Computational artificial intelligence modeling and optimization of nutrient removal in microalgae membrane bioreactors using ANN and RSM.

Biological based methods for wastewater treatment are preferred because of they are more economical and environmentally friendly. This work sought to predict microalgae membrane bioreactor (MBR) levels of ammonia nitrogen, phosphate concentration, and removal efficiency. Forecasting the values in the study using artificial neural network (ANN) modeling and response surface methodology (RSM). According to the data, the RSM model accurately estimates the effectiveness of phosphate, nitrogen, and ammonia removal. The R2 values of 0.9733 and 0.9608 help to justify the significant correlation between the predictions of the model and the removal efficiencies. The model uses time to predict the removal effectiveness of these contaminants, along with TOC content. Moreover, the performance metrics for ANN designs are R2 coefficient of determination of 0.99791, mean squared error (MSE) of 3.57E-02 for ammonia nitrogen, and R2 coefficient of determination of 0.99722 and MSE of 2.75E-02 for phosphate concentration. Nonetheless, variations between the actual data and the ANN’s predictions suggest possible limits in the model or experimental design. Furthermore, using RSM and NSGA-II in the optimization process demonstrated the model’s ability to accurately pinpoint the ideal circumstances for removing nutrients from the bioreactor.