A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Machine learning in mix design of Miscanthus lightweight concrete
Highlights A machine learning model is developed to predict the compressive strength of Miscanthus lightweight concrete. It employs the Gaussian Process Regression. The study is based on a total of 414 data sets (inputs-output pairs), which are derived from an experimental study. The influence of input variables is investigated.
Abstract This research is carried out to investigate the Gaussian process regression (GPR) based on a machine learning model to predict the compressive strength of Miscanthus lightweight concrete (MLWC). A database of 414 experimental data, which includes nine input variables such as six mix constituents of concrete, form of specimen, curing time and pre-treatment condition and an output variable of compressive strength of MLWC, is constructed from the data collected by a series of experimental tests on MLWC. Two kernel functions, namely, the squared exponential and rational quadratic are used in the GPR model. It is found from experiments that the GPR model with rational quadratic kernel gives minimum errors for predicting compressive strength of MLWC. In addition, a user-friendly graphical user interface is created using MATLAB software to deploy the GPR model which can be used at an early stage of designing the Miscanthus concrete members instead of using costly experimental investigation.
Machine learning in mix design of Miscanthus lightweight concrete
Highlights A machine learning model is developed to predict the compressive strength of Miscanthus lightweight concrete. It employs the Gaussian Process Regression. The study is based on a total of 414 data sets (inputs-output pairs), which are derived from an experimental study. The influence of input variables is investigated.
Abstract This research is carried out to investigate the Gaussian process regression (GPR) based on a machine learning model to predict the compressive strength of Miscanthus lightweight concrete (MLWC). A database of 414 experimental data, which includes nine input variables such as six mix constituents of concrete, form of specimen, curing time and pre-treatment condition and an output variable of compressive strength of MLWC, is constructed from the data collected by a series of experimental tests on MLWC. Two kernel functions, namely, the squared exponential and rational quadratic are used in the GPR model. It is found from experiments that the GPR model with rational quadratic kernel gives minimum errors for predicting compressive strength of MLWC. In addition, a user-friendly graphical user interface is created using MATLAB software to deploy the GPR model which can be used at an early stage of designing the Miscanthus concrete members instead of using costly experimental investigation.
Machine learning in mix design of Miscanthus lightweight concrete
Pereira Dias, Patrick (author) / Bhagya Jayasinghe, Laddu (author) / Waldmann, Daniele (author)
2021-07-07
Article (Journal)
Electronic Resource
English
| British Library Online Contents | 2017
| British Library Online Contents | 2017