Adaptive neuro-fuzzy interface system (ZNFIS) controller for polymerization reactor

摘要

It is a challenging task to control polymerization reactor due to the complex reactions mechanism. Moreover, the dynamic behaviour of the polymerization reactor is highly nonlinear. Thousand of reactions involed during polymerization that make the system complex in nature. Artificial intelligent appeared as promising tool to control such kind of nonlinear and complex processes. In the present work, a advanced nonlinear controller, namely adaptive neuro-fuzzy interface system (ANFIS) is proposed and designed for polymerization reactor. Sugeno type fuzzy interface system is used in ANFIS. Hybrid optimization algorithm, a combination of least-square estimation and backpropagation methods is used to optimize the neural network-based fuzzy output model. Styrene free radical polymerisation batch reactor is used as a case study. Simulation results demonstrated that the tracking performance of the ANFIS-based controller is better than the traditional neural network (NN)-based controller.