基于可视传感器阵列的鳊鱼新鲜度评价模型研究

Evaluating model of Parabramis pekinensis’s freshness based on visual sensor arrays

中文关键词:  鳊鱼  可视传感器阵列  挥发性盐基氮  BP神经网络  联合间隔偏最小二乘法

英文关键词:Parabramis pekinensis  vsiual sensor arrays  total volatile basic nitrogen  BP neural networks  synergy interval partial least squares

中文摘要:

      目的 建立基于可视传感器阵列的鳊鱼新鲜度评价模型。方法 研究采用可视传感阵列与鱼体进行无接触式反应, 提取阵列反应前后的颜色变化信息来表征鱼的气味特征; 同时根据行业标准SC/T3032-2007测得表征鱼新鲜度的挥发性盐基氮(TVB-N)含量; 将可视传感技术所得的特征信息与TVB-N指标含量进行关联, 分别建立基于可视传感技术鱼新鲜度评价的定性模型BP神经网络和联合间隔偏最小二乘法(siPLS)定量模型。结果 BP神经网络模型精度较高, 训练集正确率为86.79%, 预测集正确率为86.43%; siPLS模型次之, 模型校正集和预测集的正确率分别为82.52%和80.67%。结论 可视传感器新技术所测得指标与TVB-N相关性较大, 可快速预测出鱼在储藏期间TVB-N的变化从而能够快速、无损地评价鱼类新鲜度。

英文摘要:

      Objective To establish evaluating models of Parabramis pekinensis’s freshness based on visual sensor arrays. Methods The visual sensor array was developed to react with fish odors without contact. The color changes of the sensor array before and after exposure to fish odors were used to represent the characteristic of fish’s odor. Meanwhile, the content of total volatile basic nitrogen (TVB-N) of fish samples were measured according to SC/T3032-2007 standard. The freshness evaluating models were set up including qualitative model BP neural networks and synergy interval partial least squares (siPLS) quantitative model based on the color information obtained from visual sensor array and TVB-N measurements. Results BP neural networks model was more precise, with correct rate of 86.43% for training set, 86.07% for prediction set. The correct rate of siPLS model for calibration set and prediction set was 82.52% and 80.67%, respectively. Conclusion The visual sensor array could rapidly predict TVB-N content in fish because of the tight relationship between detection values of sensor arrays and TVB-N contents. It is feasible to evaluate the freshness of fish rapidly and non-destructively in this way.