Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Evaluating machine learning algorithms for predicting compressive strength of concrete with mineral admixture using long short-term memory (LSTM) Technique
The prediction of compressive strength in concrete holds essential significance within the construction industry, maintaining the structural reliability of important infrastructure like buildings. In the present study, a thorough investigation has been conducted to evaluate the effectiveness of using various fly ash and admixtures as input factors to forecast compressive strength at 7, 14 and 28 days. Two machine-learning algorithms, namely classification and regression trees (CART) and long short-term memory (LSTM) neural networks, are used in this investigation. The dataset has been partitioned into 80% training subset and 20% testing subset. The performance evaluation of each model has been assessed by key performance metrics such as root mean squared error (RMSE), mean squared error (MSE), and the coefficient of determination (R2). Among the two machine-learning models studied, LSTM shows superior predictive capabilities; with an R2 value of 0.9975 and 0.9948 for training and testing respectively and an impressively low root mean squared error value of 1.9175. Therefore, LSTM has demonstrated superior accuracy and predictive capabilities compared to CART; establishing it as a viable and effective technique for forecasting compressive strength of concrete. This research contributes to the progression of predictive modeling in concrete engineering, enabling more precise estimations of compressive strength using LSTM based machine learning (ML) modeling.
Evaluating machine learning algorithms for predicting compressive strength of concrete with mineral admixture using long short-term memory (LSTM) Technique
The prediction of compressive strength in concrete holds essential significance within the construction industry, maintaining the structural reliability of important infrastructure like buildings. In the present study, a thorough investigation has been conducted to evaluate the effectiveness of using various fly ash and admixtures as input factors to forecast compressive strength at 7, 14 and 28 days. Two machine-learning algorithms, namely classification and regression trees (CART) and long short-term memory (LSTM) neural networks, are used in this investigation. The dataset has been partitioned into 80% training subset and 20% testing subset. The performance evaluation of each model has been assessed by key performance metrics such as root mean squared error (RMSE), mean squared error (MSE), and the coefficient of determination (R2). Among the two machine-learning models studied, LSTM shows superior predictive capabilities; with an R2 value of 0.9975 and 0.9948 for training and testing respectively and an impressively low root mean squared error value of 1.9175. Therefore, LSTM has demonstrated superior accuracy and predictive capabilities compared to CART; establishing it as a viable and effective technique for forecasting compressive strength of concrete. This research contributes to the progression of predictive modeling in concrete engineering, enabling more precise estimations of compressive strength using LSTM based machine learning (ML) modeling.
Evaluating machine learning algorithms for predicting compressive strength of concrete with mineral admixture using long short-term memory (LSTM) Technique
Asian J Civ Eng
Gogineni, Abhilash (Autor:in) / Rout, M. K. Diptikanta (Autor:in) / Shubham, Kumar (Autor:in)
Asian Journal of Civil Engineering ; 25 ; 1921-1933
01.02.2024
13 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch