Abstract
Abstract
With the depletion of natural resources and the requirement of higher strength-weight ratio, lightweight aggregate concrete has attracted more and more attention because of its good thermal properties, fire resistance and seismic performance. However, exposure to low temperature environments accelerates deterioration of concrete, thereby, reduce the service life of lightweight aggregate concrete. Even worse, in cold and arid regions, lightweight aggregate concrete often experiences accidental impacts, wind erosion, earthquakes, and other disasters during service, these damage significantly impact its frost-resistance. Therefore, accurately and quantitatively describing and predicting the frost-resistance of lightweight aggregate concrete under specific disaster conditions is crucial. In this study, take the initial damage degree and freeze-thaw cycles as input variables, while the relative dynamic elastic modulus (RDEM) as an out variable, a frost resistance prediction models for stress-damaged lightweight aggregate concrete was established based on back propagation neural network (BPNN). The results show that the predicted values of BPNN model are in good agreement with the experimental values, and the results are also compared with the revised Loland model which is proposed by another author. Results demonstrate that the average relative error between predicted values of BPNN and experimental values is only 1.69%, whereas the one of revised Loland model is 9.13%, which indicating that the proposed BPNN prediction model can achieve a relatively accurate quantitative assessment of frost-resistance throughout the entire post-disaster lifecycle of lightweight aggregate concrete, it also broadened the idea and provided a reference for the frost resistance prediction of stress-damaged lightweight aggregate concrete.
Funder
International Science and Technology Cooperation Program of Science and Technology
Basic Scientific Research Project of Education Department of Liaoning Province