Abstract
The presented research results focus on modelling and control of the operation of a productive plant used for the 18O isotope separation. The proposed mathematical model which describes the separation plant operation is based on using two strong nonlinear differential equations. To determine the separation process structure parameters, the available experimental data are processed. In the modelling procedure, the interdependency between the operations of the two separation columns from the separation plant structure is considered, too. The behaviour of the nonlinear structure parameters associated to the proposed model is learned using neural networks. The proposed control strategy for the 18O isotope concentration at the output of the separation plant is based on the design of two controllers with complex structure, one for each of the two separation columns of the separation plant, controllers which integrate a model-based solution. The high performances of the proposed control strategy are proven through simulation, the simulations results being presented and interpreted in the final part sections of the manuscript. The previously mentioned high performances are obtained in the context of maintaining all the intermediary signals from the operation of the designed control systems between their imposed variation limits, aspect which makes feasible the proposed control solution for the practical implementation.