The aim of this study was the optimization of supercritical fluid extraction (SFE) of raspberry seed oil. Sequential extraction kinetic modelling and artificial neural networks (ANN) were used for this purpose. SFE was performed according to the broaden Box-Behnken experimental design with pressure, temperature and CO₂ flow rate as independent variables, while the influence of particle size on extraction kinetics and adjustable model parameters was additionally evaluated. Five empirical kinetic equations and mass-transfer model proposed by Sovová were utilized for extraction kinetics modelling. According to appropriate statistical parameters (R², SSE and AARD), the mass-transfer model exhibited the best fit of experimental data. The initial mass-transfer rate of extraction curve was used as a response variable in ANN optimization. SFE should be performed at elevated pressure and CO₂ flow rate, while temperature and particle size should be held at a lower level in order to achieve a maximal initial mass-transfer rate.

这项研究的目的是优化覆盆子种子油的超临界流体萃取（SFE）。为此，采用了顺序提取动力学建模和人工神经网络（ANN）。SFE根据扩大的Box-Behnken实验设计进行，压力，温度和CO ₂流量为自变量，同时还评估了粒径对萃取动力学和可调模型参数的影响。索沃瓦提出的五个经验动力学方程和传质模型被用于萃取动力学建模。根据适当的统计参数（R ²，SSE和AARD），则传质模型表现出最适合的实验数据。提取曲线的初始传质速率用作ANN优化中的响应变量。SFE应在升高的压力和CO ₂流速下进行，而温度和粒度应保持在较低水平，以实现最大的初始传质速率。