Decoupling based predictive control analysis of a continuous stirred tank reactor

摘要

This paper presents a decoupling based model predictive control strategy for a process of continuous stirred tank reactor (CSTR). The MPC strategy when used along with decoupling compensation technique seem to suffer from increase in the computational complexity of the predictive control algorithm because of the additional dynamics of decoupling compensator. In this paper, we seek to reduce the computational burden on predictive control algorithm utilizing a simple decoupling technique for a process with loop interactions. A CSTR MIMO process is a benchmark problem for studying loop interactions. We compensate for loop interactions by directly modifying the plant model without changing its response characteristics provided plant and model mismatch is at minimum. The proposed strategy is compared for its simplicity with multi-variable decoupling control strategies with application to MPC in process industry. Simulations show that proposed technique efficiently reduces loop interactions in CSTR process.