激光与光电子学进展, 2021, 58 (12): 1230001, 网络出版: 2021-06-23
基于近红外光谱的淡水鱼贮藏期质构品质的无损检测模型 下载: 574次
Nondestructive Testing Model for Textural Quality of Freshwater Fish in Storage Using Near-Infrared Spectroscopy
光谱学 近红外光谱 无损检测 淡水鱼 质构品质 连续投影算法 偏最小二乘回归 spectroscopy near-infrared spectrum non-destructive testing freshwater fish textural quality successive projections algorithm partial least squares regression
摘要
为了研究贮藏期与淡水鱼鱼肉质构品质的相关关系,以武昌鱼为研究对象,建立了近红外光谱淡水鱼鱼肉质构品质的快速无损检测模型。利用AntarisⅡ傅里叶变换近红外光谱仪采集武昌鱼鱼肉样本的光谱数据,并使用TMS-PRO型质构仪测量样本的硬度值、弹性值和咀嚼性值;采用S-G平滑法对原始光谱进行预处理,结合竞争性自适应重加权(CARS)算法、稳定性竞争自适应重加权采样(SCARS)算法和连续投影(SPA)算法进行一次特征波长的提取。基于上述3项质构指标建立的最小偏二乘回归(PLSR)模型,在一次特征波长提取的基础上再结合SPA算法进行二次特征波长的提取,根据二次特征提取的最优波长建立淡水鱼鱼肉硬度、弹性和咀嚼性的最优模型,该模型校正集的相关系数Rc分别为0.968、0.947、0.927,预测集的相关系数Rp分别为0.964、0.939、0.926,校正集的均方根误差RMSEC分别为0.753、0.827、0.986,预测集的均方根误差RMSEP分别为0.846、0.897、0.964。研究结果表明,该方法适用于淡水鱼鱼肉贮藏期质构品质的快速无损检测,具有较高的准确度,可为后续淡水鱼鱼肉品质的在线检测提供帮助。
Abstract
In this study, a rapid nondestructive testing model for the textural quality of freshwater fish using near-infrared spectroscopy is developed to study the relationship between storage periods and the textural quality of freshwater fish. Spectral data for the Parabramis pekinensis fish samples were collected using the Antaris Ⅱ Fourier transform near-infrared spectrometer, and the hardness, springiness, and chewiness values of the samples were measured using the TMS PRO type structure instrument. The S-G smoothing method was used to pretreat the raw spectra, and competitive adaptive reweighting sampling, stable competitive adaptive reweighted sampling, and successive projections algorithms were integrated to extract the characteristic wavelength for the first time. Based on the above three textural indexes,the least partial square regression (PLSR) model was established. Based on the primary characteristic wavelengths extraction, the SPA algorithm was used to extract the secondary characteristic wavelengths. Then, the optimal model of hardness, springiness and chewiness of freshwater fish was established according to the extracted secondary characteristic wavelengths. The correlation coefficients Rc and Rp of the correction and prediction sets are 0.968, 0.947, and 0.927, and 0.964, 0.939, and 0.926, respectively. The root mean square error of the correction and prediction sets are 0.753, 0.827, and 0.986, and 0.846, 0.897, and 0.964, respectively. The results show that this method is suitable for rapid and nondestructive testing of the textural quality of freshwater fish in storage and has high accuracy.
陈远哲, 王巧华, 高升, 梅璐. 基于近红外光谱的淡水鱼贮藏期质构品质的无损检测模型[J]. 激光与光电子学进展, 2021, 58(12): 1230001. Yuanzhe Chen, Qiaohua Wang, Sheng Gao, Lu Mei. Nondestructive Testing Model for Textural Quality of Freshwater Fish in Storage Using Near-Infrared Spectroscopy[J]. Laser & Optoelectronics Progress, 2021, 58(12): 1230001.