Abstract
This study develops empirical models for the prediction of the removal efficiencies of inorganic Nitrogen (N) and Phosphorus (P) from municipal wastewater using microalgae (Chlorella kessleri). Our work identified the effects of operational parameters of temperature (T), light-dark cycle (LD), and nitrate-to-phosphate (N:P) ratio on simultaneous N and P removal. Three competitive soft-computing techniques known as response surface methodology (RSM), multilayer perceptron artificial neural network (MLP-ANN), and support vector regression (SVR) were applied to construct the predictive models using real-life experimental data obtained via the BoxBehnken Design (BBD) matrix. A Bayesian optimization algorithm was applied to automatically tune the hyperparameters to develop optimized MLP-ANN and SVR models. The overall results exhibited that the SVR model is better than MLP-ANN and RSM models to assess simultaneous N and P removal efficiencies. The extra simulated data further confirmed the prediction capability of the developed SVR models under different conditions. Finally, the models developed by SVR were hybridized with a genetic algorithm (GA) to maximize the nutrient (N and P) removal efficiency (>93%) at optimum conditions as 29.3 degrees C, 24/0 h/h of LD, and 6:1 of N:P.