新疆羊肉水分近红外光谱检测方法研究

发布时间：2018-04-15 10:35

本文选题：羊肉水分 + 近红外光谱　；参考：《塔里木大学》2017年硕士论文

【摘要】：羊肉水分含量直接影响着生鲜肉加工、储藏、贸易与食用,因此对其水分含量的检测具有十分重要的意义。传统的干燥箱干燥法测定水分的过程较为复杂、耗时较长,须在测定前对样品特殊处理,较难实现大批量样品的快速检测。近红外光谱分析技术具有方便、快捷、准确、无损等优点,可以快速采集羊肉样本的光谱数据。利用近红外光谱技术检测肉类品质已经成为近年来的研究热点。本论文以新疆绵羊鲜羊肉为研究对象,以羊肉的水分含量作为检测指标,使用便携式近红外光谱仪采集波段范围在900-1700nm内的135个羊肉样本光谱数据,通过多元散射校正、小波分析等预处理方法对原始光谱进行预处理分析,研究羊肉水分含量的近红外光谱响应,并提取特征参数,通过多元线性回归和偏最小二乘法分别建立羊肉水分含量的检测模型,并针对模型的优劣进行比较,为实现羊肉品质的快速、无损检测提供技术支持。主要研究结果如下:1)对羊肉样本进行水分含量检测,获取原始光谱数据,经多元散射校正、小波变换等预处理方法对原始光谱预处理,最终得到羊肉水分含量所建模型的最佳预处理方法为小波变换法。2)利用相关系数特征选择方法对近红外光谱数据进行降维,选取水分含量的六个特征波长(932nm、985nm、1035nm、1140nm、1272nm、1540nm),用于后续检测模型的建立。3)采用偏最小二乘法建立模型稍优于多元线性回归模型及直接用原始光谱所建模型,整体来看采用偏最小二乘法建模效果最好,相关系数为0.861,均方根误差为0.446。
[Abstract]:The moisture content of mutton directly affects the processing, storage, trade and consumption of fresh meat, so it is of great significance to detect the moisture content of mutton.The traditional drying method of drying box is more complicated and time-consuming, so it is difficult to realize the rapid detection of large quantities of samples because of the special treatment of the samples before the determination.Near-infrared spectroscopy (NIR) has the advantages of convenience, rapidity, accuracy and nondestructive, so it can quickly collect the spectral data of mutton samples.Near-infrared spectroscopy (NIR) has become a hot research area in meat quality detection in recent years.In this paper, Xinjiang sheep fresh mutton was used as the research object, and the moisture content of mutton was used as the detection index. The spectral data of 135 mutton samples in 900-1700nm were collected by using a portable near infrared spectrometer and corrected by multivariate scattering.Wavelet analysis and other pretreatment methods were used to pretreat the original spectrum to study the near infrared spectral response of the moisture content of mutton, and to extract the characteristic parameters.The detection models of mutton moisture content were established by multiple linear regression and partial least square method, and the advantages and disadvantages of the model were compared to provide technical support for fast and nondestructive testing of mutton quality.The main results are as follows: (1) the moisture content of the mutton sample is detected, and the original spectral data is obtained. The original spectrum is pretreated by multivariate scattering correction, wavelet transform and other preprocessing methods.Finally, the best pretreatment method of the water content model of mutton is wavelet transform method. 2) using correlation coefficient feature selection method to reduce the dimension of near-infrared spectrum data.The six characteristic wavelengths of water content (932 nm ~ 985 nm ~ 1 035 nm ~ (1 140) nm ~ (1 140) nm ~ (1)) are used to establish a model for subsequent detection. 3) the partial least square method is better than the multivariate linear regression model and the model built by the original spectrum directly.On the whole, the partial least square method is the best, the correlation coefficient is 0.861, the root mean square error is 0.446.
【学位授予单位】：塔里木大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TS251.53;O657.33

【相似文献】