食品与发酵工业2026，Vol.52Issue(6)：211-221,11.DOI:10.13995/j.cnki.11-1802/ts.043414
基于响应面法和人工神经网络优化百香果豆清饮料发酵工艺
Optimization of fermentation process of passion fruit soy whey beverage based on response surface methodology and artificial neural network
摘要
Abstract
To prepare a fermented soy whey beverage with good sensory quality and increase the added value of soy whey,this study used soy whey as the raw material,blended with passion fruit juice,and fermented using lactic acid bacteria.Using viable bacterial count and sensory score as comprehensive evaluation indicators,the optimal fermentation process for the passion fruit soy whey beverage was constructed and optimized by combining response surface methodology(RSM)experiments with artificial neural network-genetic algorithm(ANN-GA).Gas chromatography-ion mobility spectrometry(GC-IMS)and sensory evaluation were employed to study the changes in sen-sory flavor characteristics of the passion fruit soy whey beverage during fermentation under the optimal process conditions.Results showed that the optimal fermentation process was:passion fruit juice addition at 20％,lactic acid bacteria inoculum at 0.5％,fermentation time of 18 h,fermentation temperature of 35℃,initial pH of 4.1,and soluble solids content of 13 °Brix,soy whey beverage achieving a high sensory score of 89.98.GC-IMS identified a total of 53 volatile compounds at different fermentation times,primarily alcohols and esters.As fermentation progressed,the content of aroma compounds gradually increased.This promoted the formation of alcohols,terpenes,ketones,and esters,which contributed desirable flavors such as fruity,sweet,and wine-like notes,while inhibiting the production of off-flavors.The fruity aroma compounds were most abundant at 18 hours of fermentation.However,over-fermentation(24 h)led to deterioration of the beverage's flavor.Orthogonal partial least squares discriminant analysis(OPLS-DA)identified 17 key aroma compounds as markers dis-tinguishing the flavor characteristics of the passion fruit soy whey beverage at different fermentation stages.This study confirmed that the synergy of RSM and ANN-GA overcame the limitations of single-process optimization and significantly enhanced the sensory flavor of the fermented beverage.It provides a scientific theoretical basis for accelerating the comprehensive utilization of soy whey and developing high-quality lactic acid bacteria fermented beverages.关键词
豆清液/乳酸菌/响应面/人工神经网络-遗传算法/感官风味Key words
soy whey/lactic acid bacteria/response surface/artificial neural network-genetic algorithm(ANN-GA)/sensory-flavor characteristics引用本文复制引用
胡晓婷,张胤,周李波,陈浩,赵聃,黄建芳,唐睿,张志昊,黄展锐,文明,苏伟明..基于响应面法和人工神经网络优化百香果豆清饮料发酵工艺[J].食品与发酵工业,2026,52(6):211-221,11.基金项目
邵阳学院研究生科研创新项目资助(CX2024SY018) （CX2024SY018）
邵阳学院大学生创新创业训练计划资助项目(S202410547005) （S202410547005）
邵阳市科技计划项目(2024GZ1001,2024PT6093,2024PT6091) （2024GZ1001,2024PT6093,2024PT6091）
湖南省自然科学基金资助项目(2023 JJ50256,2025 JJ70243,2023 JJ50260) （2023 JJ50256,2025 JJ70243,2023 JJ50260）
洞庭实验室专项经费项目(2024-DTKF-009) （2024-DTKF-009）
湖南省科技创新计划资助(2019TP1028,2020TP2002) （2019TP1028,2020TP2002）
湖南省教育厅科学研究项目(24C0397) （24C0397）
