Establishment Of Rapid Determination Models For Poly Unsaturated Fatty Acids

Poly unsaturated fatty acids (PUFA) is the fatty acids which contains more than two double bonds and eighteen carbon atoms in the chain. PUFA has high nutritive value and health funtion, is one of the most important materials to be developed as health care food. At present, the common methods used in determining fatty acids are Gas Chromatography (GC) and other chemical methods, these methods are so tedious that can't feed back the results in time. Compared with the conventional methods. Near Infrared spectroscopy (NIR) has many advantages. such as rapid, accurate and environmental protection. The paper takes 88 varieties of Docosahexaenoic Acid (DHA) powdered oils and 100 varieties of Arachidonic Acid (AA) powdered oils as samples. After the accuracy and reliability of GC method were investigated, spectrum of the samples were scanned using Nicolet 5700 FT-IR based on the examination of the content of DHA and AA by GC. Then establish near infrared spectroscopy calibration models of each components using different pretreatments and Partial Least Square (PLS). Finally determine the optimum model and modeling parameters. The main results as follows:(1) GC method can be used as the standard method to establish calibration model with its accuracy and reliability. The difference of PUFA contents between samples is significantly, dispersion and volatility changes relatively greatly, the samples are representative. The DHA content amplitude of testing samples is 5.79～11.61%. with the mean value and standard deviation is 8.40% and 1.31%. respectively; The AA content amplitude of testing samples is 3.88～11.09%. with the mean value and standard deviation is 7.88% and 1.96%. respectively.(2) The optimum NIR calibration model and modelling parameters of DHA were established by developing models with different spectral pretreatments. The optimum conditions of DHA powdered oils NIR calibration model are as follows: spectral range are 4142.35～4177.06 cm^-1. 5789.26-5989.82 cm^-1 and 6969.48～7062.05 cm^-1, the pretreatment is SNV and first-derivative, the number of principal component is 5. The correlation coefficient is 0.968. the RMSECV is 0.45. the RMSEP achieves 0.59. (3) The optimum NIR calibration model and modelling parameters of AA were established by developing models with different spectral pretreatments. The optimum conditions of AA powdered oils NIR calibration model are as follows: spectral range are 4250.34～4358.33 cm^-1 and 5723.6～5881.82 cm^-1, without pretreatment, the number of principal component is 3. The correlation coefficient is 0.967, the RMSECV is 0.50. the RMSEP achieves 0.57.The results showed that NIR calibration model was an effective method to predict the PUFA content of powdered oils.