A feature-enhanced approach based on joint domain alignment and multi-order derivative spectral reconstruction for predicting apple firmness using Vis-NIR spectroscopy

Food Chemistry, Год журнала: 2025, Номер 476, С. 143457 - 143457

Опубликована: Фев. 17, 2025

Язык: Английский

---

Identification of Aged Polypropylene with Machine Learning and Near–Infrared Spectroscopy for Improved Recycling

Polymers, Год журнала: 2025, Номер 17(5), С. 700 - 700

Опубликована: Март 6, 2025

The traditional plastic sorting process primarily relies on manual operations, which are inefficient, pose safety risks, and result in suboptimal separation efficiency for mixed waste plastics. Near-infrared (NIR) spectroscopy, with its rapid non-destructive analytical capabilities, presents a promising alternative. However, the analysis of NIR spectra is often complicated by overlapping peaks complex data patterns, limiting direct applicability. This study establishes comprehensive machine learning-based spectroscopy model to distinguish polypropylene (PP) at different aging stages. A dataset was collected from PP samples subjected seven simulated stages, followed construction classification analyze these spectral variations. confirmed using Fourier-transform infrared (FTIR). Mechanical property analysis, including tensile strength elongation break, revealed gradual decline prolonged aging. After 40 days accelerated aging, break dropped approximately 30%, retaining only about one-sixth original mechanical performance. Furthermore, various preprocessing methods were evaluated identify most effective technique. combination second derivative method linear -SVC achieved accuracy 99% precision 100%. demonstrates feasibility accurate identification thereby enhancing quality recycled plastics promoting automated, precise, sustainable recycling processes.

Язык: Английский

---

Transfer learning for predicting wood density of different tree species: calibration transfer from portable NIR spectrometer to hyperspectral imaging

Wood Science and Technology, Год журнала: 2024, Номер 59(1)

Опубликована: Дек. 6, 2024

Язык: Английский

---

Classification models for identifying Pterocarpus santalinus L.f. using NIR spectroscopy data

Holzforschung, Год журнала: 2024, Номер unknown

Опубликована: Дек. 24, 2024

Abstract Pterocarpus santalinus L.f. ( P. ), protected under the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES), is a high-priced, slow-growing, scarce wood primarily used crafting high-end furniture. The international timber trade currently faces issues counterfeit , with commonly substitutes including Dalbergia louvelii R.Viguier, tinctorius Welw., Gluta renghas L. Baphia nitida Lodd. This study aims to develop authenticity identification model based near-infrared spectroscopy (NIRS) technology. NIR spectral pretreatment involved use four methods, either individually or combination: multiplicative scatter correction (MSC), moving average smoothing (MAS), Savitzky-Golay (S-G), autoscaling (AUTO) standard normal variate (SNV). An for long short-term memory (LSTM) was established compared support vector machines (SVM) random forest (RF) models. results indicate that accuracy MSC-LSTM 97.1 %, precision, recall, F1 score all exceeding 0.85. In identifying test set, has an error rate only 4.8 %. LSTM performs outstandingly across multiple indicators, demonstrating its ability identify authenticity. developed shows enhanced SVM RF, significantly reducing misidentification .

Язык: Английский

---

Domain-Adversarial Approach to Neural Networks Training to Determine the Composition of Wines Using Various Techniques for Measuring IR Absorption Spectra

Bulletin of the Russian Academy of Sciences Physics, Год журнала: 2024, Номер 88(S2), С. S160 - S165

Опубликована: Дек. 1, 2024

Язык: Английский

---