研究目的
To develop accurate, rapid, non-destructive NIRS calibration models for analyzing moisture, oil, and fatty acid content in single plant canola seeds using a diode array NIRS and mirrored cup.
研究成果
The study successfully developed NIRS calibration models for moisture and oil content in single plant canola seeds using a diode array NIRS and mirrored cup, enabling rapid, non-destructive screening without pre-drying. However, models for fatty acids were not adequate, highlighting the need for broader constituent variability and possibly a wider wavelength range in future research.
研究不足
The NIRS models were not successful for fatty acid and glucosinolates content due to limited variability and low levels of these constituents, as well as the narrow wavelength range (950-1650 nm) of the instrument. Sample weight variability (2.1-4.3 g for single plant seeds) also affected prediction accuracy, especially for weights less than 3 g.
1：Experimental Design and Method Selection:
The study aimed to develop NIRS calibration models for canola seed constituents. It involved using a diode array NIRS instrument with a mirrored cup, multivariate prediction models (PLS, PCR), and various pre-processing techniques (mean center, derivatives, standard normal variate).
2：Sample Selection and Data Sources:
2500 canola samples from five North Dakota locations were collected for diversity. 100 samples were selected for calibration and 30 for validation based on uniform constituent distribution. Additional 30 single plant samples were used for further validation.
3：List of Experimental Equipment and Materials:
DA7200 NIR spectrometer, Micro Mirror Module, mirrored cup, gravity convection oven, accelerated solvent extraction unit (ASE 200), gas chromatography equipment, coffee grinder, diatomaceous earth, n-hexane, Mylar bags.
4：Experimental Procedures and Operational Workflow:
Spectra were collected at room temperature using the DA7200 with mirrored cup. Samples were scanned multiple times and averaged. Reference analyses included moisture determination by oven drying, oil extraction using ASE, and fatty acid analysis by GC. Calibration models were built using GRAMS Suite software with different pre-processing and regression methods.
5：Data Analysis Methods:
Statistical evaluation using R2, SEP, RPD values from validation sets. SEL was calculated for reference methods to ensure accuracy.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
