A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Modelling Compressive Strength of Concrete Incorporating Supplementary Cementitious Materials Using Machine Learning Technologies
This research utilised machine learning (ML) technologies to predict compressive strength of concrete that contains supplementary cementitious materials. A comprehensive database for concrete compressive strength was established, encompassing ten input parameters, including cement, slag, unique additive, fly ash, water-to-binder ratio, coarse aggregate with maximum diameter of 20 mm, coarse aggregate with maximum diameter of 10 mm, coarse sand, fine sand and superplasticiser, and one output parameter of compressive strength. Using this database, strength prediction models were developed based on four state-of-the-art ML methods, namely, artificial neural networks, support vector machines, Gaussian process regression (GPR) and ensemble decision tree. To improve the generalisation performance of developed ML models, Bayesian optimisation was employed to adjust the model hyperparameters during the training procedure. The performance of these models is evaluated and compared using several metrics The results show that the GRP model has the best performance and outperforms other models in terms of compressive strength prediction.
Modelling Compressive Strength of Concrete Incorporating Supplementary Cementitious Materials Using Machine Learning Technologies
This research utilised machine learning (ML) technologies to predict compressive strength of concrete that contains supplementary cementitious materials. A comprehensive database for concrete compressive strength was established, encompassing ten input parameters, including cement, slag, unique additive, fly ash, water-to-binder ratio, coarse aggregate with maximum diameter of 20 mm, coarse aggregate with maximum diameter of 10 mm, coarse sand, fine sand and superplasticiser, and one output parameter of compressive strength. Using this database, strength prediction models were developed based on four state-of-the-art ML methods, namely, artificial neural networks, support vector machines, Gaussian process regression (GPR) and ensemble decision tree. To improve the generalisation performance of developed ML models, Bayesian optimisation was employed to adjust the model hyperparameters during the training procedure. The performance of these models is evaluated and compared using several metrics The results show that the GRP model has the best performance and outperforms other models in terms of compressive strength prediction.
Modelling Compressive Strength of Concrete Incorporating Supplementary Cementitious Materials Using Machine Learning Technologies
Lecture Notes in Civil Engineering
Chouw, Nawawi (editor) / Zhang, Chunwei (editor) / Yu, Yang (author) / Hajimohammadi, Ailar (author) / Nezhad, Ali (author) / Hocking, David (author) / Moghaddam, Farzad (author) / Foster, Stephen (author)
Australasian Conference on the Mechanics of Structures and Materials ; 2023 ; Auckland, New Zealand
2024-09-03
9 pages
Article/Chapter (Book)
Electronic Resource
English