Systems (100) and methods are disclosed herein for proactive intervention in an industrial process involving a
defoamer composition.
Learning models are iteratively trained using historical input data sets comprising directly measured variables from the industrial process and correlations between combinations of the input data and respective process states. For a current input
data set associated with the industrial process, and responsive to detected process attributes (114), an intervention event is determined with respect to at least a first
defoamer component (116) in a current composition. Models are utilized to predict, responsive to the intervention event, changes in target values within a specified range of target values for other components in the current composition and / or for directly measured variables as corresponding with a specified
process state.
Automatic control is performed for respective actuators to produce a new
defoamer composition, and as needed to other actuators for the directly measured variables.