Research on 3D printing concrete mechanical properties prediction model based on machine learning

This study proposes an effective machine learning-based prediction method to satisfy the urgent requirement to anticipate the mechanical properties of 3D-printed concrete. The goal is to support the accurate use of 3D printing technology in the building sector. We have successfully created machine l...

Full description

Saved in:
Bibliographic Details
Main Authors: Yonghong Zhang, Suping Cui, Bohao Yang, Xinxin Wang, Tao Liu
Format: Article
Language:English
Published: Elsevier 2025-07-01
Series:Case Studies in Construction Materials
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2214509525000531
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This study proposes an effective machine learning-based prediction method to satisfy the urgent requirement to anticipate the mechanical properties of 3D-printed concrete. The goal is to support the accurate use of 3D printing technology in the building sector. We have successfully created machine learning models that can predict compressive strength and flexural strength by combining experimental data from a variety of 3D printed concrete samples and carefully preparing the data. Our study explores the fundamentals and practicality of several models, such as artificial neural networks, decision trees, random forests, support vector regression, and linear regression. We have made sure that our prediction findings are reliable and scientifically sound by implementing stringent model training and validation procedures. With a correlation coefficient between 0.96 and 0.98 with real values, experimental results demonstrate the random forest model's remarkable predicted accuracy, greatly beyond that of conventional prediction techniques. The practical use of 3D printed concrete in engineering projects is strengthened by this work, which also opens up new avenues for investigation and highlights the enormous potential of machine learning to improve the prediction of mechanical properties of building materials.
ISSN:2214-5095