Identification and model predictive control of continuous stirred tank reactor based on artificial neural networks

In this paper, system identification and neural network predictive control (NNPC) of a continuous stirred tank reactor (CSTR) is presented. The control problem with the objective of set point tracking between several modes is investigated. The real measurements of this process are used in system identification. Artificial networks such as MultiLayer Perceptron (MLP), Radial Basis Function (RBF), and Adaptive Neuro-Fuzzy Inference Systems (ANFIS) are applied as the intelligent identifiers to system identification. Neural network predictive control is utilized to control of continuous stirred tank reactor output. The predictive control strategy is used to calculate optimal control inputs. Two viewpoints are considered in neural network predictive control. One of them is based on neural network model of continuous stirred tank reactor and another one is based on dynamical model of continuous stirred tank reactor. Simulation results show the validity and feasibility of the proposed methods to neural network predictive control of continuous stirred tank reactor process.