A FUZZY NEURAL-NETWORK APPROACH FOR NONLINEAR PROCESS-CONTROL

This paper proposes an internal model control (IMC) scheme using a fuz zy neural network for process modeling. A fuzzy neural network is most useful in an environment where first-principles-based descriptions ar e difficult to obtain, but partial knowledge about the process is know n and input-output data is available. However, previously proposed fuz zy neural-network approaches are inadequate for modeling complex chemi cal process systems, as when the input dimension increases, the number of hidden nodes (rules) increases exponentially. A novel fuzzy neural -network structure using hyper ellipsoids is proposed to avoid this pr oblem. A fuzzy neural network is trained using steady-state as well as transient data by back-propagation. The inverse of the process is obt ained by a simple interval halving method. The proposed approach is ap plied to modeling and control of a continuous stirred tank reactor and a pH neutralization process. The results show significantly better pe rformances in comparison with a PID controller.