A T-S fuzzy model based on initial FCM partition to estimate the composition of commercial blended vegetable oils

Several samples of four types of vegetable oils (sunflower, soybean, corn and peanut oils) produced and/or refined in Portugal were analyzed for their fatty acid composition. Data obtained was used to build a fuzzy model to predict the amounts of different vegetable oils (VO) in commercial blends. The strategy used was a combination of a fuzzy clustering method (Fuzzy C-means) and a Takagi-Sugeno fuzzy model. Model was tested with the data from the analysis of simulated blended VO prepared in the laboratory. Results proved the ability of the model in predicting the blend composition. The model was then used to estimate the composition of 21 brands of commercial VO. Most samples have sunflower as the most representative VO in the blend.