Fuzzy Model based Predictive Control of a Hammerstein Model with Constraints Handling

Controlling complex systems, containing nonlinearities and constraints, is always a domain of interest for researchers. Different schemes such as Nonlinear Model Predictive Control (NMPC) have been proposed but these are usually computationally demanding and complex by themselves. Nowadays two powerful techniques, Fuzzy Modeling and Model Predictive control, are being blended together in different ways to attain the advantages of controlling delayed, non-minimum phase, nonlinear and constrained systems efficiently. The paper discusses this novel technique to control a class of nonlinear systems, the Hammerstein models. In order to control the Hammerstein model containing static nonlinearity with linear GPC algorithm a 0th order TS adaptive fuzzy inverse model controller is used with feedback. This strategy makes the scheme powerful enough to be used with different Hammerstein models containing different types of static nonlinearities and different linear dynamics. Furthermore the GPC algorithm is developed by using Toeplitz/Hankel matrices rather than Diophantine equation to reduce computational complexity. The scheme has been tested in MATLAB/Simulink for different kinds of nonlinearities and different parameters of system dynamics, and has produced good control results. Simulation results show that in any case, the optimization remains convex and single layer of optimization is sufficient to optimize the cost function.