Model predictive control of nonlinear affine systems based on the general projection neural network and its application to a continuous stirred tank reactor

Model predictive control (MPC) is an advanced technique for process control. It is based on iterative, finite horizon optimization of a cost function associated with a plant model. Neural network is an effective approach for on-line optimization problems. In this paper, we apply the general projection neural network for MPC of nonlinear affine systems. Continuous stirred tank reactor (CSTR) system is a typical chemical reactor widely used in chemical industry and can be characterized as a nonlinear affine system. The general projection neural network based MPC is applied to the CSTR problem with input and output constraints. This application demonstrates the usefulness and effectiveness of proposed MPC approach to industrial problems. © 2011 IEEE.