Abstract:
Detection of structural deterioration in hybrid building constructions is considered in the present work. The constructions belong to semi-rigid or ‘bending-active’ syste...Show MoreMetadata
Abstract:
Detection of structural deterioration in hybrid building constructions is considered in the present work. The constructions belong to semi-rigid or ‘bending-active’ systems. They consist of fiberglass rods, struts, cables and membrane. Hybrid constructions have found use in roofs and facades of permanent buildings and for temporary canopies as well.Detection of structural deterioration includes developing machine learning models in form of artificial neural networks. Appropriate structures of the networks and parameters of the hybrid buildings are given for two types of deterioration cases: material aging and slackening, as well as the membrane tearing.Semi-supervised learning strategy is used for training the networks. Discrepancy between the vector of structural parameters and the resultant vector, generated by the network, is considered the indicator for subsuming the hybrid construction under normal or anomalous (deteriorated) classes.The work contributes to implementation of machine learning techniques for research and development of hybrid building constructions. It facilitates automated structural health monitoring of complex modern buildings.
Published in: 2021 3rd International Conference on Control Systems, Mathematical Modeling, Automation and Energy Efficiency (SUMMA)
Date of Conference: 10-12 November 2021
Date Added to IEEE Xplore: 10 December 2021
ISBN Information: