Black box dynamic modeling of Co(II) ions removal from aqueous solution using modified maghemite nanoparticles by fixed-bed column based on deep neural networks

The removal of cobalt(II) ions from an aqueous solution using a fixed-bed glass column containing modified maghemite γ-FeO@SiO nanoparticle was investigated. The influence of various parameters such as bed depth, concentration of influent cobalt, and feed flow rate on the breakthrough curves and the column performance was also discussed. The column study indicates that the removal of cobalt(II) ions increased with increasing bed depth, whereas it decreased with the increase of both influent cobalt concentration and the flow rate. By increasing the flow rate, column saturation takes a shorter time. This study also indicated that with increasing bed depth the sites available for sorption increased in the column, therefore the input volume of cobalt solution also increased. The kinetic models indicated that the column maximum capacity increased with the flow rate and cobalt concentration, while it decreases with bed depth. Long–short term memory (LSTM) and multilayer perceptron (MLP) are merged to compose deep LSTM neural networks to form and predict the cobalt (II) ions removal onto fixed-bed column system was also concluded in this work.