【作者】 杨兰；

【导师】 赵谋明；

【作者基本信息】 华南理工大学 ， 食品科学， 2010， 博士

【摘要】 酱油是我国历史悠久、深受老百姓喜爱的传统调味品,产量位居世界第一,但目前我国酱油的品质和人均消费水平尚不令人满意。本文以广东的高盐稀态酱油为研究对象,从菌种筛选、营养源、原料的预处理和成曲制作过程中促发酵肽等对蛋白酶的诱导合成作用及神经网络对酱油制作的非稳态系统进行建模、预测,并探讨酱醪后发酵过程的变化规律。首先,从来自不同酱醪和曲精的样本中,采用马丁氏培养基进行分离、纯化,采用察氏培养基进行培养、菌落形态观察和个体形态观察。同时,考虑到主要来自酱醪的高盐环境或曲精,初步鉴定所获得的8个菌株为曲霉属的米曲霉菌。进一步用酪蛋白平板培养基进行初筛,结合种曲孢子数、酶活进行复筛,并传代6次,得到稳定性状的菌株,获得了本文研究用的菌株。为了获得性能良好的、制作酱油用的种曲,对影响米曲霉种曲的营养源进行了研究。结果表明,葡萄糖、豆粕、ZnSO₄及CaCl₂的影响较大。采用SPSS的Conjoint模块进行正交试验,得到葡萄糖1.00%、CaCl₂ 0.10%、ZnSO₄ 0.05%及豆粕3.00%的营养源优化配方。其中方差分析表明,豆粕质量分数为显著影响因子。同时,优化后的配方种曲产孢子数达157.7亿个/g干重,比对照组提高了84.87%,而种曲蛋白酶活、成曲蛋白酶活的提高幅度分别为25.04%、17.04%,表明外加营养源对种曲质量和成曲蛋白酶活均有较为显著的影响。成曲的制作是酱油发酵的关键。其中蛋白质原料的热变性对熟料消化率影响的研究表明,以126℃、6min的大豆蒸煮条件为宜。考虑在成曲制作中添加外加有机氮源作为酶的诱导物,从而达到提高成曲酶活的目的。在四种被考察的外加有机氮源中,以酵母抽提物和豆粕的效果较好。从经济和方便的角度出发,豆粕被用作下一步的研究对象。经过制作豆粕曲、酶解、超滤、冷冻干燥等工艺后,获得分子量小于3000Da的促发酵肽。用响应面分析软件对湿度、促发酵肽添加量、制曲时间的交互作用研究建立了数学模型。分析表明,两个模型的“Prob>F”值均小于0.05,模型显著,并且失拟检验表明模型与实验情况相拟合。两个模型中促发酵肽的“Prob>F”值分别为0.0221和0.0498,表明促发酵肽的添加量为显著影响因素。根据所建立的数学模型进行参数最佳化分析,取得酶活力最高所需的工艺条件为:时间45.10h、温度30.92℃、添加量0.46%,中性蛋白酶活为1887.29 U/g干重、酸性蛋白酶活为770.67 U/g干重。考虑到实际操作的可行性及简化工艺,修正后工艺为:时间45h、温度31℃、添加量0.45%,得中性蛋白酶活为1890.12 U/g干重、提高了近60%,酸性蛋白酶活776.55 U/g干重、提高了近57%,效果显著。研究结果表明,利用豆粕大曲培养过程中自身产的酶系、对豆粕进行可控酶解而获得分子量小于3000Da的促发酵肽,可调控米曲霉的代谢,对酱油成曲蛋白酶的生成具有明显的诱导和提高作用。高盐稀态酱油的后发酵过程,一直都存在时间长、难预测、因而难控制的特点。因此,BP网络被用于对酱油的后发酵过程进行建模,得到收敛精度为10-4并带有40个神经元的1个隐含层、结构为7-40-2的BP网络,其预测值和目标值的相关系数分别为0.9983和0.9953,验证了该网络具有很强的预测有效性。五个输入变量中,发酵天数、气温、酱醪pH值、中性蛋白酶活、酸性蛋白酶活对酱醪的总氮、氨氮含量的影响方式各不相同,且两个输出变量、即总氮和氨氮的相关性在0.7298到0.9916之间。另外,输入变量相对显著性的研究表明,发酵时间是最大的影响因子,其余输入变量对总氮的影响顺序是气温>酱醪pH值>中性蛋白酶活>酸性蛋白酶活,对氨氮的影响顺序为气温>酸性蛋白酶活>中性蛋白酶活>酱醪pH值,进一步证明酱醪中总氮和氨氮的形成是受多种因素的综合作用的结果。更多还原

【Abstract】 Soy sauce is a type of traditional condiments which is very liked by people with very long history. The quality and per capital consumption of soy sauce are anything but satisfactory although the soy sauce production in our country is very large. In the article, Cantonese soy sauce produced by liquid fermentation with high salt, is the objective of our study.After separation, purification, morphology and microbial populations and individual microbial, the Aspergillus oryzae strains were got. After screening with casein plate medium, comparison of spore numbers and protease activity and generation stability test, the strain L-4 was used in the following study.The study on nutrient source of seed koji revealed that, glucose, soybean meal, ZnSO₄, CaCl2 had great influence. After the orthogonal design of the formula, the best formula was obtained, that was, glucose 1.00%, CaCl20.10%, ZnSO₄ 0.05%, soybean meal 3.00%, and the content of soybean meal was the significant factor. With the formula, the spore number was increased by 84.87%, to 15.77 billion/g dry seed koji. At the same time, the protease activity of seed koji and soy sauce koji were increased by 25.05% and 17.40%, respectively.As for soy sauce koji, the pretreatment of soybean and inducing effects on protease activity of soy sauce koji were investigated. The cooking temperature of 126℃and cooking time of 6 min were determined by comparison of steamed soybean digestibility. And the soybean meal was adopted among four different proteins, which were yeast extract, peanut meal, soybean meal and gluten. After the koji-making of soybean meal, enzymic hydrolysis, ultrafiltration and freeze drying, the fermentation-promoting peptides with molecular weight less than 3000 Da were collected. Using response surface methodology, the best process condition was 31℃, 45h, 0.45%, with the result koji in which neutral and acidic protease activity were 1890.12U/g dry koji and 776.55U/g dry koji. The rates of increase reached nearly 60% and 57%, respectively. It revealed that the fermentation-promoting peptides with molecular weight less than 3000Da, obtained by controlled enaymic hydrolysis of soybean meal koji, exerted metabolic regulation on Aspergillus oryzae and induction effects on the production of protease.At last, the artificial neural network was used to predict the brewing state of soy sauce and its changing pattern with regard to total nitrogen （TN） andα-amino acidic nitrogen （AN）. After the 7-40-2 BP network with training precision of 10-4 and one hidden layer containing 40 neurons was determined statistically, the quite high R values of 0.9983 and 0.9953 between predicted and experimental data verified that the neural model was quite successful in predicting the experimental data points. Using the network, the effects of five input variables on two output variables were found in different way, and the R-values, which indicating the correlation between the TN and AN, ranged from 0.7298 to 0.9916. As for the relative significances of inputs, aging time is the key factor for both outputs during the moromi aging. As for other inputs, the order of importance was temperature > pH > neutral protease activity > acidic protease activity for TN, and temperature > acidic protease activity > neutral protease activity > pH for AN, respectively.更多还原