Laboratory of Non-Destructive Quality Evaluation of Commodities, Kasetsart Agricultural and Agro-Industrial Product Improvement Institute, Kasetsart University, Bangkok, Thailand
Improper fermentation of pineapple wine owing to volatile acidity has been associated with excessive proliferation of acetic acid bacteria during fermentation and consequent increased acetic acid concentration. Mid-infrared (MIR) and near-infrared (NIR) spectroscopies were employed to classify improper pineapple wine fermentation. Two clusters of samples possessing within the limit and over-limit acetic acid content were obtained using low-grade pineapples and prepared accordingly. Spectral data were collected for all samples in the 4000–650 cm−1 region using attenuated total reflection (ATR) with an FT-MIR spectrophotometer and in the 11,536–5800 cm−1 using sample vials and 11,536–3952 cm−1 regions using a liquid probe and a liquid cup with an FT-NIR spectrophotometer. The classification models for pineapple wine fermentation based on acetic acid content were constructed using soft independent modeling of class analogy (SIMCA) and partial least squares discriminant analysis (PLSDA). Comparisons of MIR and NIR techniques, classification methods, and spectral pretreatments have been reported. The results demonstrated that MIR spectroscopy coupled with ATR and PLSDA is highly effective for the detection of improper pineapple wine fermentation as a function of acetic acid content. The best classification model was generated using the entire MIR spectra after second derivatives transformation, which provided the highest accuracy, sensitivity, specificity, and precision.