This thesis presents laser scanning as a tool for changing the way in which monitoring of infrastructures is performed, more specifically transport infrastructure. LiDAR is a non-destructive technology able to collect accurate data about the environment where the survey is taking place. With this, the external geometrical configuration of a needed asset can be easily obtained. Information extracted using laser scanners, the so-called point clouds, is scattered in the three-dimensional (3D) space. In order to give it context, some processing algorithms need to be applied, obtaining meaningful and human readable information from them.
The approach of this thesis is to develop methodologies with the aim of inventory, monitor, and classify point cloud data of specific assets in the transport infrastructure. An especial interest has been put on masonry constructions and railway environments. Firstly, due to the functional dependency that the terrestrial transport network has on both of them. And secondly, since the location of the actual transport network was motivated by previous infrastructures constructed across the centuries and some of those ageing infrastructures are still standing. What seemed logical was to propose strategies for the automatic monitoring of those infrastructures without the need of human intervention. This standpoint was followed through the entire life of the thesis and it fitted with the definition of predictive maintenance, and more specifically predetermined maintenance, which is to be performed at scheduled times. This proved to be the most efficient way to detect problems in assets, carrying out a study of their condition before it is required [1].
This doctoral dissertation addresses the monitoring of masonry constructions using non-destructive techniques, as laser scanning systems are, and different data processing strategies. Besides that, the same is performed for tunnels in the railway network. Notionally, all the information obtained from point clouds would be stored in a digital model (i.e. digital twin), which by definition should be updated through time in order to enable the application of predictive maintenance of the stored assets. The main challenges of the research can be presented in three groups, being: (i) LiDAR point clouds as Big Data structures, containing millions of points that need to be stored and processed; (ii) small processing units, as the normal devices in the hands of final users; and (iii) storage capacity, as one the biggest challenges to be tackled by modern society.
The workflows and methods presented have been validated using different datasets, obtaining results contributing to the state of the art. This helped with the publication of peer reviewed scientific articles, which are to be consulted in three different scientific journals, indexed on the Journal Citation Report (JCR), one book chapter and two conference papers.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados