Optimizing Optical Alignment and Repeatability in SBH Spectroscopy Using a Custom Mechanical Module

Abstract

Stingless bee honey (SBH) has gained prominence in the Malaysian honey industry because of its distinct chemical composition and bioactive properties. However, the high market value of SBH makes it particularly vulnerable to economic adulteration, raising concerns over product authenticity among consumers. Adulteration in SBH is challenging to detect because conventional detection methods are often time-consuming, laboratory-bound and unsuitable for on-site screening. This study aims to evaluate how a portable Visible Near-Infrared (Vis-NIR) spectroscopic system, REVA INVITRO, can be applied for the detection of adulteration in SBH, with emphasis on the role of mechanical stabilization. The system is equipped with a custom-designed mechanical jig that holds cuvette in place, ensuring a consistent optical path length between Vis-NIR sensor and the sample. In this work, the effectiveness of REVA INVITRO was evaluated by analyzing spectral information of SBH samples contaminated with three widely used substances, distilled water (DW), apple cider vinegar (ACV), and fructose syrup (FS), at varying concentrations level. Spectral measurements acquired with and without mechanical jig were compared to assess the contribution of mechanical stabilization. Spectral trends were examined qualitatively and quantitatively to evaluate the measurements reproducibility and sensitivity. Spectral measurements obtained from using the jig achieved a classification model of 100% accuracy on the evaluation dataset, with a mean five-fold cross validation accuracy of 94.55%, indicating its robustness and consistent performance. In contrast, the classification accuracy dropped to 89.39% when measurements were acquired without using jig. For the regression models, the coefficient of determination (R²) for honey content quantification across all adulterant type ranged from 0.995 to 0.999, with predictions error between 0.34% and 1.96%. The jig improved the regression performance by reducing the prediction error (RMSE) by approximately 45-79% across all adulterant types. These findings showed a significant improvement in spectral uniformity and dependability with use of the jig. The jig integration enables the REVA INVITRO system to deliver a stable measurement suitable for SBH analysis. Overall, the ability of the system to detect adulteration trends both visually and quantitatively supports its potential as a portable tool for on-site honey quality assessment and authentication.