Multi-model approach for 2-DOF control of nonlinear CSTR process

This paper proposes a novel approach to examine the performance of nonlinear continuous stirred tank reactor (CSTR) at large set point changes where a single model adaptive controller falls short. The efficient closed-loop performance is achieved through multiple models accompanied with a bank of pole placement controllers. To select the best controller at a particular operating point, several multi-model techniques based on different scheduling variables have been adapted. To avoid the multiple switching and discontinuity in switching, the global control signal is calculated by assigning certain weight to each controller. The advantages of the proposed method are a satisfactory performance at an unstable operating point as well as at large set point changes with reduced number of models. The effectiveness of the scheme is evaluated and demonstrated through simulation results at different scenarios to achieve 2-DOF control. The comparative study between various techniques shows the superiority of gain scheduler designed on the reference signal for set point tracking and disturbance rejection to tackle the nonlinearities of CSTR.