效应面法结合BP神经网络模型优化五冬果茶提取工艺OA

To optimize the extraction process of Wudong fruit tea,based on preliminary screening,the water-to-material ration and extraction time were selected as key independent variables,with dry extract yield,psoralen content,and sensory evaluation serving as dependent evaluation indices.A comparative analysis was conducted on the applicability of three weighting method(AHP,CRITIC,and AHP-CRITIC hybrid),the AHP-CRITIC hybrid weighting method was ultimately selected to determine the index weights.A star-point design combined with response surface methodology(RSM)was employed for optimization of the extraction process,followed by a backpropagation(BP)neural network model to predict and validate the optimal conditions.The optimal extraction parameters were determined as follows:the prescribed medicinal materials was extracted with boiled water at 13-fold amount twice,and 60 minutes of extraction time(per cycle,for two cycles),followed by filtration.The integration of the star-point design-RSM and BP neural network demonstrated robust stability and reproducibility in optimizing the extraction of Wudong fruit tea,providing a scientific foundation for the optimized method in future research.