茶多酚热超声联合处理对冬瓜沙门氏菌数量的影响及其自适应神经网络生长模型的构建

Effect of tea polyphenols combined with thermosonication on the quantity of Salmonella in fresh-cut winter melon and construction of growth model using adaptive neuro-fuzzy inference system

鲜切蔬菜易受到食源致病菌的污染,然而传统的高温杀菌技术在鲜切蔬菜中的应用受到很大限制。通过研究热超声技术和茶多酚联合处理对鲜切冬瓜在不同温度储藏过程中沙门氏菌菌落数量的影响,以及自适应神经模糊推理系统(adaptive neuro-fuzzy inference system,ANFIS),构建了沙门氏菌菌落数量的预测模型。热超声和茶多酚联合处理显著减少了鲜切冬瓜中沙门氏菌的数量约2.20 lg CFU/g(P<0.05);在25、10和4℃温度下储藏时,不同温度下菌落数量变化趋势不同,但处理组的菌落数量均显著低于对照组(P<0.05),表明热超声和茶多酚联合处理是一种具有较大应用潜力的鲜切蔬菜保鲜技术。研究所建立的ANFIS模型,每个输入变量采用2个隶属函数,生成8个因果规则;采用gaussmf隶属函数的ANFIS模型对储藏过程中菌落数量的预测结果最好,其R^2大于0.99,均方根误差小于0.2。

Fresh-cut vegetables are exposed to the risk of Salmonella contamination. Effective non-thermal sterilization methods and early warning systems play important roles in ensuring food safety of fresh-cut products. The effects of thermosonication(TS) combined with tea polyphonels(TP) on the population of Salmonella enterica in fresh-cut wax gourd during storage at 25, 10 and 4 °C were evaluated. Adaptive neuro-fuzzy inference system(ANFIS) based models with 8 types of membership functions(MFs) were developed and estimated for predicting S. enterica population. The combined treatment of TS and TP significantly reduced the Salmonella population in fresh-cut melon by about 2.20 lg CFU/g. During storage at 25 ℃, 10 ℃ and 4 ℃, samples in treatment group had lower Salmonella population levels than that in control group. ANFIS model with 2 ‘gaussmf’ type input MFs and 8 if-then rules had the best performance at both training and prediction phases in present study. The root mean squared errors(RMSEs) were less than 0.2 and the R^2 for experimental populations and predicted data were greater than 0.99, indicating that ANFIS system performed well in predicting the population of Salmonella in fresh-cut wax gourd. It could be easily applied to various situations and provide a powerful tool for modelling and predicting microbe population and shelf life of food products.