RETRACTED: Implementing machine learning and statistical optimization analysis for oily wastewater treatment in a photocatalytic membrane reactor

Abstract

Photocatalytic membrane reactor systems usually face process problems, including photocatalyst deactivation, fouling formation on the membrane surface, and unwanted reactions. Optimizing effective parameters in this process is very important. In this study, the influence of operating parameters such as Trans Membrane Pressure, flow rate, and reaction time were investigated on the permeate flux in oily wastewater treatment purified by ultrafiltration process with γ-Al2O3 ceramic membrane and TiO2 photocatalyst in PMR. To find the optimized parameters, response surface methodology was used. In addition, experimental data were modeled with an artificial neural network. The correlation matrix analysis was used to determine the importance of selecting input parameters for the ANN model. To calculate the photo-reaction kinetics on the membrane surface, the Langmuir–Hinshelwood correlation was used. By doing this optimization, the efficiency reduction barriers such as sediment formation, photocatalyst deactivation, and unwanted reactions were reduced to some extent. Results of RSM showed that the interaction of TMP and flow rate had the most significant impact on the output of PMR. The ANN model predicted the experimental data with an AARD % of 1.276 %, MSE of 0.006237, and R2 = 0.999614. The value of 0.005 for the overall degradation rate constant was obtained based on the experimental data. Based on the experimental and modeling results, to control and optimize the operating parameters, this system can be used on an industrial scaleup to increase the efficiency of the purification process.