machine learning assisted prediction of fouling recovery ratio of ultrafiltration membranes

Intelligent approaches based on multilayer perceptron (mlp) and gaussian process regression (gpr) were applied for modelling to estimate the fouling recovery ratio (frr) of ultrafiltration membrane for waste water treatment. the pressure, temperature, and ph were used as variables. the gpr model showed an excellent agreement with experimental data with average absolute relative error (aare) of 0.87% relative root mean squared error (rrmse) of 1.40% and r2 of 99.29%. the performance of the gpr model for prediction frr were assessed and acceptable results were obtained. a sensitivity analysis was showed that the pressure is the most effective parameter on membrane frr, which is followed by ph and temperature, respectively.