Nonlinear and hybrid modeling, simulation, and control: current technology and future views at pavilion technologies

The need for nonlinear models in MPC is well recognized. Researchers in academia and industry have developed nonlinear models that are appropriate for control design using first principle information and/or empirical data. Of a host of modeling approaches that have been developed, few have gotten widespread industrial applications. Full endorsement by practitioners have been contingent on several factors including: (a) the cost of development and maintenance of the models, (b) their suitability for massive computations that are required to make control decisions online, and (c) the generality by which these modeling concepts and tools could be applied in different applications. The purpose of this presentation is to highlight the need for efficient hybrid modeling and control techniques for the complex industrial applications towards which today's model predictive control systems are aiming.