Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Prediction of the elastic modulus of recycled aggregate concrete applying hybrid artificial intelligence and machine learning algorithms
Recycled aggregates (RAGs) usage in concrete is surging, inspired by environmental and economic concerns. Regarding predicting various models designed the values of modulus of elasticity (MOE) of concrete with natural aggregates and, in conclusion, they would probably be unreliable when used to concrete with RAG. In the present study, two new gray wolf multi‐layer perceptron neural networks (GWMLP) and gray wolf support vector regression (GWSVR) algorithms were proposed to predict RAG concrete's elastic modulus. About 400 records were gathered from published articles to develop these models. The results show that among the GWMLP models with different hidden layers, GWM3L with three hidden layers could get the highest score (TRS) at 39. Simultaneously, in the testing phase, the GWSVR was the first‐rank model because of the lower RMSE (0.6381), MAE (0.1541), and a larger (0.9707) compared with GWMLP models. Therefore, it can result that the GWSVR model could predict the elastic modulus of RAG concrete precisely even better than GWM3L, which is well over the accuracy of the developed models.
Prediction of the elastic modulus of recycled aggregate concrete applying hybrid artificial intelligence and machine learning algorithms
Recycled aggregates (RAGs) usage in concrete is surging, inspired by environmental and economic concerns. Regarding predicting various models designed the values of modulus of elasticity (MOE) of concrete with natural aggregates and, in conclusion, they would probably be unreliable when used to concrete with RAG. In the present study, two new gray wolf multi‐layer perceptron neural networks (GWMLP) and gray wolf support vector regression (GWSVR) algorithms were proposed to predict RAG concrete's elastic modulus. About 400 records were gathered from published articles to develop these models. The results show that among the GWMLP models with different hidden layers, GWM3L with three hidden layers could get the highest score (TRS) at 39. Simultaneously, in the testing phase, the GWSVR was the first‐rank model because of the lower RMSE (0.6381), MAE (0.1541), and a larger (0.9707) compared with GWMLP models. Therefore, it can result that the GWSVR model could predict the elastic modulus of RAG concrete precisely even better than GWM3L, which is well over the accuracy of the developed models.
Prediction of the elastic modulus of recycled aggregate concrete applying hybrid artificial intelligence and machine learning algorithms
Zhang, Qian (Autor:in) / Afzal, Mansour (Autor:in)
Structural Concrete ; 23 ; 2477-2495
01.08.2022
19 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Using artificial neural networks for predicting the elastic modulus of recycled aggregate concrete
| Online Contents | 2013
Using artificial neural networks for predicting the elastic modulus of recycled aggregate concrete
| British Library Online Contents | 2013
| DOAJ | 2023