A comparative study of state estimation methods for processes with asynchronous and delayed measurements

This paper presents a comparison between two state estimation methods for processes with asynchronous and delayed measurements in order to verify which of these techniques is better in terms of dynamic behavior and possibilities of implementation in a real time processor. The methodologies are analyzed through simulations that allow a comparison in a visual form, through error index and with the time of computational load. The analyzed estimators consists in a cascade observer-predictor algorithm where the observer stage is based on a second order sliding mode algorithm, followed by a Smith predictor and the second one is based on the Kalman filter with augmented state space using the fixed-lag smoothing method. These estimators were applied on a batch bioprocess used in the production of δ-endotoxins of Bacillus thuringiensis.