Nonlinear Auto-Regressive Moving Average (NARMA-L2) Controller Design for Two CSTR

Continuous Stirred Tank Reactors (CSTRs) are important components in chemical processes. They are often interconnected to achieve desired reaction sequences. However, controlling the concentration in such interconnected systems presents significant challenges due to nonlinear dynamics and interactions between reactors. This paper proposed a Nonlinear Auto-Regressive Moving Average (NARMA-L2) controller developed to regulate the concentration for a series-connected two-CSTR system. The NARMA-L2 controller takes advantage of the inherent capabilities of neural networks to accurately model the nonlinear behavior of a two-CSTR system, and it is developed to regulate the concentration of two CSTRs that are connected in a series. The controller is designed, implemented, and trained using Simulink-MATLAB. The ability of the NARMA-L2 controller to monitor concentration target values and reject disturbances effectively is assessed through comparative analysis with the conventional Proportional-Integral-Derivative (PID) controller. The Outcomes demonstrated the exceptional performance of the NARMA-L2 controller in achieving precise concentration control in the series-connected two-CSTR system.