Structural Health Monitoring of Infrastructure elements with unmeasured inputs
Lead Research Organisation:
University of Oxford
Department Name: Engineering Science
Abstract
Estimating the properties of infrastructure elements is of paramount importance for their proper management and maintenance. This can be achieved through Structural Health Monitoring (SHM), i.e., placing sensors on the structures to infer their structural state.
An assumption commonly employed in such frameworks is that the excitations applied to the system can either be measured, or their stochastic properties are known. Railway bridges are a prominent example where such assumptions may not always be feasible. Additionally, the non-linearities in such structures, e.g., emanating from elements such as bearings, pose challenges. Yet, it is still important that such elements can be accurately monitored due to their importance.
This project aims to extend the SHM framework to cover the case of structures with unmeasured excitations allowing for their efficient monitoring under the presence of strong non-linearities.
To complete the project the applicant Manolis Chatzis, MC, has planned a visit to the Structural Mechanics Section of KU Leven in Belgium. The hosts of this visit will be Prof. Geert Lombaert, GL and Dr. Kristof Maes, KM. The Structural Mechanics Section has a long tradition in the monitoring of infrastructure elements. GL has multiple contributions in the directions explored in this project, and together with KM have recently suggested a novel SID algorithm for systems with unmeasured inputs. KM, MC and GL recently presented one of the first observability algorithms for systems with unmeasured inputs.
This project builds further on those works. It extends the concept to discrete systems aiming at improving system identification algorithms. The improved algorithms will be used to develop sensor placement strategies for systems with unmeasured inputs.
To further motivate the study, the group will demonstrate the framework for the case of an instrumented railway bridge: the KW51 railway bridge involving bearings, which has been monitored by the Structural Mechanics Section of KU Leuven since October of 2018. The bearings will be added to the model of the bridge and their effect will be accounted for in the identification framework. The Oxford Dynamics Lab accelerometer units will be added to the existing sensorial network resulting in one of the most well-instrumented bridge monitoring campaigns. The tools developed in the project will be demonstrated on this challenging application.
Additionally, the dense network used creates ideal conditions for validating the framework and the opportunity of generating a benchmark case for the SHM community.
The suggested visit extends an on-going collaboration between the visitor and the host academics and suggests the visitor spending time at a centre of excellence in SHM.
An assumption commonly employed in such frameworks is that the excitations applied to the system can either be measured, or their stochastic properties are known. Railway bridges are a prominent example where such assumptions may not always be feasible. Additionally, the non-linearities in such structures, e.g., emanating from elements such as bearings, pose challenges. Yet, it is still important that such elements can be accurately monitored due to their importance.
This project aims to extend the SHM framework to cover the case of structures with unmeasured excitations allowing for their efficient monitoring under the presence of strong non-linearities.
To complete the project the applicant Manolis Chatzis, MC, has planned a visit to the Structural Mechanics Section of KU Leven in Belgium. The hosts of this visit will be Prof. Geert Lombaert, GL and Dr. Kristof Maes, KM. The Structural Mechanics Section has a long tradition in the monitoring of infrastructure elements. GL has multiple contributions in the directions explored in this project, and together with KM have recently suggested a novel SID algorithm for systems with unmeasured inputs. KM, MC and GL recently presented one of the first observability algorithms for systems with unmeasured inputs.
This project builds further on those works. It extends the concept to discrete systems aiming at improving system identification algorithms. The improved algorithms will be used to develop sensor placement strategies for systems with unmeasured inputs.
To further motivate the study, the group will demonstrate the framework for the case of an instrumented railway bridge: the KW51 railway bridge involving bearings, which has been monitored by the Structural Mechanics Section of KU Leuven since October of 2018. The bearings will be added to the model of the bridge and their effect will be accounted for in the identification framework. The Oxford Dynamics Lab accelerometer units will be added to the existing sensorial network resulting in one of the most well-instrumented bridge monitoring campaigns. The tools developed in the project will be demonstrated on this challenging application.
Additionally, the dense network used creates ideal conditions for validating the framework and the opportunity of generating a benchmark case for the SHM community.
The suggested visit extends an on-going collaboration between the visitor and the host academics and suggests the visitor spending time at a centre of excellence in SHM.
| Description | During the visit the group defined the concept of observability of discrete systems with unmeasured inputs. This is a new theoretical tool for understanding of whether a system monitored with a system of sensors can be properly identified at the presence of unmeasured excitations. A conference presentation has already been produced on the topic and a journal paper will shortly be submitted. The group used this algorithm to understand the properties of convergence of existing input estimation kalman filters and developed two new filters based on that suggestion. Another conference presentation has already been produced on the topic and a journal paper will shortly be submitted. The group also gathered data from the KW51 bridge, the results have been uploaded to a permanent link. Details can be found in: https://eng.ox.ac.uk/non-lineardynamics/kw51-monitoring-epw0010981/ A fourth year project was dedicated on analyzing the data and a publication will follow. Other outcomes: The network of sensors for field measuring has now been completed to have 3 more GPS sensors and autonomous wifi infrastructure. |
| Exploitation Route | The findings of the work on observability and system identification will be submitted as journal papers and are expected to influence authors on the effect of discrete measurements and the lagh between measuring and being able to estimate a past input. The KW51 data will be used as benchmark data for demonstrating SHM algorithms by authors. |
| Sectors | Construction,Transport |
| URL | https://eng.ox.ac.uk/non-lineardynamics/kw51-monitoring-epw0010981/ |
| Description | Collaboration with Prof. Geert Lombaert and Dr. Kristof Maes |
| Organisation | University of Leuven |
| Country | Belgium |
| Sector | Academic/University |
| PI Contribution | This was an Overseas Travel grant so my contribution was: -moving to Leuven from March to September -bring the 7 triaxial Oxford accelerometers with me of value 50,000 £ -Working as a Visiting Researcher for 7 months -Giving a presentation to KUL |
| Collaborator Contribution | My partners provided: -Their research time (for both Prof. Lombaert and Dr Maes) and time dedicated to one meeting per week over a period of 7 months. -Their 13 triaxial accelerometers of value 91,000 Euros and other sensors -Time for their lab to set up sensors -Access to the KW51 bridge -Visit to other infrastructure projects |
| Impact | Presentations: M. N. Chatzis, K. Maes, G. Lombaert, 'An Observability Inspired Joint State, Parameter and Input Estimation Extended Kalman Filter', Engineering Mechanics Institute Conference 2022, JHU, Baltimore Maryland M. N. Chatzis, K. Maes, G. Lombaert, 'On the Observability of Discrete Systems with Unmeasured Input', Engineering Mechanics Institute Conference 2022, JHU, Baltimore Maryland |
| Start Year | 2021 |