Optimization of Canned Talang Queenfish Color Sterilized by Rotary Retort: Storage Stability, Artificial Intelligence—Adaptive Neuro Fuzzy Inference Systems Modeling <scp>TBA</scp> Based on Color Attributes

This study aimed to improve the color quality of sterilized Talang Queenfish. It also optimized sterilization conditions using Response Surface Methodology (RSM) with a central composite design. The effect of storage on color attributes was evaluated. Sterilization temperatures tested were 110°C, 115°C, and 121°C, with durations of 10, 20, and 30 min. Two methods were compared: rotary sterilizer (RS) and conventional autoclave (CA). Color parameters measured included <i>L</i>*, <i>a</i>*, <i>b</i>*, Δ<i>E</i>, hue angle (<i>h</i>), chroma (<i>C</i>), whiteness index (WI), yellowness index (YI), and browning index (BI). The Adaptive Neuro-Fuzzy Inference System (ANFIS) was used to model thiobarbituric acid-reactive substances (TBA). Optimal conditions were 116°C for 15.54 min. Under these conditions, the RS method improved <i>L</i>*, <i>a</i>*, Δ<i>E</i>, <i>C</i>, <i>h</i>, and WI compared to CA. These values reached 38.77, 31.25, 64.47, 0.16, 45.32, and 39.29, respectively. In contrast, YI and BI decreased by 27.94% and 32.82%. The RS method significantly enhanced color attributes immediately and during storage. ANFIS with a gaussmf membership function accurately predicted TBA. The sixth reduced model achieved the lowest prediction error. These results indicate that color parameters can be used as quality indicators in fish processing.