How human expertise at industrial scale and experiments can be combined to improve food process knowledge and control

The knowledge available on food processes stems from various sources: physical models, data obtained from laboratory experiments, and the expertise of operators involved in controlling of production lines. The aim of the study was to analyse various ways to associate different sources of knowledge to improve process control. The analysis was based on a real example: the manufacturing of “lady-finger” biscuits. The paper first summarized the available bibliographic knowledge. The methodology of the study was based on two parallel approaches: expertise was collected from operators in charge of the control of a continuous industrial line, and experiments were performed on a batch process at laboratory scale. The two approaches were compared in terms of aims, measured product properties and control variables. Four ways to associate human expertise extracted at the process level with laboratory results were presented and discussed: (1) How can at-line sensory measurements be used to design a measurement strategy? (2) Can batch control be improved by extracting control rules from an experimental approach? (3) Can a reverse-engineering approach be developed? and (4) Can a “kernel” of knowledge be extracted?