Frying disposal time of sunflower oil using hybrid electronic nose-fuzzy logic approach

Abstract

An electronic nose (e-nose), equipped with 18 metal oxide semiconductor (MOS) sensors (L, T, and P types), was used to monitor the disposal time of deep fried sunflower oil stabilized with natural antioxidants. E-nose was interfaced with chemometrics and fuzzy logic analyses to rank and screen the most effective MOS sensors against headspace volatiles of fried oils. The chemical indices of rancidity viz., total polar compounds (TPC) and triglyceride dimers-polymers (TGDP), among others, were measured and correlated with e-nose based rancidity index (odor index). The inherent clustering of oil samples with varying degree of rancidity was deconvoluted by principal component analysis and hierarchical clustering on principal components. Six MOS sensors (LY2/G, LY2/AA, LY2/GH, LY2/gCT1, T30/1, and P30/1) were screened and ranked using fuzzy logic analysis. A good relationship was noted between rancidity indices and odor index (R2>0.85). Upon reaching threshold discard limit of TPC (25 g/100 g oil), the frying disposal time was determined to be approximately 16 h (15.5 h (chemical test) vs. 16.24 h (e-nose)). The hybrid e-nose-fuzzy logic approach could substitute the existing chemical methods and integrated for on-line quality inspection of cooking oils.