Investigating the rheological characteristics of alkali-activated concrete using contemporary artificial intelligence approaches

• DOI: 10.1515/rams-2024-0006
• Publication Date: 2024-04-12T13:10:57Z
• Authors: Muhammad Nasir Amin; Ahmed A. Alawi Al...; Roz-Ud-Din Nassar; Omar Algassem; Suleman Ayub Khan; Ahmed Farouk Deif...
• Keywords: Perceptron; Multilayer perceptron

Abstract Using artificial intelligence-based tools, this research aims to establish a direct correlation between the alkali-activated concrete (AAC) mix design factors and their performances. More specifically, machine learning system was fed new property data obtained from AAC mixes used in laboratory experiments. The rheological parameters (yield stress [static/dynamic] plastic viscosity) of were predicted using multilayer perceptron neural network (MLPNN) bagging ensemble (BE) models. In addition, R 2 values, k-fold analyses, statistical checks, dissimilarity experimental compressive strength employed assess performance created Also, SHapley additive exPlanation (SHAP) approach for examining relevance influencing parameters. BE found be significantly accurate all prediction models, with greater than 0.90, MLPNN models moderately precise, slightly below 0.90. However, error assessment through checks analysis also validated higher precision over Building that can calculate properties different values input could save lot time money compared doing tests laboratory. order ascertain required amounts raw materials AAC, investigators, as well businesses, may find SHAP study helpful.