Compressive strength estimation of ultra-great workability concrete using hybrid algorithms

This paper aimed to analyze the capability of the hybrid adaptive neuro-fuzzy inference system (ANFIS) networks to estimate the compressive strength ( fc) of the ultra-great workability concrete (UGWC). The outline of estimation is as follows, where hybrid ANFIS networks were suggested, in which principal attributes of the model were found out using two optimization algorithms named whale optimization algorithm (WOA) and particle swarm optimization (PSO). The gathered observation rows (170 samples) were divorced randomly for two phases, train and test phase of 75% (128) and 25% (42), respectively. The calculated workability indices indicate that the ANFIS models could result in remarkable accuracy to estimate fc of the UGWC. Although both WOA-ANFIS and PSO-ANFIS models have great ability of the estimating phenomenon, the ANFIS model optimized with WOA algorithm could outperform another model and literature (R2 equals 0.801). For example, WOA-ANFIS could obtain the most proper values of all criteria, such as increasing the value of R2 from 0.8942 to 0.9494, reducing RMSE from 10.54 to 6.8553, and a 5.54% rise in the value of VAF from 89.42 to 94.95%. All in all, the integrated WOA-ANFIS model could receive larger usefulness compared to PSO-ANFIS and published article, which leads to being categorized as the best model to utilize in practical projects.