A New Strategy for Quantitative Proportions in Complex Systems of Blended Oils by Triacyglycerols and Chemometrics Tools

Many oils sold in China and India are a blend of various oils to improve performance, stability, and nutritional characteristics, which are required in their respective markets. Quantitative analysis of the proportions of constitutive components is fundamental to the conformity and adulteration checking of edible blended oil products. A multi linear regression model with constrained linear least squares and exhaustion calculation was applied in this study. The source of the varieties in the model is a database (614 pure oils) of triacylglycerols (TAGs) collected by GC–FID and HPLC–RID. There were 20 groups of binary and ternary blended oils consisting of two or three oils out of five kinds, namely soybean, corn, peanut, rapeseed, and sunflower, which were analyzed and processed separately. Results showed that the method was able to predict the proportions of constitutive components in the edible blended oils, given that relative errors required less than 20%, the accuracy was 98.2% for the binary system if the proportion of each oil in blended oils was more than 20%, while the accuracy was 84.7% for the ternary system if the proportion of each oil in blended oils was more than 10%. The quantitative method is based on a simple analysis to determine the TAGs composition and thus it is useful for quick segregation and quality control of blended oils in routine analysis.