张 娟

（烟台汽车工程职业学院电子系，山东烟台 265500）

摘要：基于样品集划分、特征波长选择、偏最小二乘法（PLS）等基本理论，利用近红外光谱技术对草莓糖度建立定量分析模型。首先，采用光谱-理化值共生距离算法（SPXY）将草莓样品集划分为40个校正集和15个预测集。其次，采用小波变换（WT）结合竞争性自适应重加权算法（CARS）对原始光谱进行分解和重构。最后，利用偏最小二乘法（PLS）建立草莓糖度预测模型。结果表明，SPXY样品集划分合理有效，有利于建立稳健的预测模型。小波变换能够有效剔除高频噪声干扰，重构得到的光谱特征波形轮廓清晰。PLS预测模型不仅能够提高模型预测精度和稳定度，而且还能降低建模变量和模型复杂度。该研究结果为实际生产中利用近红外光谱技术快速无损检测其它水果糖度提供了技术可行性。

关键词：草莓糖度；近红外光谱；SPXY算法；WT算法；CARS算法；PLS算法；特征波长

中图分类号：TS207.3   文献标识码：A   文章编号：1674-506X（2020）06-0136-0005

Research on the Detection of Strawberry Sugar Content by NIR Based on SPXY-WT-CARS Algorithm

ZHANG Juan

（Electronic Department, Automotive Engineering Vocational College, Yantai Shandong 265500, China）

Abstract：Based on the basic theories of sample set division, characteristic wavelength selection and partial least squares (PLS), the quantitative analysis model of strawberry sugar content is established by NIR. Firstly, the strawberry sample set is divided into 40 training sets and 15 prediction sets by SPXY. Secondly, wavelet transform (WT) and competitive adaptive reweighting algorithm (CARS) are used to decompose and reconstruct the original spectrum. Finally, the prediction model of strawberry sugar content is established by PLS algorithm. The results show that SPXY sample set partition is effective and reasonable, which is conducive to the establishment of a robust prediction model. Wavelet transform can effectively eliminate the interference of high-frequency noise, and the reconstructed spectral characteristic waveform is clear. PLS prediction model can not only improve the prediction accuracy and stability of the model, but also reduce the modeling variables and complexity. The research results provide a technical feasibility for the rapid and non-destructive detection of sugar content of other fruits by NIR spectroscopy in practical production.

Keywords：sugar content of strawberry; NIR spectroscopy; SPXY algorithm; WT algorithm; CARS algorithm; PLS algorithm; Characteristic wavelength

doi：10.3969/j.issn.1674-506X.2020.06-023

收稿日期：2020-04-29

基金项目：山东省高等学校科技计划项目（J17KB131）；烟台汽车工程职业学院科技项目（2020kj02）

作者简介：张娟（1982-），女，硕士，副教授。研究方向：光电弱信号获取与处理，农产品品质近红外无损检测研究。

引文格式：张娟.基于SPXY-WT-CARS算法的草莓糖度近红外光谱检测研究[J].食品与发酵科技,2020,56(6):136-139,142.