Integrating experimental design and machine learning for efficient dye adsorption in textile wastewater treatment

The textile industry generates large volumes of dye-contaminated wastewater, posing significant environmental risks. This study aims to optimize the adsorption of textile dyes from wastewater using activated carbon through a general factorial design. The research investigates the effects of key variables, including activated carbon type (powdered and granular), contact time, carbon dosage, salinity, and temperature, on dye removal efficiency. A comprehensive analysis using ANOVA identifies the most significant factors and their interactions. The study also integrates machine learning models, such as Gaussian process regression (GPR) and artificial neural networks (ANN), to predict adsorption outcomes. The results demonstrate that granular activated carbon (GAC) and lower salinity significantly enhance dye removal efficiency. ANN models outperformed GPR in predictive accuracy, achieving an R 2 value of 0.85 compared to GPR’s 0.71. The findings provide optimized operational parameters that can be applied in the textile industry for efficient wastewater treatment and environmental compliance.