猪肉pH值的可见近红外光谱在线检测研究

pH值是猪肉关键品质之一，实施在线检测对优化肉品加工工艺、保证产品质量、提高肉及肉制品的经济价值有重要意义。研究应用可见近红外光谱对新鲜猪肉pH值进行在线检测，实验时样品以0.25 m·s-1的速度运动，采集其可见近红外漫反射光谱(350～1 000 nm)，进行反射距离校正后应用偏最小二乘回归法建立猪肉pH值在线检测模型。研究通过Kennard-stone算法划分样品校正集与预测集，对比了不同的光谱预处理方法(多元散射校正，微分等)对预测结果的影响，并对建模所用光谱变量进行优化。研究发现经过多元散射校正结合一阶微分预处理的模型效果最好，模型预测相关系数为0.905，预测均方根误差为0.051，经过优化的模型建模所用波长变量数减少一半，模型的预测相关系数提高到0.926，预测均方根误差下降至0.045。结果表明可见近红外光谱可用于新鲜猪肉pH值的在线检测。

Online Determination of pH in Fresh Pork by Visible/Near-Infrared Spectroscopy

The present research was focused on determination of the pH value online by visible and near-infrared spectroscopy.In the part of data gathering,fresh pork longissimus dorsi was moving at the constant velocity of 0.25 m·s~(-1) on the conveyor belt,and the visible and near-infrared diffuse reflectance spectrum (350-1 000 nm) was captured.In the part of data processing,band of 510-980 nm of the spectra was chosen to calibrate reflex distance,then to set up online detection model of pH value in fresh pork by partial least squares regression (PLSR).Kennard-stone algorithm was applied to divide the samples to the calibration set and validation set.The performances of several PLSR models employing various preprocessing methods including multiple scatter correction,derivative and both of them combined were compared.Further,the best performance model was optimized by interval PLSR to decrease the modeling variables of wavelength.The results indicated that the PLSR model based on preprocessing of multiple scatter correction (MSC) combined with first derivative gave the best performance with 0.905 of the correlation coefficient for validation set and 0.051 of the root of mean square errors for validation set.For the best PLSR model performance,the correlation coefficient of validation set increased to 0.926 and the root of mean square errors for validation set to 0.045 in the optimization interval PLSR model.However,only half of variables were used.The research demonstrates that using visible and near-infrared spectroscopy to determine fresh pork pH online is feasible.