Abstract
The manuscript introduces an innovative method for modeling and simulation of the 18O isotope concentrations dynamics inside a separation cascade which contains within its structure 2 separation columns. The proposed model is designed as a nonlinear distributed parameter one. The present study, in order to maximize the model accuracy, does not use simplification and linearization techniques. The functional form of the separation cascade structure parameters is established through unique identification procedures, which rely on experimental data obtained from the real plant. A major difficulty in the study approach was the experimental data take-off, taking into account that the considered separation process is a very slow one and it has strict operation conditions. Due to the nonlinearity of the functions which describe the structure parameters of the plant, two neural networks are used to establish their dynamics. The two neural networks are included in the proposed model to facilitate simulation capabilities. The validation of the model is made through simulation as well as the comparison of the obtained results with the experimental data. Also, the suggested model is simulated in various feasible alternatives, thus resulting in some interesting interpretations and conclusions.
DOI: 10.61416/ceai.v26i4.9095