IoT based Condition Monitoring of Generators and
Predictive Maintenance
D.Swathi
Department of EEE
Anurag group of Institutions
Hyderabad, India
swathi.diddi15@gmail.com
MD.Yaseen
Department of EEE
Anurag group of Institutions
Hyderabad, India
yaseeneee@cvsr.ac.in
Dr.T.Anil Kumar
Department of EEE
Anurag group of Institutions
Hyderabad, India
thalluruanil@gmail.com
Abstract—This paper presents a IOT (Internet of Things)
based solution for condition monitoring and predictive
maintenance of generator set by establishing a communication
between Electronic Hardw are and Cloud Computing, popularly
known as IOT based applications used especially for online and
real time monitoring. Generator is heart of any power system
used for power generation. Generally Generator suffers from
abnormal conditions such as over loading and vibrations, to
predict these abnormalities an IoT based system is proposed to
avoid unnecessary shutdowns. In the implementation of IOT
based system, different sensors such as vibration sensors,
current, temperature sensors are used to capture some essential
parameters to monitor health condition of generator. Faults are
being simulated using IOT based hardware equipment for test
purpose and to create alters to the respected field operator on
web page application. The signal obtained from vibration sensors
of time domain are converted in to frequency domain by using
FFT algorithm in gateway and waveform patterns are analyzed
for fault detection. Edge analytics has been carried out locally
and efficient early warning system was created based on trends
observed on web application.
Keywords—generators; condition monitoring; internet of
things; edge analytics; vibration analysis.
I. INTRODUCTION
Due to increase in renewable energy resources, especially
wind energy and solar energy day by day, it is making
necessary fossil power plants to operate in medium and peak
load range to stabilize electrical grid. The higher number of
load cycles associated with the medium and peak load
operation leads to thermo mechanical stressing of core
components such as generators. This typically causes
unexpected failures, aging of the machines, risk of damage,
loss of production, unplanned maintenance etc. Therefore
vibration and temperature analysis are the important
monitoring techniques for the rotating machines [1].
One important measure which can help unexpected
generator damage is to monitor the generator operation and
analyze the measured data irregularities. A continuous
machine condition monitoring is needed which can reliability
and accurately transfer the real time data from the equipment
to the monitoring system such that any abnormality in the
system can be founded easily and necessary corrective action
can be taken without any time delay [2]. A successful online
monitoring can help to avoid unexpected trips and optimize
outages. The differences between the real time data and
desired performance behavior of a system allow the operators
to predict and identify the problems before they cause the
equipment to damage thereby reducing the unexpected failures
and the consequences [2].
Some measuring devices like sensors will add significant
knowledge of the performance of the machine. Different types
of sensors can be used which can sense the key parameters
which are needed to monitor continuously [3]. Data received
directly from the sensors are presented in a variety of plots
that depict exactly what is occurring through online
monitoring. Typical signal processing techniques is applied to
these sensor signals to extract particular features which are
sensitive to the presence of faults. Finally, in the fault
detection stage, a decision needs to taken to check whether a
fault exists or not. The principal objective of these efforts is to
increase overall reliability and availability of the equipment
without loss of production as well as to plan for systematic
maintenance in a cost effective way. While vibration, thermal
and current signature are the key analysis which can be
applied to machines to monitor its health condition. In this
paper these analysis is done through various sensors like
accelerometer sensor, temperature sensor and Hall Effect
current sensor [3].
Proceedings of the 2nd International Conference on Communication and Electronics Systems (ICCES 2017)
IEEE Xplore Compliant - Part Number:CFP17AWO-ART, ISBN:978-1-5090-5013-0
978-1-5090-5013-0/17/$31.00 ©2017 IEEE 725
Data collection, storage, analysis and early alert
system are important for efficient monitoring of a machine.
Recent advances in technology such as IOT ( internet of
things) and cloud computing created a new era in this process
and which will also provide a platform for edge analytics and
provides best GUI (graphical user interface) [3,4].IOT is a
computing concept which provides an effective intelligence to
the system by establishing a proper internet based
communication[4].
Due to electrical and mechanical forces in rotating
machines there exits vibrations which vibrate at different
frequencies and amplitudes in the system and these vibrations
may increase along with time and may lead to damage of the
equipment [5]. Therefore it is essential to mitigate and analyze
the behavior of these vibrations for safe and reliable operation
and there are several standards proposed to regulate the
vibration levels in machines [6].
Analyzing the vibrations can be done by using real
time data of ADXL345 sensor from data acquisition gateway
and by applying proper signal conditioning technique called
FFT. The time domain data from accelerometer sensor is
converted into frequency domain by using FFT algorithm in
the gateway and required software. The trends in the vibration
frequency spectrum is simulated and observed frequently to
know the problems such as unbalance, misalignment, bearing
defect, resonance, looseness and electrical problems [6].
Overloading, transient fluctuations and insulation
failure will increase the temperature at bearings, windings and
there will be variations in the stator current. Thermal and
overload current monitoring is important to extend machine
performance as the life of insulation decreases because of over
temperatures, so these monitoring can be done by using
temperature and current sensors and in this paper the signal
obtained from these sensors is analyzed by using analytics
applied in edge devices [7].
Edge computing provides a corrective solution for the
delay in IOT applications and will not only act like
intermittent connectivity and data analytics. Data from the
sensors will be sent to edge enabled gateway in which local
database has been created to store the data locally and rules
are written such that any variation in the temperature and
current parameters greater than its limits will create an alert at
the end user instead of sending data for centralized data store
where the data from all generator nodes will be sent for cloud
analytics. This real time running mechanisms provides light
weighted algorithms in edge level reduced the complexity in
real time decision [8].
II. RELATED WORK AND REQUIRMENTS
Fig.1 shows the IOT solution for monitoring of generators
in which sensors are connected to gateway and internet
connectivity which provides a better online monitoring when
compared to traditional methods. The advantages of this
method is once the data is transmitted it can be accessed from
anywhere at any time.
A. Sensors Used
The monitoring parameters considered to diagnose the
generators health are vibrations at different locations, driving
end and non driving end bearing temperatures, winding
temperature, current which is done by using following sensors.
1) ADXL345
The type of vibration sensor to be used is a Micro
Electrical-Mechanical System (MEMS) accelerometer shown
in Fig.2. This type has advantages over other vibration sensor
technologies as it is very compact, can measure a wide range
of frequencies, is reliable and can have high sensitivity.
Accelerometers with analogue or digital interfaces are
available, and a SPI and I2C digital interface was chosen as it
is more resilient against signal interference and again, does not
require the ADC of the MCU. The chosen accelerometer, the
ADXL345 from Analog Devices, offers three axes of
measurement, with the ability to measure ± 16 g along each
axis, can survive a shock of up to 10,000 g and it can measure
static and dynamic vibrations. The sensor returns data in a 16-
bit, twos complement format. Some tuning and adjustment of
mounting location and orientation will be required in order to
detect the suitable range of frequencies for health monitoring
[9].
2) LM135
LM135 is used as temperature sensor which can be
used to sense winding temperature and bearing temperature of
the motor. The LM135 sensor shown in Fig.3 is precision,
easily-calibrated, integrated circuit temperature sensors which
can give linear output. The LM135 operates over a –55°C to
150°C temperature rang and has a breakdown voltage directly
proportional to absolute temperature change of 10 mV/°K
performance[10].
Fig.1. Structure of IOT framework
Fig.2. Accelerometer Fig.3. LM 135
Proceedings of the 2nd International Conference on Communication and Electronics Systems (ICCES 2017)
IEEE Xplore Compliant - Part Number:CFP17AWO-ART, ISBN:978-1-5090-5013-0
978-1-5090-5013-0/17/$31.00 ©2017 IEEE 726
3) ACS712
ACS712 sensor shown in Fig.4 is used to measure the
stator current of a generator for electrical analysis. Generally,
this kind of is called as Hall Effect sensor, it works based on
principle Hall Effect, The Hall effect is the production of
a voltage difference (the Hall voltage) across an electrical
conductor, transverse to an electric current in the
conductor.ACS712 current sensors can be easily used with
microcontrollers like arduino [11].
B. Toggle Switch
It is a versatile switch and can be easily used to integrate
with different applications. Toggle switch shown in Fig.5
consists of two states momentary on and momentary off and it
can be described as a mechanical or software switch that can
be altered between two states each time when it is actuated .In
this paper it is used as mechanical simulation switch to
simulate a fault and test the system whether the data
transmission and alert system is incorporated correctly or not
[12].
C. Gateway
IOT gateway device bridges the communication gap
between devices, sensors, equipment, systems and the cloud.
Devices can communicate with gateway with wired
communication like USB, I2C, SPI or wireless communication
such as WIFI, Bluetooth, Zigbee, wireless USB, RFID etc.,
Gateways can control equipment based on data received from
sensors and they offer local pre processing and storage
solutions [13].
1) Arduino Mega
Arduino is an open source hardware and software
which acts like a gateway to collect sensor data, analyzing it
and to connect to a network. The Arduino Mega 2560 shown
in Fig.6 is a microcontroller board based on the ATmega2560
is designed to be programmed in C and C++ programming
languages. It has 54 digital input/output pins and 16 analog
inputs and it have 256 KB of flash memory for storing code.
Arduino integrated development environment (IDE) which
simplifies the coding, debugging, compiling and loading of the
program onto the MCU. There are many shields are boards
that can be plugged on top of the arduino or can be serial
communicated with it [14].
2) Raspberry Pi
Raspberry pi shown in Fig.7 is a fully featured single
board micro computer with WIFI and Bluetooth connectivity.
It consists of Broadcom BCM2837 64bit quad core 1GB
RAM processor running at 1.2GHz and 4 USB ports, 40
GPIO pins, 4 pole Stereo output, Composite video port, Full
size HDMI CSI camera port, DSI display, Micro SD port for
loading your operating system and storing data, Micro USB
power source[15].
D. Cloud Platform
Cloud platform provides sufficient processing, storage,
analytics, networking and user interface for IOT applications.
IOT is a computing concept that describes a scenario where
physical objects are connected to the internet and network
connectivity enables these objects to collect and exchange data
and can identify themselves to other devices via IP address
[16].
III. IMPLEMENTATION OF PROPOSED METHOD
The block diagram of the proposed work shown in
Fig.8, it describes the monitoring of generators. It consists of
different sensors are attached to various generators and
simulation switches
The data from the sensors and switches are given to
microcontroller unit, Algorithms are written in arduino ide as
shown in Fig.9 such that the logics for FFT and simulations
are written and processed according to the data received and
data received from sensors will be sent to edge gateway
Raspberry pi is used as edge gateway in this method which
is used for network connectivity to send the data to cloud and
also for local database as shown in Fig.10 for edge analytics. It
clearly shows data storage of considered parameters for
monitoring of generators. Rules are written in edge database
based on slope variation of parameters and to send commands
to particular operator
Fig.4. Current Sensor Fig.5. Toggle Switch
Fig.6. Arduino Mega Fig.7. Raspberry pi
Fig.8. Block Diagram
Proceedings of the 2nd International Conference on Communication and Electronics Systems (ICCES 2017)
IEEE Xplore Compliant - Part Number:CFP17AWO-ART, ISBN:978-1-5090-5013-0
978-1-5090-5013-0/17/$31.00 ©2017 IEEE 727
Generator set data under normal operating conditions is
used as reference and is compared with faulted condition by
using the below formula.
K=max(P0)-min(Pn)/max(S0)-min(Sn)
Where k is called as slope, P0 and Pn are the maximum and
minimum variation of a parameter in the number of samples
taken. Parameter can be a winding temperature, bearing
temperature, current and vibration and S0 and Sn are the first
and last samples taken in order to calculate the variation of
parameter, Usually number of samples are taken in the order
of 2n by the microcontroller.
By equation (1) rate of change of parameter are
calculated and compared with k under normal operating
condition and based on trends observed real time alerts are
created. Web application as shown in Fig.11 is created for
graphical user interface for online monitoring of generators.
Separate login credentials are given for administrator and field
operator. When any parameter exceeded its limits alerts will
be sent to particular field operator.
IV. RESULTS
Once the sensor nodes and gateways are connected the data
can be accessed on GUI through application program
interface(API) keys.API provides an web interface between
generator sensor platform and internet by providing a secure
authentication [17].The main idea of the model is to transfer
data for monitoring and realize the data analytics functions by
simulating faults through switches. This will ensure the
accuracy and reliability to know the performance whether
alerts are sent to field operator effectively or not when there
are any abnormalities present. Fig.12 shows the serial plotter
of arduino ide software.
The graph shows the presence of various frequencies with
respect to sampling frequency and it clearly describes the
presence of second and first harmonic exits with respect to
samples received which represents a misalignment of coupling
[18]. Fig.13 shows the time domain data of vibration sensor
and frequencies magnitudes present on web page, where time
domain plot represents the sinusoidal waveform which
consists of various harmonics and frequency plot is called as
spectrum which need to be analyzed in order to differentiate
the fault.
Database created for edge analytics in local host shows the
variation of slope in non driving end bearing temperature and
winding temperature as shown in Fig.14. Large variations in
slope of temperatures with respect to pre defined value will
cause insulation failure.
Fig.9. Arduino IDE
Fig.10. Database
Fig.11. Web Application
Fig.12. Serial Plotter
Fig
.13. Time a
nd Frequency
Plots
Proceedings of the 2nd International Conference on Communication and Electronics Systems (ICCES 2017)
IEEE Xplore Compliant - Part Number:CFP17AWO-ART, ISBN:978-1-5090-5013-0
978-1-5090-5013-0/17/$31.00 ©2017 IEEE 728
Fig.15 shows the variation of non driving end temperature
and winding temperature and alert shown due its variation in
winding temperature in local host as shown in Fig 16.
V. CONCLUSION
Successful monitoring of the generator depends on
the availability of real time information which reflects the
health state of a machine. In this paper, mainly describes the
need of transmitting real time data from the sensors to GUI.
The main advantage of this method is to minimize the outages,
reduce loss of production, decision based maintenance due to
early recommendations provided and thereby extending the
life of operation of generators. The reliability of monitoring
system has to be high in order to meet the need of continuous
monitoring and because of communication which is
established due to this network connectivity provides a
efficient online condition monitoring solution for generators in
IOT platform. The proposed method is simulated under
various operating conditions as described above, and is found
to work satisfactorily.
REFERENCES
[1] Peter Kung, lutang Wang, Maria Lulia Comanici,”Stator end winding
vibration and temperature rise monitoring,”2011 Electrical Insulation
Conference (EIC), pp. 10-14, June 2011.
[2] Bernhard Fruth, Herbert Looser, Detlef Hummes, Wojciech Betlej,
“Integrated remote insulation condition monitoring of generators and
motors using traditional and wireless sensor technologies,” 2014
International Conference on Computer, Communications and Control
Technology (14CT), pp. 91-96, September 2014.
[3] Suratsavadee Koonlaboon Korkue, Wei-Jen Lee, “Wireless Sensor
network for permance monitoring of electrical machines,” North
American Power Symposium(NAPS), pp. 1-5, Octobor 2009.
[4] Lane Thames and Dirk Schaefer, “Software Defined Cloud
Manufacturingfor Industry 4.0,” Vol. 52, pp. 12-17, 2016.
[5] Luciane Agnoletti S. Pedotti, Ricardo Mazza Zago, Fabiano Fruett,
“Instrument based on MEMS accerometer for vibration and unbalane
analysis in rotating machines,” 2016 1
st International Symposium on
Instrumentation Systems, Circuits and Trasducers (INSCIT), pp.25-30,
DOI: 10.1109/INSCIT.2016.7598212, September 2016.
[6] ISO 2372: Mechanical vibration of machines with operating speeds from
10 to 200 rev/s – Basis for specifying evalution standards, ISO Std,1974.
[7] Hongye Guo, Lin Ruan, “Design of wireless powered RF temperature
monitoring system applied to the rotor of generators,” 2011 International
Conference on Electrical Machines and System, pp. 1-4, DOI:
10.1109/ICEMS.2011.6073631, August 2011.
[8] Stefan Nastic, Thomes Raush, Ognjen Scekic, Schahrak Dustdar, Marjan
Gusev, Bojana Koteska etc., “A Serverless Real Time Data Anaiytics
Platform for Edge Computing,” IEEE Internet Comput ing, Vol. 2 1,
Issue. 4, pp. 64-71, July 2017.
[9] Georgios A. Skrimpas, Ivaylo G. Dragiev, Reynir Hilmisson, Christian
W. Sweeney Bogi B. Jensen, Nenand Mijatovic, Joachim Holboll,
“Detection of generators bearing race creep by means of vibr ation and
temperature analysis,” 2015 IEEE 10th International Symposium on
Diagnostics for Electrical Machines, Power Electronics and Drives, pp.3
303-309, DOI:10.1109/DEMPED.2015.7303706, September 2015.
[10] http://www.ti.com/product/LM135
[11] Medilla Kusriyanto, Bambang Dwi Putra, “Smart home using local area
network (LAN) based arduino mega 2560,” 2016 2nd International
Conference on Wireless and Telematics (ICWT), pp. 127-131, DOI:
10.1109/ICWT.2016.7870866, August 2016.
[12] https://learn.sparkfun.com/tutorials/switch-basics
[13] http://internetofthingsagenda.techtarget.com/feature/Using-an-IoT-
gateway-to-connect-the-Things-to-the-cloud/
[14] Humayum Rashid, Sarjahan Shekha, S M Taslim Reza, Iftekhar Uddin
Ahmed, Qader Newaz, Md.Rasheduzzaman, “A low cost automated
fluid control device using smart phone for medical application,” 2017
International Conference on Electrical, Computer and Communication
Enginerring, pp. 809 – 814,DOI: 10.1109/ECACE.2017.7913014, 2017.
[15] http://in.element14.com/raspberry-pi/raspberrypi-modb-1gb/raspberry-
pi-3-model-b/dp/2525226
[16] Abdur Rahim Biswas; Raffaele Giaffreda, “ Iot and cloud convergence:
Oppurtunities and Challenges”, 2014 IEEE World Forum on Internet of
Things,” pp. 375 – 376, DOI: 10.1109/WF-IOT.2014.6803194, 2014.
[17] https://cloud.google.com/docs/authentication/api-keys
[18] http://www.vibescorp.ca/learn-about/advanced-vibration-analysis/
Fig.14. Slope Variation in Local Data base
Fig.15
.
Unstable Temperature plots
Fig.16. Alerts in Local host
Proceedings of the 2nd International Conference on Communication and Electronics Systems (ICCES 2017)
IEEE Xplore Compliant - Part Number:CFP17AWO-ART, ISBN:978-1-5090-5013-0
978-1-5090-5013-0/17/$31.00 ©2017 IEEE 729