کد مقاله | کد نشریه | سال انتشار | مقاله انگلیسی | نسخه تمام متن |
---|---|---|---|---|
1186010 | 963422 | 2010 | 6 صفحه PDF | دانلود رایگان |
The aim of this study was to develop optimal NIRS calibration for ash content prediction in legumes by using the thermogravimetric (TGA) and gravimetric (GA) analytical methods. The calibration was performed on the basis of whole and structured sample sets (n = 143 and n = 99, respectively). Samples were scanned using a Rapid Content Analyzer in reflectance mode (400–2500 nm). Different mathematical treatments of the spectra preceded modified partial least squares (MPLS) regression analyses. The performance of the models was assessed by cross validation and external validation (n = 44). Models developed for the whole sample set on the basis of the TGA and GA methods were characterised by standard error of calibration (SEC) ranged from 0.28 to 0.50, standard error of cross validation (SECV) ranged from 0.43 to 0.60, coefficient of determination (R2) ranged from 0.97 to 0.89, explained variance (1 − VR) ranged from 0.94 to 0.85 and residual predictive deviation (RPD) ranged from 4.23 to 2.68, respectively. Models developed for the structured sample set on the basis of the TGA and GA methods were characterised by standard error of calibration (SEC) ranged from 0.32 to 0.42, standard error of cross validation (SECV) ranged from 0.53 to 0.56, coefficient of determination (R2) ranged from 0.97 to 0.94, explained variance (1 − VR) ranged from 0.91 to 0.89 and residual predictive deviation (RPD) ranged from 3.52 to 2.98, respectively. The obtained results showed the potential of NIRS method to accurately predict the ash content of legume grass samples that correspond to ash content determined by the TGA and GA methods.
Journal: Food Chemistry - Volume 123, Issue 3, 1 December 2010, Pages 800–805