Identification of a Parameter-Varying Mist Reactor for Cell Culture

Abstract

The identification of a multivariable mist reactor for cell culture is considered, for application in model predictive control. A unified identification framework for linear parameter-varying systems is proposed based on a combination of local and global approaches. The technique is able to achieve high model accuracy with low computational complexity. The model is applied in a model predictive controller to control the temperature and humidity of the reactor. It achieves an average of 40% and 28% improvement in the integral absolute error with respect to the linear approach and a competing parameter-varying approach, respectively.