Modeling of bleaching process of camellia oil using multiple surrogate models

Abstract

The objective of this article is to study the effect of bleaching conditions on the quality of camellia oil. To achieve this goal, the relationship between bleaching conditions (e.g., heating temperature, stirring speed, stirring time, and adsorbent content) and quality indexes are established using four surrogate models. The performance of different surrogate models is then compared, and the best one is chosen to analyze the influence of process parameters on quality. Additionally, the global sensitive analysis (GSA) is conducted to determine the relative importance of the four process parameters on oil quality. The results show: the RSM yields the best modeling accuracy for bleaching rate (BR), while SVR performs best for peroxide value (PV); Activated clay is the most important factor affecting both BR and PV, followed by heating temperature, stirring time, and stirring speed; The BR of 96.33% and PV of 0.03 meq/kg are obtained under the optimum bleaching conditions (temperature of 80 °C, stirring time of 10 rmp, stirring speed of 60 rpm, and adsorbent content of 7 g/hg), which is validated by experiment under the same conditions.