Mariusz Ostrowski from the Department of Intelligent Technologies IPPT PAN won sixth place in the prestigious international project competition on civil infrastructure condition assessment, The 2nd International Competition for Structural Health Monitoring - IC-SHM, 2021.
The competition was organized by international universities, i.e. the University of Illinois at Urbana-Champaign, USA, and Harbin Institute of Technology, China. 167 teams from 20 countries applied to participate. In the end, 53 projects from 12 countries were submitted.
Participants were offered a choice of three categories for their projects on computer vision-based structural health monitoring of civil infrastructure. The first two ones consisted in computer vision-based assessment of concrete structures with the aid of machine learning, e.g. artificial neural networks. The third category consisted in computer vision-based measurement of truss structure vibrations and extraction of information about the location and degree of structural damage.
Mariusz Ostrowski prepared his project in the third category. He developed software in the Matlab® environment enabling computer vision-based measurement of structural vibration and assessment of the technical condition of the truss structure. He tested various methods for computer vision-based object tracking. Finally, he chose a method based on normalized cross-correlation. Ostrowski also investigated several methods of detection, location and estimation of the degree of damage in order to assess the technical condition of the monitored structure. He chose methods employing the mathematical model of the structure, because model-free methods (based solely on measurement) turned out to be too inaccurate due to the significant noise content in the computer vision-based measurement, i.e. from 5 to even 60% in this case. Ostrowski developed a computer model of the truss structure using the finite element method (FEM). Then, the model was updated by changing the stiffness parameters of the monitored members of the truss structure so that it reflected the behavior of the potentially damaged monitored truss structure. Elements of the model with reduced stiffness parameters were considered damaged. Ostrowski finally selected the augmented inverse estimate method that minimized the difference between the measured modal data and the ones obtained from the model. The selected method is based on the pseudoinverse of the extended sensitivity matrix of modal data to the changes in stiffness parameters. The method preliminarily rejects measurements with the significant noise content. It also uses the regularization of the solution with the aid of the truncated singular value decomposition (TSVD). The selected technique solves the problem by repeatedly removing in subsequent iterations the unknowns from the system of equations corresponding to the truss elements classified undamaged. This improves regularization and numerical conditioning. The method described above proved to be more effective than other proven model updating techniques – including the Bayesian methods which are popular nowadays. The framework proposed by Ostrowski allowed for automated assessment of the technical condition of the truss structures, having at its disposal only filmed vibrations and basic construction data from technical documentation, i.d. dimensions and materials used.
Mariusz Ostrowski is a fifth-year PhD student at the IPPT PAN. Here is how he describes himself: "I have been interested in science since my youth. I have always been fascinated by physics in its broad sense, and the practical use of mathematics. This was the reason why I decided to study at the Lublin University of Technology (2011-2016). Having graduated from my first degree studies in mechatronics, I started second-cycle studies in the field of Electrical Engineering, specializing in the control of electric drives.
My professional work reflected my passion for science. As a student, I participated in a project concerning the electric drive of go-karts, which had been carried out by Findewal Sp. z o. o. Later, I worked as an opto-mechatronics specialist at Astri Polska Sp. z o. o. on the UV2VIS project developing a measuring instrument for the verification of a spectrometer prototype, which would ultimately be launched into space on a satellite (Sentinel 5 mission). The project was funded by the European Commission and the European Space Agency.
Later, I researched simulations of electromechanical systems in the Matlab® environment within a project financed by the Polish National Center for Research and Development (NCBR), implemented at Findewal Sp. z o. o. In 2017, I decided to start my doctorate at IPPT PAN, choosing a research that would combine applied mechanics and control theory."
As part of his doctorate at IPPT PAN under the supervision of Associate Professor Bartłomiej Błachowski, PhD, DSc., Mariusz Ostrowski develops a strategy of semi-active modal control. The control is implemented by dynamically blocking rotational connections in frame structures. This makes it possible to induce the effect of modal coupling – and consequently – the directed transfer of vibration energy to or from selected vibration modes. The developed control strategy is used in vibration damping and in vibration-based energy harvesting.
In the case of vibration damping, energy is transferred from lower-order vibration modes, which usually are weakly damped, to higher-order vibration modes characterized by strong material damping, in which energy is then dissipated. This results in effective damping of structural vibration while maintaining the advantages of the semi-active control – no need to use an efficient power source, low energy consumption, and the inability to destabilize the structure by the control. The developed control strategy is applicable in frame structures, in particular those ones that have low stiffness due to weight requirements, e.g. space construction modules.
Regarding energy harvesting, the proposed control is able to transfer energy to a selected vibration mode that is tuned to an energy harvesting device, significantly improving its work. In this case, the controlled structure equipped with lockable rotational connections operates as a small adaptive energy buffer between the vibration source and the energy harvesting device.
The subject matter of the research is interdisciplinary. In addition to control theory, Ostrowski uses knowledge in the field of mathematical and computational modeling of the structural dynamics and proprietary methods of optimal placement of actuators and sensors. Proprietary programs and FEM computer models are also used for his research. The control strategy under study is beginning its experimental validation in the laboratory.
