A methodology for the closed loop control of biscuit quality during baking
is proposed and was validated during several experiments. This methodology
involves expert knowledge which is captured in a modular way. Three modules
are involved in the control strategy of the process. The first module repr
oduces sensory evaluation by the operator. The second module diagnoses the
state of the process with respect to quality state of biscuits. Finally, a
decision module supervises the actuators of the process. Different fuzzy co
ncepts are applied within these modules and fuzzy information is processed
through the control chain. Results show that using only II rules, all compo
nents of quality are controlled with accuracy and robustness. This is parti
cularly interesting in the light of the complexity of the process. Such a m
odular approach makes it possible to follow the evolution of quality during
the process, in successive steps which are easily recorded and interpreted
by experts. This approach could be envisaged as an operator decision suppo
rt system.