Efficient and robust state estimation: Application to a copolymerization process

Polymerization processes are highly non-linear systems that require strict control of their dynamic operation to be competitive. The unscented Kalman filter is a filtering strategy has shown rewarding performance for state estimation. Besides, filters based on robust statistics have been proposed deal with the presence outliers. However, reported employed only Huber M-estimator as loss function estimation problem. This work presents new state-estimation procedure transformation and concepts. When outliers present, estimates more accurate than when using conventional filter. In contrast previous research, our methodology also efficient there no performances different functions solving problem presented. results show redescending M-estimators outperform function. behaviour technique analyzed copolymerization process.