Quality Analysis Of Extra Virgin Olive Oil By Chemometrics Methods And FT-IR Spectra

The classification and quantification of adulteration in extra virgin olive oil based on Fourier transform infrared spectroscopy(FT-IR) were carried out in this paper. In order to handle the massive amounts of spectral data, several data dimension reduction methods were introduced. Those variables selection methods combined with classification and quantification algorithm were implemented for olive oil analysis.1、 The significance of olive oil quality analysis was introduced firstly. Various analysis methods for adulteration in olive oil were presented, including chemical methods and chemometrics algorithm. Fourier transform infrared Spectroscopy(FT-IR)was also illustrated emphatically.2 、 We proposed an adaptive fuzzy c means by particle swarm algorithm combined with Fourier transform infrared spectroscopy to obtain classification and semi-quantitative information of extra virgin olive oils adulterated with peanut oils. In order to implement classification and quantification procedure simultaneously, the fitness function was modified. The results shown that this chemometrics method can classify and semi-quantify the adulterated oil simultaneously and is a stable and reliable method for identifying extra virgin olive oils.3、Detection the adulteration and contamination is one of the main aspects in the quality control of extra virgin olive oil(EVOO). In this paper, we sought to identify the adulterated oil from the EVOO, to discriminate the type of adulterants and to quantify the levels of the adulteration by FT-IR coupled with chemometrics.The supervised locally linear embedding was employed to reduce the dimensionality of variables and compared with principal component analysis and locally linear embedding. The results have shown that we can clearly identify which edible oils are adulterated and accurately quantify the percentage of adulteration in EVOO.4 、 Particle swarm optimization combined with Gaussian mixture model and Gaussian mixture regression(GMMRPSO) was proposed for olive oil analysis.Expectation maximization(EM) was used to obtain the parameters in classical GMM.Considering the limitations of EM, PSO was introduced to obtain optimal parameters.In order to reduce the amount of calculation, kernel principle component analysis was used for dimensionality reduction. For comparison, PCA and LLE were also implemented in dimensionality reduction. The result has shown that KPCA combined with GMMRPSO can obtain a satisfactory results and is an efficient method for adulteration analysis.