基于导数增强与子空间夹角法的甜味剂快速识别

Rapid Identification of Sweeteners Based on Derivative Enhancement and Subspace Angle Method

摘要:

为满足食品中甜味剂快速、精准定性的需求，解决传统分析方法前处理复杂、结构相似物识别率低的问题，本研究提出一种基于傅里叶变换红外光谱(FTIR)精细解析与向量-子空间夹角法的分析策略。以三氯蔗糖对照品光谱为示例，通过对比光谱累加，Savitzky-Golay(S-G)与模拟双曲卷积(SH)算法的平滑降噪、信噪比提升以及导数光谱分辨增强效果，确定光谱叠加结合SH2阶导数为最优预处理方案，据此构建了15种常见甜味剂特征增强基准谱库。引入向量-子空间夹角法作为相似度判别依据，设定夹角阈值θ<0.1，实现单一甜味剂定性分析，结合逐级扣除策略解析了复配体系，并应用于市售三氯蔗糖样品的品质评估。结果表明，该方法可100%准确识别单一甜味剂样本，复配甜味剂体系的组分解析结果与实际配比一致，市售三氯蔗糖样品的品质评估结果与HPLC分析结果吻合。该方法无需复杂前处理，兼具快速性与准确性，可为食品中甜味剂现场筛查及品质监管提供可靠技术支撑。

Abstract:

To address the need for rapid and precise qualitative identification of sweeteners in food products, a novel strategy was developed, integrating refined Fourier Transform Infrared (FTIR) spectral analysis with a vector-subspace angle approach. This methodology was designed to overcome the limitations of traditional analytical methods, such as cumbersome sample pretreatment and low discrimination rates for structurally similar compounds. Employing the sucralose reference spectrum as a case study, spectral point stacking was systematically compared with Savitzky-Golay (S-G) and Simulated Hyperbolic Convolution (SH) algorithms. The comparison revealed the superior performance of the SH algorithm in spectral smoothing, noise reduction, signal-to-noise ratio enhancement, and derivative spectral resolution. Consequently, spectral stacking combined with SH second-order derivatization was established as the optimal preprocessing protocol, enabling the construction of a feature-augmented reference spectral library encompassing 15 common sweeteners. The vector-subspace angle method was adopted as the similarity metric, with a discriminative threshold of θ < 0.1 established for single-sweetener qualitative analysis. This framework was further extended via a stepwise subtraction algorithm to deconvolute multicomponent systems and to evaluate the quality of commercial sucralose samples. The results demonstrated 100% accuracy in the identification of individual sweeteners, high concordance between resolved components in compounded systems and their actual formulations, and strong alignment of commercial sample quality assessments with findings from High-Performance Liquid Chromatography (HPLC). As no elaborate sample preparation is required, this method exhibits both expediency and precision, offering robust technical support for on-site sweetener screening and regulatory oversight in food safety.