Soft-sensors in industrial process control

Two methods for inferring (soft-sensing) process outputs which can only be measured at infrequent intervals of time, from other more rapidly sampled secondary outputs, have been developed. One estimator is based upon a general input-output representation of the plant as the starting point for synthesis, the second method, starts from state space considerations and uses an observer canonical form for estimator derivation. The developments are supported by results from some industrial scale site tests: on-line determination of product composition in a large industrial distillation tower and in a companion study, biomass concentration in industrial mycelial fermentation. The technique uses measurements from established instruments. The structure of the estimators (soft-sensors or sensors in software) is not application dependent and their parameters can be continuously estimated and updated. As a result, slow variations in plant or disturbance characteristics can be tracked.<
>