|M.Sc Student||Kattel Uri|
|Subject||Semantic Enrichment of Geometry Representation Models|
of Highway Bridges
|Department||Department of Civil and Environmental Engineering||Supervisor||Professor Rafael Sacks|
|Full Thesis text - in Hebrew|
Highway bridges are important structures. We are dependent on their proper functioning for our daily commutes, especially in developed countries with complex highway networks and large numbers of bridges. The aging population of bridges in many countries, together with the limited budgets invested in inspections and maintenance of the bridges, highlights the need to find new and more efficient ways to gather information and allocate the limited maintenance budgets. In the last years, several studies have been conducted in the field of implementing building information modeling (BIM) tools in different facility management applications. Those studies found many benefits of such an implementation. The existing manual method of creating BIM models requires many man hours of modeling experts, which casts doubt on its economic viability in an environment where bridge networks include many bridges that lack a BIM model. To profit from use of BIM tools in bridge management systems we have to automate the process of BIM model creation. Such a process has been investigated in some current studies and is often called "Scan to BIM". The research described in this paper was conducted as a part of a large-scale study named SEEBridge (Semantic Enrichment Engine for Bridges). SEEBridge itself was part of the Infravation initiative of the European Union. The research was aimed to develop a scan to BIM process composed of four stages: The first stage is Acquisition of geometrical information (creation of the point cloud) with help of laser scanning, photogrammetry or videogrammetry. The Second stage is creating a geometric representation model according to the acquired data. The third stage is Enriching the geometry model into a useful BIM model. The last, fourth, stage is Mapping the findings of damage to the bridge on the enriched model. The research described here concentrates on the third stage. In this research we define semantic enrichment as the addition of meaningful information to the geometry model using rule inferencing. Our inputs were geometry representation models and our outputs were BIM models that could be used for bridge inspection and maintenance. The research method was design science. The success in enriching geometry models proves our ability to develop such a process, but it does not prove that the process we have developed is the only possible process or even the most efficient one. In the research we interviewed several experts in bridge design and bridge inspection domains. After processing the data gained, we defined what must be added to the geometry model to make it useful. We developed a rigorous method to compile rules for object identification, which we call “the Matrix Method”. We have developed algorithms for numbering objects, axes reconstruction, assignment to function and placement groups, and we have developed a special algorithm that reconstructs the geometry of objects partly occluded during the scan process. We have successfully validated the developed algorithms on models created by manual and automatic tools. This validation proves the ability to enrich a bridge model by means of rule inferencing.