inst. of Technol. Japan,Proceedings of the ISCA 9th International
Conference Intelligent Systems.pp. 158. Int. Soc. Comput. & Their Appl. - ISCA Cary NC USA.
[42] Hsiao, I.-L., et al. (1999) On-line fault diagnosis of rotor vibration by using signal-based feature generation
and neural fuzzy inference. Journal of the Chinese Society of Mechanical Engineers, Transactions of the
Chinese Institute of Engineers, Series C/Chung-Kuo Chi Hsueh Kung Ch'eng Hsuebo Pao, 20(4), 345-352.
[43] Altmann, J and Mathew J (2001), Multiple Band-Pass Auto-Regressive Demodulation for the Detection and
Diagnosis of Faults in Low Speed Rolling-Element Bearings, Journal of Mechanical Systems and Signal
Processing, Volume 15, Issue 5, pp. 963-977, September.
[44] Siu, C., Q. Shen, and R. Milne.(1997) A fuzzy expert system for vibration cause identification in rotating
machines. in Fuzzy Systems, 1997., Proceedings of the Sixth IEEE International Conference on. Pp. 555-560,
vol. 1. Dept. of Artificial Intelligence, Edinburgh Univ., UK.
[45] Feng, E., H. Yang, and M. Rao. (1998) Fuzzy expert system for real-time process condition monitoring and
incident prevention. Expert Systems with Applications, 15(3-4), 383-390.
[46] TI Liu, JH Shigonahalli, NR Iyer.(1996) Detection of Roller Bearing Defects using Expert systems and
Fuzzy logic. Journal of Mechanical Systems and Signal Processing, 10(5), 595-614.
[47] Nabeshima, K., et al. (2002) Nuclear reactor monitoring with the combination of neural network and expert
system. Mathematics and Computers in Simulation, In Press.
[48] Jack, L.B. and A.K. Nandi. (2000) Genetic algorithms for feature selection in machine condition monitoring
with vibration signals. IEE Proceedings: Vision, Image and Signal Processing, 147(3), 205-212.
[49] Wang, W.J. and P.D. McFadden. (1993) Early detection of gear failure by vibration analysis--ii.
interpretation of the time-frequency distribution using image processing techniques. Journal of Mechanical
Systems and Signal Processing, 7(3), 205-215