Optimizing aerobic granular sludge process performance: Unveiling the power of coupling experimental factorial design methodology with artificial intelligence modeling

This research explored innovative approaches, integrating artificial intelligence (AI) and design of experiments, to enhance the performance of the aerobic granular sludge (AGS) process in wastewater treatment. A hybrid model coupling artificial neural networks and random forests (ANN-RF) with response surface methodology (RSM) via central composite design (CCD) and Box-Behnken design (BBD) was developed to improve the optimization process. This model was compared to another one pairing fuzzy logic (FL) and genetic algorithm (GA), using CCD for random run generation. The experimental datasets were obtained from three AGS batch reactors operated under varying hydraulic retention times (HRT) and different chemical oxygen demand (COD), N, and P ratios. Both RSM-AI and FL-GA models optimized these variables toward maximum COD removal. RSM-AI achieved a high fit (desirability index of 1), resulting in 99.1 % COD removal efficiency at COD:N and COD:P ratios of 35–145 and 145–355, respectively, and an HRT of 7.6 h. FL-GA achieved 96.8 % COD removal efficiency at different COD:N and COD:P ratios with an HRT of 6 h. Integrating RSM, AI, FL, and GA presented a comprehensive approach, contributing to AGS reactor scale-up for sustainable wastewater treatment.