Rapid detection of peanut oil oxidation by data fusion strategy based on SERS and electronic nose technology

Abstract

In this study, surface-enhanced Raman spectroscopy (SERS) and electronic nose (E-nose) technology were combined to detect the oxidation of peanut oil rapidly and accurately. The most characteristics spectral region associated with peanut oil oxidation was found to be from 799 to 1073 cm−1 using Raman spectroscopy with a surface enhanced gold substrate. Three different data fusion strategies with various machine learning methods were compared to obtain the best qualitative and quantitative prediction model. The results showed that the data fusion strategy effectively improved the model’s prediction performance. BPNN and RF in the low-level fusion strategy were able to identify peanut oil peroxide levels with 100 % accuracy. The most powerful prediction model for peroxide value was obtained using the t-SNE-RF model in the intermediate fusion strategy. It can be concluded that the combination of SERS and E-nose technique is an effective method for rapidly determining the oxidation status of peanut oil.