期刊
CONSTRUCTION AND BUILDING MATERIALS
卷 317, 期 -, 页码 -出版社
ELSEVIER SCI LTD
DOI: 10.1016/j.conbuildmat.2021.125876
关键词
Concrete mixtures; Natural zeolite; Compressive strength; Electrical resistivity; Strength prediction; Artificial neural network
The study demonstrates the potential of using artificial neural networks to predict the compressive strength and electrical resistivity of natural zeolitic concrete, significantly speeding up the process and improving accuracy. Experimental results from 324 different designs of natural zeolitic concrete specimens validate the accuracy and reliability of the model. This has important implications for cost reduction and time saving.
To decrease the environmental and climatic effects of rising concrete consumption, more environmentally friendly concretes are required. One approach to achieve this goal is using natural pozzolans (NP) in concrete mixtures. Natural zeolite (NZ) as a highly reactive NP can improve concrete's durability and mechanical properties. However, tests to estimate concrete strength may take a long time and be costly. Therefore, using computational intelligence techniques, particularly artificial neural networks (ANNs), can help speed up and simplify the process. Hence, this study aims to explore the potential of employing an ANN model to predict compressive strength (CS) and electrical resistivity (ER) of natural zeolitic concrete (NZC). The experimental results of 324 NZC specimens made from 54 different mix designs are used to develop the ANN model. Seven variable inputs are considered when designing the ANN model to predict CS and ER values, including specimen age, water-to-cementitious materials ratio, cement, NZ, gravel, sand, and superplasticiser contents. Extensive computational tests were conducted to evaluate the performance of the proposed model against results obtained by experimental tests and existing gene expression programming (GEP) in the literature. The RMSE values for CS and ER are 1.65 MPa and 3.96 omega-m, respectively, which confirm the model's accuracy and robust predictive capability. The study's findings have the potential to assist in cutting costs and saving time by using a reliable prediction technique rather than conducting costly and time-consuming tests.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据