Predictive analysis of different mechanical parameters of concrete using an efficient data-driven algorithm and principal component analysis
The superior mechanical characteristics of concrete have made it a versatile material for construction purposes. Therefore, it is important to have an early understanding of these characteristics based on the concrete mixture characteristics. While recent studies have demonstrated the capability of...
Saved in:
| Published in: | Construction & building materials Vol. 489; p. 141883 |
|---|---|
| Main Authors: | , , |
| Format: | Journal Article |
| Language: | English |
| Published: |
Elsevier Ltd
29.08.2025
|
| Subjects: | |
| ISSN: | 0950-0618 |
| Online Access: | Get full text |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The superior mechanical characteristics of concrete have made it a versatile material for construction purposes. Therefore, it is important to have an early understanding of these characteristics based on the concrete mixture characteristics. While recent studies have demonstrated the capability of ensemble learning for this purpose, many models such as metaheuristic algorithms still suffer from issues such as high dimensionality and time-consuming analysis. In this study, a novel hybrid algorithm is proposed to deliver efficient predictions of four concrete mechanical parameters, namely compressive strength (CS), tensile strength (TS), load capacity (LC), and slump (SL). The model relies on a feed-forward multi-layer perceptron (FFMLP) neural network that is optimally trained by a quick and robust metaheuristic algorithm called electromagnetic field optimization (EFO). Based on the results, the proposed algorithm not only provides excellent predictions of the CS, TS, LC, and SL but also presents improvements in terms of time-effectiveness and accuracy compared to several earlier metaheuristic-based models. The EFO, therefore, can be an efficient alternative to earlier burdensome algorithms for non-destructive analysis of concrete parameters. Moreover, the used datasets underwent a well-known statistical technique called principal component analysis (PCA) for determining the key factors that influence the analysis of the CS, TS, LC, and SL. In short, the integrated outcomes of this study can contribute to improving the design of concrete with respect to both intended mechanical parameters and influential factors.
•Different mechanical parameters of concrete are predicted.•Feasibility of improved artificial neural networks for this task is approved.•Proposed EFO model provided more efficient solutions than various benchmarks.•PCA is used for feature analysis within the dataset. |
|---|---|
| ISSN: | 0950-0618 |
| DOI: | 10.1016/j.conbuildmat.2025.141883 |