本文已被：浏览 711次 下载 247次
投稿时间：2025-03-20
投稿时间：2025-03-20
中文摘要: 该文基于近红外光谱，建立快速检测腌制生食虾蛄新鲜度模型，并筛选出腌制生食虾蛄挥发性盐基氮（total volatile basic nitrogen，TVB-N）含量预测的最佳模型。采用一阶导数、多元散射校正（multiplicative scatter correction，MSC）、标准正态变换，通过组合预处理方法对近红外光谱数据进行处理，并运用偏最小二乘法（partial least squares，PLS）和卷积神经网络（convolutional neural network，CNN），建立定量预测TVB-N 模型。通过预测准确度、稳定性及对不同样本的适应性，进行多维度评估分析。结果表明，基于近红外光谱建立PLS 和CNN 定量预测模型，模型均可靠；利用MSC 对去壳虾仁光谱进行预处理后所构建的CNN 模型，展现出最佳的预测性能，训练集与测试集的相关系数分别为0.93、0.84。
Abstract:This study aims to establish a model for the rapid detection of the freshness of marinated raw mantis shrimps based on near-infrared spectroscopy and screen out the optimal model for predicting the content of total volatile basic nitrogen （TVB-N） in marinated raw mantis shrimps. First，first-order derivative，multiplicative scatter correction（MSC），and standard normal variate transformation were used. The near-infrared spectroscopy data were processed by a combination of pretreatment methods. Partial least squares （PLS） and convolutional neural network （CNN） were applied to establish a quantitative prediction model for TVB-N. A multidimensional evaluation and analysis were carried out based on prediction accuracy，stability，and adaptability to different samples. The results showed that the quantitative prediction models of PLS and CNN established based on near-infrared spectroscopy were both reliable. The CNN model constructed after preprocessing the spectra of peeled shrimps using MSC showed the best prediction performance，with the correlation coefficients of the training set and the test set being 0.93 and 0.84，respectively.
keywords: near-infrared spectroscopy total volatile basic nitrogen convolutional neural network partial least squares marinated raw mantis shrimp
文章编号：202514020 中图分类号： 文献标志码：
基金项目:国家市场监督管理总局科技计划项目（2023MK131）；大连市科技人才创新支持计划项目（2023RJ003）；2023 年辽宁省自然科学基金计划项目（2023-BS-200）
引用文本：
津公网安备12011602301139号