A Proposal of a Neural Predictor of Residual Compressive Strength in an SCC Exposed to High Temperatures for Resilient Housing

This paper reports a proposal of a neural predictor, which could usefully improve performance in predicting the compressive strength of SCC, when used for innovative housing. The predictor is intended for its ability to provide accurate predictions on structures where verification against high temperatures should be required, thus offering an interesting possibility for resilient housing systems.In detail, predictions of the mechanical properties of Self-Compacting Concretes (SCC) after exposure to high temperatures will be obtained by designing a neural predictor for residual compressive strengths of this type of concrete for resilient housing. All considered data refer to SCC samples heated up to 800°C and subjected to standard compression tests.In this work, the proposed neural predictor will be related to temperature and geometric ratio of each SCC sample.An interesting behavior in capturing the complex relationships between input variables and residual compressive strengths of SCCs will be shown by the proposed neural network. Results in terms of residual compressive strengths are reported.