Concrete, an artificial stone composed of cement, aggregate, water, and additives, is extensively utilized in contemporary civil projects. A pivotal characteristic of concrete is its capacity to efficiently serve various purposes and structural requirements. Cement, wat
More
Concrete, an artificial stone composed of cement, aggregate, water, and additives, is extensively utilized in contemporary civil projects. A pivotal characteristic of concrete is its capacity to efficiently serve various purposes and structural requirements. Cement, water, aggregate, and additives are pivotal parameters wherein even minor alterations can significantly impact concrete strength. Among these parameters, the Water/Cement (W/C) ratio holds particular significance due to its inverse correlation with strength. Traditionally, predicting concrete strength solely based on the water-to-cement ratio has been challenging. However, with advancements in AI and machine learning techniques coupled with ample data availability, accurate strength prediction is achievable. This paper presents an analysis of a diverse dataset comprising various concrete tests utilizing machine learning methodologies, followed by a comparative examination of the outcomes. Furthermore, this study scrutinizes a renowned dataset encompassing 1030 experiments, featuring diverse combinations of cement, water, aggregate, etc., employing artificial intelligence and machine learning techniques. Model accuracy and result fidelity are evaluated through rigorous sampling methodologies. Initially, the dataset is subjected to analysis utilizing the linear regression algorithm, followed by validation employing the random forest algorithm. The random forest algorithm is employed to predict the water-to-cement ratio and corresponding compressive strength for concrete with a density of 300 kg/m3. Notably, the obtained results exhibit a high level of concordance with experimental and laboratory findings from prior studies. Hence, the efficacy of the random forest algorithm in concrete strength prediction is established, offering promising prospects for future applications in this domain.
Manuscript profile