AI-driven neural time series network forecasting and cost analysis for dye removal prediction in packed bed adsorption using ultrasonic biomass composites for sustainable wastewater management.

The study investigates the application of Artificial Intelligence (AI) driven neural network time series (NNTS) model for the forecasting prediction of dye removal using ultrasonic activated mixed biomass. Surface and functional characterization of ultrasonic activated algal-jack fruit seed biomass (UAJSB). An experimental evaluation of the influence of column adsorption parameters indicated that higher bed height (15 cm) and lower flow rate (10 ml/min) resulted in higher dye removal. Column adsorption performance was analyzed through different breakthrough models - Yoon-Nelson, Adams-Bohart, and Bed Depth Service Time (BDST) model. The Bed Depth Service Time model resulted in a higher correlation coefficient, indicating a proportional relationship between bed height and dye removal, with adsorption capacities up to 2189 mg/L and a strong linear correlation of greater than 0.99. NNTS modeling applied for dye removal prediction analysis exhibited better performance with a correlation coefficient of 0.99992 for basic orange dye adsorption, with minimal prediction error, indicating strong correlation with the experimental data. Cost analysis showed the economically viable nature of UAJSB with a cost of ₹178/kg. The study highlights the integrated use of experimental and AI-based modeling for forecasting the removal prediction in column systems and demonstrating the potential of UAJSB composites for dye removal.