Applications of artificial intelligence in bridge engineering

No Thumbnail Available
Journal Title
Journal ISSN
Volume Title
Insinööritieteiden korkeakoulu | Master's thesis
Date
2019-03-11
Department
Major/Subject
Mcode
Degree programme
Master's Programme in Building Technology (CIV)
Language
en
Pages
48+6
Series
Abstract
The term artificial intelligence is related to computer engineering, but since 1950s the applications of AI have been done in the field of bridge engineering. Machine learning is one of the branches in artificial intelligence, which allows the new way of access to the hidden structure in data. Initially, in this thesis, a basic knowledge of AI and machine learning is studied. Secondly, some of the applications done by neural networks, genetic algorithms and expert systems are reviewed. This thesis investigates the possibilities of using AI to enhance the bridge design and maintenance which could help the designer in their tedious process. A survey was conducted with the bridge engineering professionals to find the areas in the design process where AI could help to make the designer’s process less tedious. For the thesis, two simple case studies were demonstrated by utilizing machine learning techniques, imputation and symbol recognition. Methods were applied against a simple bridge related dataset. The imputation techniques were used to impute the missing parameters in the bridge dataset though the results seemed to be biased, the work was made as a general demonstration application rather than for real use case. A convolutional neural network model was developed to recognize the presence of a coordinate arrow symbol from the noisy image, model has obtained an accuracy of 99.8% was obtained. The thesis concluded that in today’s world, AI has not been still used in bridge design. According to interviews, there could be multiple possibilities of AI to enhance the routine tasks and error checks. On basis of case studies at least symbol recognition has the real potential in present technologies. In future, symbol recognition can be extended to teach old drawings to machines which are in the design office.
Description
Supervisor
Singh, Vishal
Thesis advisor
Hartikainen, Ari
Keywords
artificial intelligence, machine learning, convolutional neural network, bridge engineering, imputation
Other note
Citation