Analysis of mechanical properties of fly ash and bauxite residue based geopolymer concrete using ANN, Random Forest and Counter propagation neural network

In recent years, geopolymer concrete has attracted considerable interest as a sustainable substitute for conventional Portland cement-based concrete. The geopolymer concrete (GPC) was developed using fly ash (FA) and bauxite residue (BR). FA and BR, both are the industrial by-products which contained silica and alumina that are essentials for the synthesis of the geopolymerisation reaction of the GPC. This study presents the synthesis of the GPC with the partial replacement of the FA with BR at different molar concentration of the NaOH 8, 10, 12 and 14 M of NaOH. The maximum compressive strength was obtained at 12 M NaOH 61.04 MPa. The maximum strength was also obtained when the 30% of FA was replaced with BR. Achieving R² values of up to 0.98 on training sets and 0.91 on testing sets, the artificial neural network (ANN) model showcases its robustness and dependability in making predictions. Conversely, while the Random Forest (RF) model proves proficient in forecasting compressive strength, its performance fluctuates across flexural and split tensile strengths, notably when confronted with new data. The prediction of the GPC was evaluated using counter propagation neural network (CPNN). The CPNN model was best suited for the compressive strength with R² values of 0.99.