VIP High Class Call Girls Jamshedpur Anushka 8250192130 Independent Escort Se...
Cpen 11 presentation abhisar
1. Artificial Neural Network Based Fault Diagnostics for
Induction Motors in Different Machine Tool Applications
by
Mr. Abhisar Chouhan
Dr. Puruhottam Gangsar
Dr. Raj Kumar Porwal
Dr. Manoj Chouksey
SGSITS Indore (MP)
International Conference on Precision, Meso, Micro and Nano
Engineering [COPEN-11], IIT Indore
December 12-14, 2019
2. Overview of presentation
Introduction & Literature Survey
Introduction to ANN
Experimental Set-up and Procedure
Induction Motor Fault Diagnosis
Conclusion
References
2
3. Introduction and Literature Review
Introduction
Induction motors (IMs) or squirrel-cage motors,
critical component as prime mover in the simple
and complex industrial processes.
Low manufacturing and maintenance cost,
robustness and simplicity in the construction.
Reliable operation in extreme condition and
adaptability to wide variety of operation conditions
(Siyambalapitiya and McLaren, 1990).
An lathe machine
3
http://www.watertechonline.com/vfds-reduce-power-factor-
and-demand-penalties-caused-by-across-the-line-motors/
A Milling Machine
4. Introduction
In spite their reliability, faults occurs in IM.
IM fault causes, complete motor failure and
shut down of production process.
Failure results in, monetary losses,
unpredicted breakdown and serious human
injury.
A Drill Machine
4
Introduction and Literature Review
An Industrial Crane
http://www.te.com/usa-en/industries/truck-bus-off-road-
vehicles/applications/solutions-for-marine-applications.html?tab=pgp-
story
5. Introduction and Literature Review
Introduction
All IMs failure occurs due to electrical or mechanical faults.
Induction motor failure
EPRI Study motor-failure
[Nandi et al., 2005]
5
Induction
motor faults
Electrical
faults
Mehcnaical
faults
Stator faults Rotor faults
Bearing faults Rotor faults
Electrical supply
faults
· Turn-to-turn fault
· Coil-to-coil fault
· Phase-to-phase fault
· Phase-to-ground fault
· Broken rotor bar
· Broken end ring
· Unbalanced rotor
· Bowed rotor
· Misaligned rotor
· Phase unbalance
· Single phasing
· Inner race fault
· Outer race fault
· Rolling element fault
· Cage/train fault
6. Literature Review
During past twenty years, many
condition monitoring and fault
diagnosis techniques, developed to
avoid IMs failure. (Basak et al., 2006).
Machine current signature analysis
(MCSA)
Vibration
Air gap torque
Acoustic noise measurement
Temperature measurement
Introduction and Literature Review
Many literatures reported, monitoring of
IM faults based on current, vibration,
temperature and etc.
Some monitoring techniques are invasive
and deals with a single fault at a time
(Benbouzid, 2000).
An integrated monitoring based on current
and vibration can determine the root-cause
of any mechanical and electrical faults in
IMs (Noor al-deen et al.,)
6
7. Introduction and Literature Review
The current and vibration
signal analysis, meet the
aim of identifying the IM
fault successfully,
preferred in industry due
to, (Jigyasu et al.)
High precision,
Ease of measurability
Non-intrusiveness.
Methods reported for fault
diagnosis in induction
motor, (Bhavsar et al., Kia et
al., 2009)
Spectral analysis
Discrete wavelet transform
Hilbert-hung transform
Park’s vector approach etc.
Three signal processing
approaches used for fault
diagnosis of IM, (Mehrjau
et al., 2011)
Time domain,
Frequency domain
Time-frequency domain
Literature Review
7
8. Introduction and Literature Review
Literature Review
AI-techniques,
successfully used for fault
diagnosis in IM, such as
(Hamdani et al; Gangsar
and Tiwari, 2016, 2018): ).
Artificial neural network
Fuzzy logic
Immune genetic system
Support vector machine
AI techniques are data
based techniques and
advantageous. (Bhavsar et
al)
Requires only the prior-
knowledge or the history
of data.
No need of knowledge of
IM parameter and detail
modeling of faulty
8
9. After reviewing most of the papers on AI based fault diagnostics of IM, it is found that
many researchers have only considered the detection, when data is trained and test at
same operating condition.
However in actual case there are the chances that we do not have testing data at the
same operating condition has the training condition.
Therefore the diagnosis is very necessary when the data is available on limited
operating condition.
Introduction and Literature Review
10. Artificial Neural Network(ANN)
ANNs are digitized models of a human brain, it intended to simulate the manner by
which human mind forms data. learn (or are prepared) through involvement in proper
learning models as similarly as individuals do, not by programming.
Neural Network accumulate their insight by identifying the examples and connections
in information.
Artificial Neural Network
11. An ANN is based on a collection of connected units or nodes called artificial neurons,
which loosely model the neurons in a biological brain.
Each connection, like the synapses in a biological brain, can transmit a signal to other
neurons. An artificial neuron that receives a signal then processes it and can signal
neurons connected to it.
The connections are called edges. Neurons and edges typically have a weight that
adjusts as learning proceeds. The weight increases or decreases the strength of the
signal at a connection.
Neurons are aggregated into layers. Different layers may perform different
transformations on their inputs. Signals travel from the first layer (the input layer), to
the last layer (the output layer), possibly after traversing the layers multiple times.
12. Introduction
A test rig consist of Machinery Fault Simulator (MFSTM), used for fault diagnosis of IM.
Experimental test-rig
Experimental Set-up and Procedure
).
Figure 2 The IM experimental set-up
Current
probes
MFS
Accelerometer
Test-motor
Constant
DC Source
DAQ
Signal-
monitor
Tri-axial accelerometer AC current probes
Speed controller of IM
Torque controller with gear box
12
Motor-0.5 hp, 50 Hz, 3- phase IM
13. IM with various seeded faults
Ten faulty conditions of IM are considered
No. Motor fault conditions Generation of Faults
1 Healthy motor (HM) Motor with no defect
2 Broken-rotor bar fault (BRB) By drilling a hole in number of bars of rotor cage
3 Phase unbalance level-1 (PUF1) By adding resistance (max)
4 Phase unbalance level-2 (PUF2) By adding resistance (min)
5 Stator winding fault level-1 (SWF1) By adding resistance (max)
6 Stator winding fault level-2 (SWF2) By adding resistance (min)
7 Bearing fault (BF) Fault on outer race way of bearing near shaft end
8 Unbalanced rotor (UR) By attaching a balanced weight of 8.4 g to the
aluminum pins protruding from both end of the
rotor
9 Bowed rotor fault (BR) By carefully bending the rotor at the center
10 Misaligned rotor (MR)- Angular
misalignment
By displacing one end of bearing more than the
other end
Figure 2.1 Induction motors with various seeded fau
PUF
SWF BRB
MR
BR UR
BF
Rheostat-1 Rheostat-2
Experimental Set-up and Procedure
13
14. Experimental Set-up and Procedure
Data Acquisition (DAQ) and Signal Processing
Data acquisition, carried out for vibration (in three orthogonal directions)
and current signals (in three phases), acquired using tri-axial
accelerometer and AC current probes, respectively.
Raw data, acquired in time as well as frequency domain. Data acquisition system
▪ Experiments, sampling rate of 1 KHz and 10,000 sampling points. 25 raw data-sets, in 250 Sec.,
▪ Measurements, speed i.e., 10 Hz to 40 Hz in 5 Hz interval at no load condition.
14
15. Fault diagnosis of IM
The multi-class induction motor fault diagnosis is performed after obtaining the
necessary data sets from time domain data with the help of ANN classifier.
The total number of standard data sets is split into two class training and test data i.e., in
respective proportions of 80% and 20%.
An ANN model is created using the machine learning language, python, and is trained
using 80% of the data extracted from the experiments. Using the test data sets, This
model would be used to predict the faults.
For solving problems where there is shortage of data i.e., The data available is for one
speed and we have to check it for another. An ANN model is created in this work in
which two different speeds are used to train the model and is tested at intermediate
speed between the two.
16. Table1 Fault diagnosis for intermediate speed case
Train Test Train_ Test_ Individual Prediction Accuracy (%)
Speed Speed Accuracy Accuracy BF BRBF BRF MPUF MSWF ND PUF RMF SWF URF
(Hz) (Hz) (%) (%)
10,20 15 89 89.2 92 91 88 93 88 87 85 90 87 91
20,30 25 88.9 89.1 89 92 94 87 86 89 83 85 88 97
30,40 35 89.1 89.4 91 90 88 86 87 89 93 87 88 94
Here, three speed cases are considered when the testing are done at 3 distinct
intermediate speeds i.e., 15 Hz, 25 Hz and 35 Hz respectively at no load conditions.
From the table, it can be seen that, the training and testing accuracy are very close
which means the developed model is very effective.
17. Individual accuracy of the fault classification of all the 10 faults with their class labels
are also shown in the table.
Minimum and maximum individual accuracies of all the considered faults are found to
be 83 % and 97 %, respectively, in all three speed cases.
Overall accuracy of the model is found to be 89% approximately for all the speeds.
Overall accuracies for all the three different speed cases are found to be approximately
same that shows effectiveness of developed model.
Also, two different levels of severities of same faults i.e., SWF and PUF are classified
successfully here.
18. Conclusion
In this work the IM fault detection (mechanical as well as electrical faults) depending
on ANN is developed for intermediate speed cases.
The vibration (in three directions) and current data (in three phases) are used for all the
faulty conditions of IM. The data are used which are generated at 7 distinct speeds (10
Hz to 40 Hz at 5 Hz intervals).
The raw time domain data obtained by MFS is used for training and testing of an ANN
model. This optimal ANN model is finally used for diagnosis of IM faults for
intermediate speed case.
The ANN based detection of mechanical as well as electrical faults in IM is found to be
satisfactory (testing accuracy 89%, approximately) using vibration and current data of
time domain.
Also, the developed model is able to detect different severity levels of the same defects
i.e. MSWF and SWF, MPUF and PUF, effectively.
19. References
[1] Bhavsar, R. C., Patel, R. A., & Bhalja, B. R. (2014, July). Condition monitoring of induction motor using
artificial neural network. In 2014 Annual International Conference on Emerging Research Areas: Magnetics,
Machines and Drives (AICERA/iCMMD) (pp. 1-6). IEEE.
[2] Hamdani, S., Touhami, O., Ibtiouen, R., & Fadel, M. (2011, September). Neural network technique for induction
motor rotor faults classification-dynamic eccentricity and broken bar faults. In 8th IEEE Symposium on Diagnostics
for Electrical Machines, Power Electronics & Drives (pp. 626-631). IEEE.
[3] Kareem A. Noor Al-Deen, Marina E. Karas, Ahmed M. Abdel Ghaffar, CyrilleCaironi, Bernhard Fruth and
DetlefHummesSignature Analysis as a Medium for Faults Detection in Induction Motors.
[4]Gangsar, P., & Tiwari, R. (2017). Comparative investigation of vibration and current monitoring for prediction of
mechanical and electrical faults in induction motor based on multiclass-support vector machine
algorithms. Mechanical Systems and Signal Processing, 94, 464-481.
[5]Gangsar, P., & Tiwari, R. (2016). Taxonomy of induction-motor mechanical-fault based on time-domain vibration
signals by multiclass SVM classifiers. Intelligent Industrial Systems, 2(3), 269-281.
[6]Gangsar, P., & Tiwari, R. (2018). Multifault diagnosis of induction motor at intermediate operating conditions
using wavelet packet transform and support vector machine. Journal of Dynamic Systems, Measurement, and
Control, 140(8), 081014.
[7]Zhongming, Y., & Bin, W. (2000, August). A review on induction motor online fault diagnosis. In Proceedings
IPEMC 2000. Third International Power Electronics and Motion Control Conference (IEEE Cat. No. 00EX435) (Vol.
3, pp. 1353-1358). IEEE.
[8] Nejjari, H., & Benbouzid, M. E. H. (2000). Monitoring and diagnosis of induction motors electrical faults using a
current Park's vector pattern learning approach. IEEE Transactions on industry applications, 36(3), 730-735.
[9] Yang, B. S., Oh, M. S., & Tan, A. C. C. (2009). Fault diagnosis of induction motor based on decision.