Studies Of The Fast Evaluation Method To Rapeseed Oil Quality Based On Near Infrared Reflectance Spectroscopy

Rapeseed oil is a kind of important edible oil in China, rapid detection and scientific evaluation of its quality is the main research content. At present, the detection methods of rapeseed oil are complex, time-consuming, high labor intensity etc. In this paper, the near infrared spectroscopy analysis technology was employed, the rapeseed oil fatty acid composition, physical, chemical index of near infrared spectral characteristics, as well as quality of the fast evaluation method was researched, the prediction model of rapeseed oil key indicators was built, hence, a rapid nondestructive determination of rapeseed oil quality was implemented. The main conclusions are as follows:1. Establishing the optimum scanning spectrum condition which is collecting near infrared transmission spectroscopy of rapeseed oil.By using rapeseed oil as raw material, the best rapeseed oil near infrared spectroscopy scanning parameters which are determined by employing Ju Gang Science NIR-2600 near infrared analyzer: path 5mm, scanning 10 times, scanning temperature is 40℃. Under such conditions, the collection of rapeseed oil near infrared spectroscopy is not only stable, but also the quantitative model has higher matching precision.2. Establishing the multivariate linear prediction model of evaluation detection of rapeseed oil fatty acid composition.145 rapeseed oil samples were collected, the gas chromatography detected 7 kinds of fatty acids including oleic acid, linoleic acid, palmitic acid, stearic acid, eicosenoic acid and erucic acid, through spectral acquisition, abnormal samples alimentation, pretreatment scheme optimization, feature optimization wavelength selection steps and et, al. It’s built the multivariate linear prediction model of 7 kinds of fatty acid composition which is based on near infrared spectroscopy. The precision of 7 models are high fitting(models’ Rc were all greater than 0.9943, RMSEC are all less than 0.2631). The near infrared fingerprint combination of rapeseed oil seven kinds of main fatty acids is determined which the range of wavelength is 1000nm~1250nm, 1350nm~1408nm, and 1660nm~1818nm.3. Establishing multivariate linear regression forecasting model of rapeseed oil chemical index.126 rapeseed oil samples were collected, by employing national standard methods to detect 3 kinds of chemical indicators including the acid value, peroxide value, iodine value, through spectral acquisition, abnormal samples eliminating, pretreatment, feature optimization method of wavelength selection steps, the multivariate linear prediction model of 3 chemical indexes were built which are based on the near infrared spectroscopy. The precision of 3 models is high fitting(models’ Rc were all greater than 0.9684, RMSEC are all less than 0.5154). The near infrared fingerprint combination of rapeseed oil acid value is determined which the range is 1000nm~1200nm; the ranges of peroxide value are 1000nm~1250nm, 1360nm~1460nm and 1680nm~1790nm; iodine value is in the range of 1660nm~1770nm.4. Build the multivariate linear prediction model of the evaluation detection of rapeseed oil physical index.126 rapeseed oil samples were collected, by using Lovibond colorimeter to detect color yellow value and red value, rotary Digital Viscometer to detect viscosity index, and through spectral acquisition, eliminating abnormal samples, pretreatment, feature optimization method of wavelength selection steps, the multivariate linear prediction model of 2 kinds of physical index was built which was based on the near infrared spectroscopy. The precision of color models is high fitting(models’ Rc were all greater than 0.9838, yellow value RMSEC is 2.1058, red value RMSEC is 0.2335); viscosity model Rc is 0.8395, RMSEC is 2.9731. The near infrared fingerprint combination of rapeseed oil color is determined which the ranges are 1000nm~1180nm, 1350nm~1480nm and 1640nm~1700nm; the viscosity ranges are 1160nm~1300nm and 1600nm~1700nm.