Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Artificial neural networks and support vector regression for predicting slump and compressive strength of PET-modified concrete
Laboratory experiments for estimating concrete properties can be costly and time-consuming. Alternatively, predictive models based on artificial intelligence (AI) methodologies offer a viable solution. This paper presents predictive modeling employing artificial neural networks (ANNs) and support vector regression (SVR) to forecast two critical properties, slump, and compressive strength, of concrete incorporating plastic waste as fine aggregate, with a focus on PET material. Over 100 data points from literature were carefully selected to train these models, considering ten input variables including the percentage of PET content (PET_%), water-cement ratio(w/c), minimum size of PET (P_min), maximum size of PET (P_max), minimum size of sand (S_min), maximum size of sand (S_max), minimum size of gravel (G_min), maximum size of gravel (G_max), cement (C) and superplasticizer (PS). The results indicated that SVR outperforms ANN in accuracy for predicting these properties. Additionally, the study acknowledges limitations and points to avenues for further research to enhance predictive modeling’s applicability in sustainable concrete design.
Artificial neural networks and support vector regression for predicting slump and compressive strength of PET-modified concrete
Laboratory experiments for estimating concrete properties can be costly and time-consuming. Alternatively, predictive models based on artificial intelligence (AI) methodologies offer a viable solution. This paper presents predictive modeling employing artificial neural networks (ANNs) and support vector regression (SVR) to forecast two critical properties, slump, and compressive strength, of concrete incorporating plastic waste as fine aggregate, with a focus on PET material. Over 100 data points from literature were carefully selected to train these models, considering ten input variables including the percentage of PET content (PET_%), water-cement ratio(w/c), minimum size of PET (P_min), maximum size of PET (P_max), minimum size of sand (S_min), maximum size of sand (S_max), minimum size of gravel (G_min), maximum size of gravel (G_max), cement (C) and superplasticizer (PS). The results indicated that SVR outperforms ANN in accuracy for predicting these properties. Additionally, the study acknowledges limitations and points to avenues for further research to enhance predictive modeling’s applicability in sustainable concrete design.
Artificial neural networks and support vector regression for predicting slump and compressive strength of PET-modified concrete
Asian J Civ Eng
Mouzoun, Kaoutar (Autor:in) / Zemed, Najib (Autor:in) / Bouyahyaoui, Azzeddine (Autor:in) / Abdelali, Hanane Moulay (Autor:in) / Cherradi, Toufik (Autor:in)
Asian Journal of Civil Engineering ; 25 ; 5245-5254
01.11.2024
10 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Predicting the compressive strength and slump of high strength concrete using neural network
| British Library Online Contents | 2006
Predicting the compressive strength and slump of high strength concrete using neural network
| Online Contents | 2006