Prestressed concrete bridges are susceptible to deterioration over time which might significantly affect their capacity and overall performance. In previous decades, infrastructure owners have found that continuous monitoring of these assets is a valuable tool for their management as it facilitates the decision-making process regarding the intervention strategies required. However, as data acquisition and measurement technologies have advanced tremendously in recent years, the amount of information that can be retrieved daily is not easy to manage and analyse. This study presents an evaluation of the effectiveness between different machine learning methods regarding prediction and interpretation of structural responses as well as the feasibility of mapping an independent variable, aspects such as metric performance, learning curves and residual plots was analysed. A comparison was made on the machine learning algorithms performing regression analysis where each model scored over 98% in the R-square metric. This study utilised data collected from a prestressed concrete bridge located in Autio, northern Sweden, that has been continuously monitored for more than three years.
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:umu-214572 |
| Date | January 2023 |
| Creators | Coric, Vedad |
| Publisher | Umeå universitet, Institutionen för datavetenskap |
| Source Sets | DiVA Archive at Upsalla University |
| Language | English |
| Detected Language | English |
| Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
| Relation | UMNAD ; 1441 |
Page generated in 0.0014 seconds