Experiment design, identification and control in large-scale chemical processes

Experiment design for parameter identification, state and parameter estimation, and model reduction have been studied extensively in the literature. However, most of the methods proposed in the literature are not suitable for, or have not been tested for, large scale and complex systems. In this contribution, we investigate modifications to methods developed for the design of optimal experiments and system identification in order to make them suitable for application to large scale systems. The optimal experiment design and system identification are demonstrated on two different examples. Parameter clustering and principal component analysis are used with D optimal design of experiments for a catalytic kinetic system, the preferential oxidation of carbon monoxide on platinum catalyst. A reparameterization of autoregressive integrated moving average models are used for identification and control of a multiscale stochastic thin film growth process.