鲜猪肉水分近红外光谱在线检测方法研究

本研究目的在于应用近红外光谱技术对冷鲜猪肉深层水分含量进行在线检测方法研究。根据稳态空间分辨光谱技术设计了具有多个检测器的传感器,以750nm、805nm、850nm和970nm的LED作为光源,检测器距离为30-48mm之间,用于检测肉块样品一定深度和范围内的水分含量,从而为无损、快速地给出鲜猪肉水分含量提供一种可靠方法。实验中针对猪背最长肌进行检测,获得184个样品,其水分含量在71%-78%之间,采用多元校正预测模型计算肉块水分含量。所建模型相关系数（r）为0.783,呈显著相关,建模集标准残差（RMSEP）为0.988,验证集标准残差（RMSEP）为1.037。由此得出结论,在短波近红外范围内,利用稳态空间分辨光谱技术能够对猪肉水分含量进行在线检测。

Online Nondestructive Determination of Moisture Content in Intact Fresh Pork by Near-Infrared Spectroscopy

Pork is the main meat for the consumers in China.The moisture content is one of the most important indicators of the quality of meat,and it will affect the level of texture and nutritional value of the meat.Comparing other moisture content detecting methods,near infrared spectral technology is effective and fast by using chemometrics.Although many researchers have tried to detecting and modeling the minced pork moisture content using this technology,by now there is few report about fresh inta ct meat moisture.The objective of this study is to determine the moisture content of fresh intact pork and establish a fast non-invasive method using near-infrared spectroscopy.In this study,we designed a portable moisture measuring instrument.It includes data acquisition module and optical probe.The optic probe was designed using three detectors,four light sources based on Steady spatially-resolved spectroscopy（SRS）.The probe consists of three photodiodes and four LEDs at different wavelength,i.e.750 nm,805nm,850 nm and 970 nm.To accurately detect t he moisture,the space at cen ter-to-center distances between the light source and the detector should be set rationally.As the source–detector separation increases,greater tissue depth is probed.Here,source–detector separation was set between 3 0 to 48 mm to ensure the depth of detection for the pork.Before experiments,the instrument linearity and stability were verifi ed by using a solid-model which optical properties was known.In the experiments,184 longissimus muscle samples were used,and the moisture content was between 71%-78%.The samples were divided into a calibration set and a validation set.149 samples were selected to be the calibration set to establ ish the quantitative analytical model for the experiment.Multivariate calibration model was used to predict meat moisture content.The results showed that the model correlation coefficient（r） was 0.783,which was significantly correlated.Modeling sets RMSEP = 0.988,validation set standard residuals（RMSEP） was