A FUZZY SYMBOLIC MODEL FOR FOOD BROWNING MEASUREMENT

摘要

In food industry, controlling sensory properties on the manufacturing line represents a key issue for companies, because the sensory properties condition the choice and the preference of the consumers. A relevant control of a sensory property depends on the reliability of the measurements of the sensory variables implied in the perception of the sensory property. The browning variable is an important sensory variable, assessed by an operator, to control the visual quality of the browned products. In this paper, we propose to develop a model ensuring a reliable measurement of the browning variable by the use of the fuzzy symbolic approach. The browning variable is decomposed in descriptive variables, which are formalized by three sensory indicators: the area and colour of the spots on the product surface, and the colour of the area without spots. These variables are assessed by the operators with repeatability indexes superior to 90%. The decision support system has been validated on a database with a percentage of compatibility of 92% at a sensibility level of half a symbol.