Low-Cost Vibration Measurement for Behavior of Small-Scale Steel Modeling using MEMS, Raspberry Pi-3 and Arduino Mega 2560

Abstract

The paper presents structural health monitoring (SHM) approach for measuring vibration responses of small-scale steel structure by using low-cost MEM Sensors, Raspberry Pi-3 and Arduino Mega 2560. The MPU-6050 (GY521) with 6-DOF including 3-axis gyroscope and 3-axis accelerometer uses standard I2C bus for data transmission through cheap microcontroller boards and open-source softwares to send commands and receive data in real-time with different sampling rates. Vibration signals of MEMS are analyzed by fast fourier transform (FFT) algorithm under the frequency spectrum to predict fundamental frequencies and compared with measured signals via the wireless structural testing system (STS-WiFi) manufactured by the Bridge Diagnostics incorporation in the USA using industrial accelerometers.

Full text

Hội nghị Khoa học toàn quốc Cơ học Vật răn lần thứ XIV,
Đại học Trần Đại Nghĩa, Thành phố Hồ Chí Minh Nội, 19-20/7/2018
Low-Cost Vibration Measurement for Behavior of Small-Scale Steel
Modeling using MEMS, Raspberry Pi-3 and Arduino Mega 2560
Nguyen Cong Duc1,*, Huynh Quoc Hung2, Phan Cong Ban2, Tran Van Mot2,
Nguyen Cong Minh3, Pham Bao Toan4 and Ngo Kieu Nhi4
1,2 Mien Trung University of Civil Engineering (MUCE) - Minister of Construction, Phu Yen
3 Air Force Officers College - Ministry of National Defence, Nha Trang
4 Ho Chi Minh City University of Technology (HCMUT), Viet Nam National University
*Email: nguyencongduc@muce.edu.vn
Abstract. The paper presents structural health monitoring (SHM) approach for measuring
vibration responses of small-scale steel structure by using low-cost MEM Sensors, Raspberry Pi-3
and Arduino Mega 2560. The MPU-6050 (GY521) with 6-DOF including 3-axis gyroscope and 3-
axis accelerometer uses standard I2C bus for data transmission through cheap microcontroller
boards and open-source softwares to send commands and receive data in real-time with different
sampling rates. Vibration signals of MEMS are analyzed by fast fourier transform (FFT) algorithm
under the frequency spectrum to predict fundamental frequencies and compared with measured
signals via the wireless structural testing system (STS-WiFi) manufactured by the Bridge
Diagnostics incorporation in the USA using industrial accelerometers.
Keywords: Arduino, Raspberry Pi-3, MPU6050, MEMS, Vibration, Acceleration, FFT.
1. Introduction
In recent year, the structural health monitoring (SHM) play an important role in load testing and
rating for bridges, train rail systems, dams, buildings and other structures (steel, concrete, timber,
composite-fiber material) throughout live load tests and several long-term tests. One of the commonly
used live load testing for structures is vibration analysis based on the FFT algorithm. In Vietnam, the
demand for load testing approaches and vibration monitoring is rapidly increasing. The spending of
money for issues has been very high cost, partly because technologies and devices are supplied by
companies and manufacturers oversea. Part of explanation lies in engineers need to training as
certified technical experts dependent on types of different systems and programming software also
must response with requirements of standards in the country. The specific objectives of the study are
select low-cost efficiency devices for monitoring vibration of structures.
Some previous studies applied monitoring vibration to predict fundamental frequencies of
structures. Jerome P. Lynch et al. [8] reported in their work that the 32-bit Motorola MPC555
PowerPC microcontroller was selected to connect with piezoresistive and ADXL210 accelerometer for
measuring time-history responses of structural modeling and comparing frequency response of the
system from two sensors. Samuels, Reyer, Hurlebaus, Lucy, Woodcock and Bracci [11] have
investigated long-term vibration testing (8 days) of Frankford church (Dallas, Texas, USA) using
Mica2 mote and MTS310 sensor board. The wireless structural health monitoring (SHM) system with
low-cost Tmote Sky devices carried out by Tat S. Fu1, Amitabha Ghosh, Erik A. Johnson and Bhaskar
Krishnamachari [5] which can identify the structural characteristics of Four-story building model by
estimating fundamental frequencies and calibrating mode shapes.
Recently, Spencer et al [12] developed wireless SHM application including: QuickFilter
QF4A512 and TI ADS1274, Imote2 and ITS400 board for monitoring vibration, strain, temperature
and wind on structures. Thanh-Canh Huynh, Jae-Hyung Park and Jeong-Tae Kim [7] performed in
vibration monitoring of Hwamyung Cable-Stayed Bridge (Busan, Korea) by using twelve Imote2
hardware nodes (Memsic Co.) and using power spectral density (PSD) analysis to investigate

Nguyen Cong Duc, Huynh Quoc Hung, Phan Cong Ban, Tran Van Mot, Nguyen Cong Minh,
Pham Bao Toan and Ngo Kieu Nhi
2
historical responses. Moreover, Ge-Wei Chen, Sherif Beskhyroun and Piotr Omenzetter [1] conducted
experimental studies of the vibration responses and numerical FEM modal analysis of the Nelson St.
off-ramp Bridge (Auckland, New Zealand) under Eccentric mass shakers and weak ambient
excitations through the Battery-powered USB Tri-axial MEMS Accelerometer X6-1A produced by
Gulf Coast Data Concepts, LLC (Waveland, USA).
More recently, Feldbusch, Sadegh-Azar and Agne [4] used the smartphone “iDynamics”
application with MEMS-Accelerometers to measure vibration amplitudes and natural frequencies of
pedestrian bridge, to evaluate vibration strength according to the German standards: DIN 4150-2, DIN
4150-3 and VDI 2038. David Hester, James Brownjohn, Mateusz Bocian and Yan Xu [6] worked on
different low-cost MEMS devices including: JA-70SA triaxial accelerometer (Japan Aviation
Electronics), QA-750 Accelerometer (Honeywell), Single Axis K-Beam MEMS 8315A (Kistler),
triaxial MEMS Opal (APMD Inc) and triaxial MEMS Kionix KXRB5-2050 (Gulf Coast Data
Concepts), which the goal of this study could calculate displacement from the double integration of the
acceleration signals, and also compared capacity of MEMS devices.
The most familiar paper of this issue is Guido Morgenthal and Hagen Höpfner [9] studied
MEMS technology of modern smartphones for estimating transient displacements and predicting
natural frequencies of the structures, also show limitations of vibration measurement using MEMS-
based sensors of HTC phones via Electrodynamic Shaker and sonar distance measurement. Also,
Guido Morgenthal, Sebastian Rau, Jakob Taraben and Tajammal Abbas [10] performed
experimentally to predict the cable forces of The Queensferry Crossing (QFC) in Scotland, by
programming the application of smartphones (Sony Xperia Z5, Nexus 4) and the setup with RPi
together with the MPU6050 sensor for determination of the fundamental frequency. Furthermore,
Research in the area of infrastructure SHM, Kosmas Dragos and Kay Smarsly [3] have studied to
update FEM modeling and the wireless Oracle SunSPOTs nodes with 8-bit MMA7455L accelerometer
for measuring vibration of the four-story shear frame structure.
2. Experimental investigation and Results
The test facility used to conduct vibration measurement is shown Figure 2, the system consists
of BDI Accelerometer (±5g), LVDT sensors, two wifi Nodes and a Base Station which will send and
receive data from sensors to wireless laptop using Internet Protocol Version (IPv4) (address:
192.168.10.2). The node-wifi module can be connected with four different sensors, and wifi base
station can control nodes and managed by WinSTS software.
Figure 1. Low-cost vibration measuring devices
Figure 2. The wireless structural testing
system (STS-WiFi, USA)
MPU6050
Arduino Mega 2560
Raspberry Pi 3
Battery 500mAh

Low-Cost Vibration Measurement for Behavior of Small-Scale Steel Modeling using MEMS,
Raspberry Pi-3 and Arduino Mega 2560
3
The low-cost instruments used in the experimental study is Arduino Mega 2560, Raspberry Pi3
with with MPU6050 (±2g, 4g, 8g, 16g) for recording vibration signal of steel model described in
Figure 1. The sample rate and scale of low-cost MEMS will compared with the STS-WIFI (BDI -
USA) of the Mien Trung University of Civil Engineering (license 2015). The mounting of the test
steel model and the arrangement of the MEMs and LVDT transducers and vibration motor on the top
steel floor model are shown in Figure 3 and 4.
Figure 3. Schematic diagram of the test setup
Displacement sensors: LV9804, LV9648;
Acceleration Sensors: A2271, A2267, A2272
Figure 4. View of the test facility for vibration
measurement
Using WinSTS software can run on the window PC to measure and predict the first natural
frequency of steel model at the sample rate of a 50Hz signal shown in Figure 5. The curves of Figure
5a, b corresponding to acceleration sensors, and Figure 5d, e of displacement sensors. The frequency-
response curves shown in Figure 5c, f, g, are determined by FFT analysis, and displayed the amplitude
of vibration and the fundamental frequency about 3.512 Hz. Figure 6 shows the experimental data at
the sample frequency of 291.511 Hz, and also can estimate the natural frequency of steel model.
However, the main objectives of the study has been done to investigate vibration of steel
model which will be measured by low-cost MEMS (MPU6050). In order to solving this problem,
experimental determination of characteristic frequencies can use microcontroller boards (Arduino
Mega 2560 and Raspberry Pi 3) to connect with MPU6050 via the SDA and SCL pins. Interfacing
Arduino Mega 2560 and MPU6050 can be programmed to run open-source code libraries supplied by
Arduino Software (IDE) [13, 14], and receive acceleration signal through Matlab software using
“serial” function. Raspberry Pi 3 is single-board computer which is installed Raspberry Pi Desktop
(the Foundation’s operating system) with many open-source software such as: Python, Java, C++ [16];
also can be connected with MPU6050 by the SDA and SCL pins. Using Python with Smbus, NumPy,
Matplotlib, SciPy modules [17, 18, 19, 20] on Linux of Raspberry Pi 3 can access through the I2C/dev
to interface MPU6050.
LVDT
Motor
Accelerometer
MPU6050

Nguyen Cong Duc, Huynh Quoc Hung, Phan Cong Ban, Tran Van Mot, Nguyen Cong Minh,
Pham Bao Toan and Ngo Kieu Nhi
4
Figure 5. Vibration and displacement responses of steel model with the sample frequency of 50Hz
Figure 6. Dynamic responses of steel frame at the sample rate of 291.511Hz
The frequency-response curves are shown in Figure 7b, d corresponding to commercial MEMS
of A2267 (BDI-USA), MPU6050 using Arduino Mega 2560 with the sample rating of 291.511Hz. The
results presented in Figure 7 can determine the fundermental frequency which are related to maximum
peak of modal amplitude, and other frequencies are shown in this Figure. Figure 7a,c illustrate
acceleration signals of model, amplitudes from 1400 mm/s2 to 2000 mm/s2 . WinSTS (BDI-USA) can
control accelerometers to obtain data at different sampling frequencies while MPU6050 in Arduino
has limitations to remain constant sample rating, and WinSTS must also change the same frequency
with Arduino Mega for predicting first modal shape and natural frequency. This problem can be
solved through data processing technique of embedded codes in Arduino Mega board.
a)
b)
d)
c)
f)
e)
g)
a)
b)
d)
c)
f)
e)
g)

Low-Cost Vibration Measurement for Behavior of Small-Scale Steel Modeling using MEMS,
Raspberry Pi-3 and Arduino Mega 2560
5
Figure 7. Vibration responses of Acceleration (A2267, BDI-USA) and MPU6050 (±2g) on
Arduino Mega 2560 at the sample rate of 291.511Hz
In the Figure 8a, c, Dynamic responses of steel frame are measured by Raspberry Pi 3 and
MPU6050 (±2g) with the sample frequency of this experiment under 66.208 Hz, compared to MEMS
signal of BDI-USA (±5g). The first mode shape of steel modeling is predicted by FFT analysis with
the frequency range of 0.001 Hz to 33 Hz shown in Figure 8b, d at peak-frequency of 3.621 Hz
(A2267) and 3.587 Hz (MPU6050). There are small noise peaks in spectral analysis of signals, partly
because there are limitations in fixed supports and manufacturing of steel frame. Based on open-
source codes of Python software including: Smbus [20] designed to read signal from Raspberry Pi 3,
Numpy module [19] in FFT analysis, Matplotlib [18] for plotting signal in time.
Figure 8. Acceleration (A2267, BDI-USA) and MPU6050 (±2g) on Raspberry Pi3 at the sample
rating of 66.208Hz
a)
b)
c)
d)
a)
b)
c)
d)

Nguyen Cong Duc, Huynh Quoc Hung, Phan Cong Ban, Tran Van Mot, Nguyen Cong Minh,
Pham Bao Toan and Ngo Kieu Nhi
6
Table 1. First modal shape of steel frame in amplitude-frequency curves
Sample Rating
MPU6050 (±2g)
Arduino Mega 2560
MPU6050 (±2g)
Raspberry Pi 3
A2267 (±5g)
BDI, USA
50 Hz
-
-
3.512 Hz
66.208 Hz
-
3.587 Hz
3.621 Hz
291.511Hz
3.547 Hz
-
3.558 Hz
Figure 9. FEM modeling of scaled steel floor
using BEAM4 and SHELL63 in ANSYS
Figure 10. Using 3D model of steel frame in
ABAQUS
Moreover, In the theoretical analysis, it is possible to calculate modal shapes of scaled steel
floor for predicting natural frequency of structure through finite element method in ABAQUS and
ANSYS software, as shown in Figure 9 and 10. With FEM tools can support for vibration analysis in
test field and simulation which can also calibrated and updated the model's parameters in the future
study.
3. Conclusions
The purpose of the present work was to study vibration measurement for behavior of steel
specimen based on low-cost MEMS, Raspberry Pi-3 and Arduino Mega 2560. The scaled 3-story steel
frame was simulated based on first modal shape, and experimentally analyzed under periodic force.
The following principal conclusions are drawn from the results that obtained in this study:
- Using Raspberry Pi-3 and Arduino Mega 2560 can connect with various types of low-cost
accelerometer via I2C communication protocol of boards for experimental modal analysis, and
embedded open-source codes.
- The present method of measuring structural vibration via MPU6050 is efficient, fast, accurate
and cheap, can predicts well the fundamental frequency in vibration analysis of this steel frame and
other structures. The apparatus is simple to operate and is particularly recommended for extensive
studies of vibration measurement.

Low-Cost Vibration Measurement for Behavior of Small-Scale Steel Modeling using MEMS,
Raspberry Pi-3 and Arduino Mega 2560
7
Acknowledgments
The authors would like to thank the Mien Trung University of Civil Engineering (MUCE, LAS-
XD 162) in Phu Yen province, Ministry of Construction of the Socialist Republic of Vietnam (MOC)
for equipment funding in this research project.
References
[1] Ge-Wei Chen, Sherif Beskhyroun and Piotr Omenzetter. Experimental investigation into amplitude-
dependent modal properties of an eleven-span motorway bridge. Engineering Structures, vol. 107, (2016), pp.
80-100
[2] Ge-Wei Chen, Sherif Beskhyroun and Piotr Omenzetter. Operational modal analysis of an eleven-span
concrete bridge subjected to weak ambient excitations. Engineering Structures, vol. 151, (2017), pp. 839-860
[3] Kosmas Dragos and Kay Smarsly. Decentralized Infrastructure Health Monitoring Using Embedded
Computing in Wireless Sensor Networks. Dynamic Response of Infrastructure to Environmentally Induced
Loads, Lecture Notes in Civil Engineering, vol 2, (2017), pp 183-201
[4] A. Feldbusch, H. Sadegh-Azar and P. Agne. Vibration analysis using mobile devices (smartphones or
tablets). X International Conference on Structural Dynamics – EURODYN, Procedia Engineering, vol. 199,
(2017), pp. 2790-2795
[5] Tat S. Fu1, Amitabha Ghosh, Erik A. Johnson and Bhaskar Krishnamachari. Energy-efficient deployment
strategies in structural health monitoring using wireless sensor networks. Structural Control and Health
Monitoring, vol. 20, (2012), pp. 971–986
[6] David Hester, James Brownjohn, Mateusz Bocian and Yan Xu. Low cost bridge load test: Calculating bridge
displacement from acceleration for load assessment calculations. Engineering Structures, vol. 143, (2017), pp.
358–374
[7] Thanh-Canh Huynh, Jae-Hyung Park and Jeong-Tae Kim. Structural Identification of Cable-Stayed Bridge
under Back-to-Back Typhoons by Wireless Vibration Monitoring. Measurement, vol. 88, (2016), pp. 385-401
[8] Jerome P. Lynch; Aaron Partridge; Kincho H. Law; Thomas W. Kenny; Anne S. Kiremidjian and Ed
Carryer. Design of Piezoresistive MEMS-Based Accelerometer for Integration with Wireless Sensing Unit for
Structural Monitoring. Journal of Aerospace Engineering, vol. 16, ( 2003), pp. 108-114
[9] Guido Morgenthal and Hagen Höpfner. The application of smartphones to measuring transient structural
displacements. Journal of Civil Structural Health Monitoring, vol. 2, Issue 3–4 (2012), pp. 149–161
[10] Guido Morgenthal, Sebastian Rau, Jakob Taraben and Tajammal Abbas. Determination of Stay-Cable
Forces Using Highly Mobile Vibration Measurement Devices. Journal of Bridge Engineering, Vol. 23, Issue 2
(2018)
[11] Julie M. Samuels, Michael Reyer, Stefan Hurlebaus, Stephen H. Lucy, David G Woodcock and Joseph M.
Bracci. Wireless sensor network to monitor an historic structure under rehabilitation. Journal of Civil Structural
Health Monitoring, vol.1, issue 3–4, (2011), pp. 69–78
[12] Billie F. Spencer Jr., Hongki Jo, Kirill A. Mechitov, Jian Li, Sung-Han Sim, Robin E. Kim1, Soojin Cho,
Lauren E. Linderman, Parya Moinzadeh, Ryan K. Giles and Gul Agha. Recent advances in wireless smart
sensors for multi-scale monitoring and control of civil infrastructure. Journal of Civil Structural Health
Monitoring, vol. 6, Issue 1, (2016), pp. 17–41
[13] https://www.arduino.cc
[14] https://github.com/jrowberg/i2cdevlib
[15] https://www.mathworks.com
[16] https://www.raspberrypi.org
[17] https://www.python.org
[18] https://matplotlib.org
[19] http://www.numpy.org
[20] https://pypi.org/project/smbus-cffi
[21] https://github.com/pyserial/pyserial