Mechanical behaviour optimization of saw dust ash and quarry dust concrete using adaptive neuro-fuzzy inference system

Adaptive neuro-fuzzy inference system (ANFIS) which possesses adaptive framework that integrates neural network and fuzzy-logic principles was adapted in this study for the optimization of mechanical strength properties of green concrete. The gains derived from deploying soft-computing methods help to handle multiple constraints associated with recycling and re-use of waste materials/derivatives to achieve sustainability and eco-efficient construction materials. Sixty-two datasets extracted from the experiments evaluating the mechanical properties of green concrete with varying replacement ratios of saw dust ash (SDA) and quarry dust (QD) for cement and fine aggregates, respectively, from 0 to 50% were deployed for the ANFIS model development. The outcome of the mechanical strength response shows that SDA and QD can be effective concrete admixture with significant improvement recorded at 0.5% and 0.225% replacement by cement and fine aggregates, respectively. Using ANFIS-toolbox in MATLAB software a smart-intelligent model was developed with hybrid method of optimization and subclustering method of FIS using Gaussian membership function. Mix variations of SDA cement and QD fine aggregates in the matrix were the input variables, while compressive strength result is the output variable. The developed model performance was evaluated using statistical methods, and the computed results showed indicate MAE of 0.144, RMSE of 0.429, MSE of 0.1837 and correlation of 97.20%, while MLR model presents a coefficient of determination of 82.46%.