论文部分内容阅读
为实现鲫鱼新鲜度的快速测定,本文基于近红外漫反射光谱定量分析技术和化学计量学方法,采集了144个鲫鱼鱼肉样品在1000~1799 nm范围内的光谱数据,测定了鲫鱼样品的p H、TVB-N含量、TBA含量和K值四种新鲜度指标;在确定近红外光谱数据最佳预处理方法和适宜波段的基础上,分别采用偏最小二乘法、主成分分析和BP人工神经网络技术、偏最小二乘法和BP人工神经网络技术建立了鲫鱼新鲜度定量预测模型。结果表明,鲫鱼样品四种指标数据范围均较大,可满足建模要求。以p H为鲜度指标时,采用偏最小二乘法和BP人工神经网络技术建立的模型最好,其定标相关系数为0.9945;以TVB-N、TBA和K值为鲜度指标时,采用偏最小二乘法建立的模型最好,其定标相关系数分别为0.9857、0.9985和0.9952。建立的四种鲜度指标定量模型均具有较好的预测能力。
In order to determine the freshness of Carassius auratus, the spectroscopic data of 144 fish samples of crucian carp in the range of 1000 ~ 1799 nm were collected based on quantitative analysis of near-infrared diffuse reflectance spectroscopy and chemometrics methods. The p H , TVB-N content, TBA content and K value. On the basis of determining the best pretreatment method and appropriate band of near infrared spectroscopy data, partial least square method, principal component analysis and BP artificial neural network Technology, partial least squares and BP artificial neural network technology to establish a quantitative prediction model of freshness of the carp. The results showed that the four indicators of crucian carp samples range are large, to meet the modeling requirements. When p H is the freshness index, the model established by partial least square method and BP artificial neural network is the best, and the calibration coefficient is 0.9945. When TVB-N, TBA and K are the freshness indexes, Partial least squares method to establish the best model, the calibration correlation coefficients were 0.9857,0.9985 and 0.9952. The established four quantitative indicators of freshness indicators have better predictive ability.