Predicting the quality of wheat flour dough at mixing using an expert system

Abstract

Modelling the links between the mixing process conditions of wheat flour dough and the properties of the dough is a challenge. This paper presents a systematic modelling approach based on qualitative algebra to represent human expertise in this domain. Qualitative models of wheat dough mixing have been implemented as an expert system, called Ascopain. The relations between the process conditions – flour specifications and kneading conditions – and the dough sensory properties, have been formalised by means of qualitative functions. An extensive evaluation of Ascopain is provided by comparing the simulation results, first to experts' predictions, and second, to experimental results of sensory evaluation of mixed dough properties. The good matching level proves the accuracy and the robustness of the expert-system and, overall, its ability to implement a reasoning on the influences of process conditions to predict actual dough properties, starting from ingredient characteristics.