基于嗅觉可视化的鲳鱼TVB-N预测模型  

Total volatile basic nitrogen prediction models of Pampus argenteus based on olfaction visualization detection technique

作　　者：黄星奕[1] 蒋飞燕[1] 穆丽君[1] 孙宗保[1] 

机构地区：[1]江苏大学食品与生物工程学院,江苏镇江212013

出　　处：《江苏大学学报（自然科学版）》2013年第1期38-43,共6页Journal of Jiangsu University：Natural Science Edition

基　　金：国家自然科学基金资助项目(31071549);公益性行业(农业)科研专项经费资助项目(201003008-04);高等学校博士学科点专项科研基金资助项目(20093227110007);江苏省青蓝工程(2008)项目;江苏省高校优势学科建设工程项目

摘　　要：采用半微量凯氏定氮法对4℃恒温下储藏的冰鲜鲳鱼的TVB-N进行检测,同时利用嗅觉可视化技术提取其挥发性气味信息,将嗅觉可视化传感器阵列与鲳鱼挥发性气体发生反应前后的RGB颜色变化值作为传感器响应值,分别建立TVB-N的偏最小二乘法(PLS)、遗传偏最小二乘法(GA-PLS)预测模型.经比较,GA-PLS模型预测效果更优,预测集相关系数达到0.851 7;可视化传感器阵列响应信号与TVB-N之间存在较高的相关性,可以快速预测出鱼储藏期间TVB-N变化,从而无损评价鱼的新鲜度.TVB-N contents of Parnpus argenteus samples stored in refrigerator at constant temperature of 4 ~C were measured by semimicro-fixation of nitrogen method. The headspace volatile gases of Pampus argenteus samples were extracted by olfaction visualization detection system. The RGB value changes of the colorimetric sensor array before and after exposure to the headspace gas were obtained as the sensor array responses. Partial least squares（PLS） model and genetic algorithm-partial least square（GA-PIS） model were established and compared. The GA-PLS model is good with the correlation coefficient（R~,） between actual value and predicted TVB-N content of 0. 851 7 for the validation dataset. There exists a good correlation between the eolorimetric sensor array responses and TVB-N content. The olfaction visua- lization detection technique can be used as a rapid way to predict TVB-N content and to evaluate fish freshness with non-destructive test.

关 键 词：鲳鱼 嗅觉可视化 检测 挥发性盐基氮 遗传-偏最小二乘法 

分 类 号：S126[农业科学—农业基础科学] S983