Control of methylamine removal reactor using neural network based model predictive control

Methylamine (MA) removal process using mixed bacteria strains depends highly on constant temperature (303 K), at which the mixed bacteria strains provide highest activity in removing MA. Controlling MA removal reactor is extremely difficult for its inherent process nonlinearities and complex reaction kinetics and other uncertain factors. In the designed approach, a network predicted model is trained as a nonlinear process to predict the future output of the controlled process according to current and previous input and output over the specified horizon. The advanced predictive control strategy is used to minimize the cost function in order to calculate the optimal output of the controller. In this work, a neural network based predictive control (NNMPC) algorithm was implemented to control the temperature of MA removal reactor and the controller performance in set-point tracking and disturbance rejection was investigated, and the performance results of NNMPC was compared with conventional PID controller. It is concluded that the NNMPC performance is superior to the conventional PID controller in the control of MA removal reactor.