2018 Volume 30 Issue 1 Pages 509-516
This research intends to improve quality control while maintaining the skill level of distributers of marine products by designing models which ensure accurate non-destructive estimates of the freshness of fish meat (K value) by using fuzzy inference. A total of 240 sample fish of the genus Seriola (Japanese amberjack, greater amberjack, and yellowtail amberjack) were used to construct the models. Relationships between fish coloration and K value from sample acquisition until 72 hours later under refrigeration at -2°C, +2°C, and +6°C were investigated. By analyzing the results of the relationships between fish coloration and K value statistically, it was found that a combination of the body surface color indexes within 7 reflected well the degree of freshness of fish meat. The fuzzy inference models constructed from these index sets, as antecedent-part variables, were evaluated. The results of both simulation and experimental evaluations demonstrate that the models are robust, with the residuals of the K value indicating an accuracy within 9.34%. Therefore, the models were shown to be highly useful for quality control in the distribution of fresh fish.
