Distributed parameter bioprocess plant identification and I-term control using centralized recurrent neural network models

In the paper, it is proposed to use a recurrent neural network model, and a real-time Levenberg-Marquardt algorithm of its learning for centralized data-based modeling, identification and control of an anaerobic digestion bioprocess, carried out in a fixed bed and a recirculation tank of a wastewater treatment system. The analytical model of the digestion bioprocess, used as process data generator, represented a distributed parameter system, which is reduced to a lumped system using the orthogonal collocation method, applied in four collocation points plus one- in the recirculation tank. The paper proposed to use three types of I-term adaptive control: direct adaptive integral plus states neural control, indirect adaptive I-term sliding mode control and real-time I-term optimal control. The comparative graphical simulation results of the digestion wastewater treatment system control, exhibited a good convergence and precise reference tracking, giving slight priority to the direct control with respect to the other methods of control applied.