LabVIEW Based Implementation of Fuzzy Logic

for Vibration Analysis to Identify Machinery Faults

Rohit V. Mukane *, Nandkumar M. Gurav † , Shilpa Y. Sondkar ‡ , Neville C. Fernandes §
Department of Instrumentation, Vishwakarma Institute of Technology
Pune, Maharashtra, India 411037
[email protected] *, [email protected] ‡
Research & Development, Forbes Marshall Pvt. Ltd.
Pune, Maharashtra, India 411018
[email protected] † , [email protected] §
Abstract—Rotating machine is device that supports an impor- about machine faults; therefore, it is essential to examine the
tant category of manufacturing industry. Machine fault detection non-stationary signals.
is very advantageous for diagnosis of faults before it occurs and Time-frequency analysis has demonstrated to be an effec-
to protect the machinery from catastrophic damages. Vibration
monitoring is important for running rotating equipment evenly tive resource for analyzing the behavior of non-stationary
for years. There are diverse reasons that causes vibrations in signals. There are diverse time-frequency analysis systems,
the rotating machinery like rotor unbalance, machine looseness, like Short-Time Fourier Transform (STFT), Bilinear Time-
permanent bow (bend shaft or warped shaft ), bearing damage, Frequency Distribution function and Wavelet Analysis (WA),
misalignment, eccentricity, oil whirl, cracked tooth, etc. There are which may applied for fault detection of machinery in diverse
several methods used to detect machinery faults such as, model
based technique and signal based techniques which includes
operating conditions. Fuzzy logic approach is advantageous in
Hilbert-Huang Transform (HHT), Wavelet Transform (WT), Ar- manufacturing industries and classification of faults. Vibration
tificial Neural Networks (ANN), Support Vector Machine (SVM), signals are not always exactly repeatable, it may sometimes
Shock Pulse Monitoring (SPM), Fuzzy Logic. The proposed work even be contradictory. Fuzzy technology has the efficiency to
shows a novel approach of classification using fuzzy logic for understand the uncertainty that exist during fault diagnosis.
decision making about the machine condition. The experimental
results shows that, algorithm implemented identifies severity of
Hence, a rule based fuzzy system is applied for machinery faults
faults to appropriate condition. classification problem. This paper put forward the fuzzy logic
Index Terms—Rotating machine, Fuzzy logic, Fault diagnosis, algorithm for identification & analysis of different machinery
Signal processing, LabVIEW. faults.

II. L ITERATUR E S URVEY

I. I NT RODUC T I ON
There are diverse studies and researches that describes the
Rotating machines are often applied in a class of mechanical precise monitoring and control of machinery fault diagnosis.
systems, together with aircraft engines, automotive transmis- It is an important task to develop effective algorithm for
sion systems and power plants. Emergent failures in machinery machinery fault diagnosis which can adapt the changes in
may account in breakdowns, leading to significant losses and, machines at various stages of operation. Fateh Harrouche [1]
even worse, human casualties. To avoid disastrous failures have chosen a fuzzy system to track and diagnose defects
and minimize defective machinery breakdown, machinery fault in rotating machines (application to bearings only). M.Yang
evaluation is of great industrial importance. Condition of [2] has offered a fuzzy system for turbo machinery diagno-
machines rely on, the feature exploration of vibration data, sis, which combines fuzzy logic & weight approaches for
it is essential that the feature of the signal can be delicately representing and reasoning about domain specific knowledge.
extracted at the state change of a machine. The system is not applicable for unknown fault condition.
Signal analysis techniques are the most significant methods Hichem Merabet [3] have studied the fault detection of short-
used for machinery fault detection, in which the objective is circuit fault in turns of stator windings and open stator phases
to find effective transform of the original signals. In this way, in doubly-fed induction generator by fuzzy logic. The work
essential information implied in the signals may be shown then represented have studied only on the simulation of healthy
dominant features of signals extracted for fault diagnostics. machine. Shorouk Ossama Ibrahima [4] has developed an
However for machinery operating under unsteady conditions, intelligent system for discovering fault of the 3-phase induc-
even if the machine is in the natural state, the spectrum aspect tion motor by using Matlab/Simulink without faults (healthy
of vibration signal is always changing because of difference drive) and under faults (faulty drive). In this case modelling of
in rotating motion and operating load. The frequency features induction motor is applied as an identifier for different faults.
of a vibration signal will change with time. Usually, non- Mariela Cerrada [5] have applied fuzzy logic for determining
stationary components contain large amount of information fault intensity condition in helical gears. The algorithm links

978-1-5386-4008-1/17/$31.00 ©2017 IEEE

Authorized licensed use limited to: Rajamangala Univ of Technology Srivijaya provided by UniNet. Downloaded on August 31,2023 at 06:23:31 UTC from IEEE Xplore. Restrictions apply.
 Mamdani models & hierarchical clustering to define the mem-
bership degrees to fault intensity levels of samples extracted
from historical vibration signals. The system is validated on
the basis of historical vibration data and assumptions also
training of samples is required for calculating Weights of
Fuzzy Transitions(WFT). Huaqing Wang [6] have developed
the system that it can effectively detect faults for machinery
in the variable speeds, a proposed method rely on time-
frequency analysis, fuzzy inference, and automatic feature
extraction method. This approach can only be applied when Fig. 1. System Description
a bearing has constant rotating speed and relies in the fact
that relative crossing information(RCI) allows to automatically
extract feature spectra from the time-frequency distributions of
the vibration signals. V.P. Mini [7] have presented fuzzy logic
analysis for defective rotor bar detection and for analyzing
the severity of fault. The model of a 3-phase induction
motor developed, under healthy condition is contracted as the
context for fault pick- ups and analysis. R.Saravana Kumar
[8] have developed system which states that, motor condition
is recognize using a compositional rule of fuzzy inference.
The work were restricted to the induction motor only and not
tested for other faults in machinery. Vijay Prakash Pandey
[9] have applied fuzzy logic to diagnosed induction motor stator
and phase conditions, based on amplitude features of stator Fig. 2. Experimental Setup
currents. Matlab/Simulink model were developed for
monitoring system output.
It is observed that most of researchers uses various types IV. P ROPOSED M ET HODOLOGY
of algorithms for monitoring faults in machinery but there is
need for more efficient method to be implement for better The proposed methodology utilizes fuzzy logic for ma-
performance in fault diagnosis. Classification of machinery chinery fault classification. Fig. 3 indicates the flowchart of
faults using fuzzy logic has been proven ability in imitate proposed system. This algorithm is rely on the rule base system,
human decisions and various condition monitoring problems IF-THEN statements were applied for creating rule base. The
has typically been solved. output data shows us multiple faults severity. Fuzzy logic
III. S YS TEM D ESCR IP TIO N classifier is very much helpful in case, new parameters are to
be added to algorithm and existing rule base is to be modify
The proposed test system consist of a shaft, one balancing according to system requirements.
disc, two bearing blocks, a base, electric motor and VFD
Fuzzy systems depends on a set of rules. Thus, an engineer
(Variable Frequency Drive). Two sensors are connected on the
might refer to a machine faults as normal, medium or high [12]
first bearing block & two sensors on second bearing block.
[15]. In the proposed algorithm frequency signal amplitudes
Non-contact type phase marker is attached to the system for
used as input variables and machine fault conditions like
calculation of phase. The rotor kit consist of electric motor
unbalance, bend shaft, machine looseness were taken as output
which is coupled to the rotating shaft. The fuzzy logic is ap-
variables of the classifier. Interpretation of result is difficult as
plied for the classification of machinery faults. It consist of six
relationships between the machine condition and the vibration
stages as, Input variables, Normalization, Fuzzification, Rule
amplitudes are vague. Therefore, using fuzzy sets numerical
base, Defuzzification, Denormalization [11]. Fig. 1 indicates
data are indicated as linguistic information.
general block diagram of system.
The experimentation was conducted on rotor kit. Fig. 2 Fuzzy rules and membership functions are made by ob-
shows the real time physical system. The data acquisition serving the data set. For the measurements concerned to the
device for the experimentation included a NI USB-4432 card rotor kit vibration amplitudes and machine conditions, more
with a BNC connector, an accelerometer CTC AC102-1A & presence into the data is needed, so membership functions will
SEC RN-M/AC102-1A and velocity sensors FM-VEL-004B. be generated for all input/output variables. Proposed algorithm
LabVIEW was used to record both acceleration and velocity implemented using LabVIEW is shown in fig. 4. Membership
signals. Sampling rate for analysis was 25600 Hz. This test plot of input/output is indicated in Fig. 5 and Fig. 6. Input
rig was tested under several speeds and load conditions. In variables are splits into 3 subsets as, small, medium & high.
validation process, only data collected at maximum load and Output variables are splits into 3 subsets as, normal, medium
speed were used in the research. & high.

2017 Third International Conference on Computing, Communication, Control And Automation (ICCUBEA)
Authorized licensed use limited to: Rajamangala Univ of Technology Srivijaya provided by UniNet. Downloaded on August 31,2023 at 06:23:31 UTC from IEEE Xplore. Restrictions apply.
 suitably high for better results viz. 25600 samples/sec. It is
observed that because of variable nature of rotor kit used in fault
detection system all sets of readings have different values. Fig. 7
shows the original vibration signal in time domain, acquired for
10 seconds, through the sensors placed on test rig. In proposed
method FFT algorithm is applied on the acquired vibration
signals to get frequency waveform. Fig. 8 shows the FFT plot
of no load machine condition. It shows the amplitude of
vibration data in frequency domain.

Fig. 7. Vibration Signal under No load Condition

Fig. 3. Flowchart of Proposed Methodology

Fig. 8. FFT of Vibration Signal under No load Condition

Fig. 4. Proposed Algorithm Implemented using LabVIEW

Fig. 9. FFT of Vibration Signal under Unbalance Condition

Fig. 5. Input Variable Membership Function

Fig. 10. FFT of Vibration Signal under Bend Shaft Condition

Fig. 9 shows the frequency change in amplitude in accor-

dance to unbalance condition as 1X amplitude is observed in
spectrum. In fig. 10 1X and 2X harmonics are observed axially,
Fig. 6. Output Variable Membership Function it indicates that there is bend in shaft. In fig. 11 multiple
harmonics like 1X, 2X, 3X and 4X are observed it indicates
V. E XPERIM ENTAT I ON that there is machine loosening. For the validation process,
machinery fault characteristics were used as reference [10]
During experimentation different sets of readings with same [13][14].
operating conditions were taken. Sampling rate were made

2017 Third International Conference on Computing, Communication, Control And Automation (ICCUBEA)

Authorized licensed use limited to: Rajamangala Univ of Technology Srivijaya provided by UniNet. Downloaded on August 31,2023 at 06:23:31 UTC from IEEE Xplore. Restrictions apply.
 Fig. 11. FFT of Vibration Signal under Machine Looseness Condition

VI. R ESU LTS AND D IS C U SS ION

During experimentation rotor test rig were considered for
Fig. 14. Output of Proposed Algorithm for Machine Looseness Condition
modeling. Different faults were introduced in test rig for
the validation of the proposed algorithm. Fig. 12 shows the
proposed output of classifier under unbalance condition. It
indicates that, rule no. 2 is invoked and normal unbalance
is detected. Fig. 13 shows the proposed output under bend
shaft condition, It also indicates that rule no. 11 is invoked
and showing the medium bend shaft fault. Fig. 14 shows
the proposed output for machine looseness condition, it also
indicating that rule no. 27 is invoked and showing medium
fault condition.

Fig. 15. Output of Proposed Algorithm for Multiple Fault Condition

VII. C ON C L U SI ON
This paper presents LabVIEW based vibration analysis for
classification of machinery faults using fuzzy logic algorithm.
Experimental results for proposed classifier shows, unbalance,
bend shaft and machine looseness fault conditions and rule
no. 2, 11 and 27 resp. were invoked and it shows that
Fig. 12. Output of Proposed Algorithm for Unbalance Condition normal unbalance of 20.0049, medium bend shaft of 50.0345,
and medium looseness of 65.0705 were indicated as output
values of machine faults. The proposed classifier is applied
for multiple faults occurring at the same time, as experimental
results shows that, multiple faults like unbalance, machine
looseness and bend shaft were observed with output values
79.9987, 36.2086 and 50.0142 resp. This simulation and exper-
imental studies shows the implementation of fuzzy system for
vibration analysis to identify machinery faults. This LabVIEW
based machinery fault classifier can be further used in various
machines fault classification with modifying the pre-defined
rule base according to the system requirements.
A C KNOW L EDGM ENT
We are sincerely grateful to Forbes Marshall Pvt. Ltd,
Fig. 13. Output of Proposed Algorithm for Bend Shaft Condition
Pimpri-Pune for extending the support to complete this re-
search work.
In fig. 15, there are multiple faults are observed at same time
with different fault severity. Unbalance, Machine Looseness, R EF ERENC E S
and Bend Shaft conditions are observed in the output of [1] Fateh Harrouche, Ahmed Falkaoui, Hocine Feniniche and Ishak Touafek,
classifier. As it indicates that multiple rules are invoked. In the Application of Fuzzy Logic (FL) in Diagnosis of Rotating Machines, 15th
proposed LabVIEW-VI algorithm off-line analysis of system International Conference on Experimental Mechanics, Porto/Portugal, 22-
27 July 2012.
is taken into consideration.

2017 Third International Conference on Computing, Communication, Control And Automation (ICCUBEA)

Authorized licensed use limited to: Rajamangala Univ of Technology Srivijaya provided by UniNet. Downloaded on August 31,2023 at 06:23:31 UTC from IEEE Xplore. Restrictions apply.
 [2] M.Yang and Q.Shen, Fuzzy Diagnosis of Turbo Machines, International
Journal of Knowledge-Based and Intelligent Engineering Systems, Vol
12, no. 2, pp. 137-146, 2008.
[3] Hichem. Merabet, Tahar. Bahi and Noura Halem, Condition Monitoring
and Fault Detection in Wind Turbine Based on DFIG by the Fuzzy Logic,
Energy Procedia 74,518 - 528, 2015.
[4] Shorouk Ossama Ibrahima, khaled Nagdy Farisb and Esam Abo Elzahab,
Implementation of fuzzy modeling system for faults detection and diag-
nosis in three phase induction motor drive system, Journal of Electrical
Systems and Information Technology 2, 27 - 46, 2015.
[5] Mariela Cerrada, Chuan Lic, Rene-Vinicio Sanchezb, Fannia Pachecob,
Diego Cabrerab and Jose Valente de, A fuzzy transition based approach
for fault severity prediction in helical gearboxes, Fuzzy Sets Syst. (2016).
[6] Huaqing Wang and Peng Chen, Fuzzy Diagnosis Method for Rotating
Machinery in Variable Rotating Speed, IEEE SENSORS JOURNAL,
VOL. 11, NO. 1, JANUARY 2011.
[7] V. P. Mini and S. Ushakumari, Rotor Fault Detection and Diagnosis of
Induction Motor using Fuzzy Logic, AMSE JOURNALS 2014-Series:
Modelling A; Vol. 87; N 2; PP 19-40.
[8] R.Saravana Kumar, K.Vinoth Kumar and Dr. K.K.Ray, Fuzzy Logic
based fault detection in induction machines using Lab view, IJCSNS
International Journal of Computer Science and Network Security, VOL.9
No.9, September 2009.
[9] Vijay Prakash Pandey and Prashant Kumar Choudhary, Induction Motor
Condition Monitoring Using Fuzzy Logic, Advance in Electronic and
Electric Engineering. ISSN 2231-1297, Volume 3, Number 6 (2013), pp.
755-764.
[10] Donald E. Bently, Charles T. Hatch and Bob Grissom, Fundamentals of
Rotating Machinery Diagnostics, Bently Pressurized Bearing Press, 2002.
[11] Lotfi A. Zadeh, Is there a need for fuzzy logic?, Information Sciences
178, 27512779, 2008.
[12] Ying Bai and Dali Wang, Fundamentals of Fuzzy Logic Control Fuzzy
Sets, Fuzzy Rules and Defuzzifications, Advanced Fuzzy Logic Technolo-
gies in Industrial Applications. ISBN 978-1-84628-468-7, XXV, 334p,
2006.
[13] Xiaofeng Liu, Lin Ma and Joseph Mathew, Machinery Fault Diagnosis
Based on Feature Level Fuzzy Integral Data Fusion Techniques, Proceed-
ings IEEE International Conference on Industrial Informatics, pages pp.
857-862, Singapore, 2006.
[14] SKF Reliability Systems, Vibration Diagnostic Guide, CM5003 (Revised
9-00), 2000.
[15] Rui Francisco Martins Maral and Kazuo Hatakeyama, Expert System
Based on Fuzzy Rules for Monitoring and Diagnosis of Operation
Conditions in Rotating Machines, Proceedings of PICMET, Infrastructure
and Service Integration, 2014.

2017 Third International Conference on Computing, Communication, Control And Automation (ICCUBEA)

Authorized licensed use limited to: Rajamangala Univ of Technology Srivijaya provided by UniNet. Downloaded on August 31,2023 at 06:23:31 UTC from IEEE Xplore. Restrictions apply.