Trajectory tracking of batch product quality using intermittent measurements and moving window estimation

In order to meet tight product quality specifications for batch/fed-batch processes, it is vital to monitor and control batch product quality throughout the batch duration. The ideal strategy is to control batch product quality through trajectory tracking of a desirable batch product quality evolution during the batch run. However, due to the lack of in-situ sensors for continuous measurements of batch product quality, the measurement of batch product quality is usually implemented by laboratory assay of samples and thus these measurements are generally intermittent. Therefore direct trajectory tracking of batch product quality is not feasible for such scenarios with intermittent measurements. This paper proposes an approach to use intermittent measurements to realize trajectory tracking control of batch product quality through moving window estimation. The first step of the approach is to identify a partial least squares (PLS) model using intermittent measurements to relate process variable trajectories and batch product quality. Then the identified PLS model is further applied to predict product quality trajectory during the batch run so as to realize trajectory tracking of a desirable product quality evolution. An example from fed-batch fermentation for penicillin production is used to illustrate the principle and the effectiveness of the proposed approach.