Suitability testing of micro-electro-mechanical systems (MEMS) for integrating into a geosensor network for the automated monitoring of oscillating structures

Suitability testing of micro-electro-mechanical systems (MEMS) for integrating into a geosensor network for the automated monitoring of oscillating structures

Looking after:  Andreas Becker, Mohammad Omidalizarandi, Jürgen Rüffer, Ingo Neumann
Researcher:  Eva Kemkes
Year:  2017
Lifespan:  2016
Is Finished:  yes

There has always been a risk of collapses of structures (e.g. bridges, buildings and so on) by enormous forces and external influences such as winds, movements of vehicles in case of bridges and traffic load and so on. Therefore, structures are needed to be monitored continuously to avoid any failures such as cracks. Today, low cost deformation analysis is of utmost interest and subsequently, low cost micro-electro-mechanical sensors (MEMS) sensor as a cost effective and alternative sensor to the classical geodetic measurements can be used. MEMS sensor can consists of multiple types of sensors, e.g. accelerometer, gyroscope, temperature and pressure sensors. The output of accelerometers and gyroscope are accelerations and rotation rates in all three directions respectively. Furthermore, accelerations and rotations rates can be integrated over time to determine relative movements and orientations. The aim of this research is to perform suitability test of low cost MEMS accelerometers for integrating into a geosensor network by mounting at different positions of bridge structures for an automated monitoring. The choice of sensors is based on suitable properties such as, e.g., accuracy, stability, costs. The price of used sensors in this work is up to approximately 60 Euros. Concerning suitability test, static calibration is performed for three low cost accelerometers (i.e. ADXL (ADXL345 chip from Analog Devices company with Arduino UNO Board), NAMS (BNO055 chip with Arduino UNO board and Nine-Axis-Motion-Shield) and IMU (BNO055-Chip Tinkerforge-Board)) based on well known six positions static calibration test, here measurement period of about 15-20 minutes at fourteen positions, to determine biases, scales and non-orthogonalities in all three directions with applied Gauss Helmert model. Variance component estimation is carried out to assign optimal weights to the observations. To this end, deterministic errors of the accelerometer are defined based on calibration with optimal estimation of standard deviations of the observations. From standard deviations of accelerations in all three directions, it is observed that z axis has a less accuracy comparing to the other axes and should be taken into the account to set it up towards less deformable direction of the bridge, i.e. transverse direction. In addition, measurements are also conducted in the climate chambers to observe effect of temperature variations which shows strong influence on the measurements and tendency of increasing noises up to maximum 50%. Moreover, long-term measurement over 24 hours is carried out to observe noise behavior of the sensors under static conditions and to prove stability and correctness of the computed parameters. At the end, test sensors measurements are compared with a highly accurate accelerometer as a reference sensor in the frequency domain. Results shows 96% or better equivalence of the test sensor characteristics to the reference sensor.