2017 Volume 53 Issue 4 Pages 107-115
This study focused on improving quality control of marine products and their distribution. Discrimination models were designed and calibrated based on the quality assignments made by fish market auctioneers. Eight kinds of fish caught in the waters off Yamaguchi Prefecture were graded into five classes by experienced auctioneers. By statistically analyzing the results of the auctioneers’ evaluations and the relationship between the color of fish skin and the freshness of fish meat, it was found that four-color indexes for each type of fish species accurately reflected the auctioneers’ evaluations and the degree of meat freshness. Fuzzy inference models constructed from these sets of indexes, as antecedent variables of fuzzy rules, were evaluated. The results of the model simulations and expert evaluations revealed that the models were robust, agreeing with the quality assignments of the auctioneers by more than 84.6%.
