Prediction of Compressive Strength of Rice Husk Ash Mix Concrete Using Support Vector Regression and Artificial Neural Networks Integrated with Artificial Bee Colony and Particle Swarm Optimization Algorithms

Rice husk ash (RHA) is one of the most abundant agricultural wastes. The application and usage of RHA as a supplementary cementitious material (SCM) within the concrete mix are environmentally valuable. The compressive strength (CS) of concrete depends on the composition and ratio of its constituent materials. The present study employed two commonly used and popular data-driven methods, namely artificial neural network (ANN) and support vector regression (SVR), for predicting the CS of concrete containing RHA. These data-driven models were hybridized with meta-heuristic algorithms, including particle swarm optimization (PSO) and artificial bee colony (ABC) algorithms, to examine the accuracy of the prediction of CS using the developed hybrid models. The findings revealed that the hybrid models were more accurate than the standalone models. In addition, the performance of hybrid ANN models (i.e., ANN-ABC and ANN-PSO) was more accurate than that of hybrid SVR models (i.e., SVR-ABC and SVR-PSO). Statistical measurements demonstrated that the combination of ABC compared to PSO with ANN and SVR led to a slightly better prediction of CS. This study highlighted the potential of integrating data-driven approaches with meta-heuristic algorithms for improving the accuracy of the prediction of CS in RHA blended concrete.