Abstract In the present research work, portable near-infrared (NIR) spectroscopy coupled with different types of chemometric algorithms like partial least-squares (PLS) regression and some effective variable selection algorithms, i.e., synergy interval-PLS (Si-PLS), genetic algorithm-PLS (GA-PLS) and synergy interval genetic algorithm-PLS (Si-GA-PLS) were used for the quantification of chemical compositions of peanut seed samples; also the Principle Component Analysis (PCA) and Linear Discriminant Analysis (LDA) models were applied for discrimination of peanut of different regions. The compositional parameters, i.e., total phenolic content (TPC), fat, protein, fiber, carbohydrate, moisture, ash and pH, were estimated. The results of the developed model estimated by applying correlation coefficients of the calibration (Rc) and prediction (Rp); root mean standard error of cross-validation, RMSECV; root mean square error of prediction, RMSEP and residual predictive deviation, RPD. The efficiency of the developed model was significantly enhanced with the use of Si-PLS, GA-PLS and Si-GA-PLS correlated with the classical PLS model. The results of Rp determined for prediction and Rc calibration set differ from 0.7473 to 0.9420 and 0.7794 to 0.9623 correspondingly. These results showed that portable NIR spectroscopy coupled with different chemometric algorithms having the potential to be applied for the prediction of the chemical compositions of peanut seed samples.

中文翻译：

---

使用便携式近红外光谱法快速测定花生种子 (Arachis hypogaea.) 的化学成分

摘要 在目前的研究工作中，便携式近红外 (NIR) 光谱结合不同类型的化学计量算法，如偏最小二乘法 (PLS) 回归和一些有效的变量选择算法，即协同间隔-PLS (Si-PLS)，采用遗传算法-PLS（GA-PLS）和协同区间遗传算法-PLS（Si-GA-PLS）对花生种子样品的化学成分进行定量；还应用主成分分析（PCA）和线性判别分析（LDA）模型对不同地区的花生进行判别。估计了组成参数，即总酚含量 (TPC)、脂肪、蛋白质、纤维、碳水化合物、水分、灰分和 pH 值。通过应用校准 (Rc) 和预测 (Rp) 的相关系数估计开发模型的结果；交叉验证的均方根标准误差，RMSECV；预测的均方根误差 RMSEP 和残差预测偏差 RPD。通过使用与经典 PLS 模型相关的 Si-PLS、GA-PLS 和 Si-GA-PLS，开发模型的效率显着提高。为预测和 Rc 校准集确定的 Rp 的结果相应地从 0.7473 到 0.9420 和 0.7794 到 0.9623 不同。这些结果表明，便携式 NIR 光谱结合不同的化学计量算法具有用于预测花生种子样品化学成分的潜力。GA-PLS 和 Si-GA-PLS 与经典 PLS 模型相关。为预测和 Rc 校准集确定的 Rp 的结果相应地从 0.7473 到 0.9420 和 0.7794 到 0.9623 不同。这些结果表明，便携式 NIR 光谱结合不同的化学计量算法具有用于预测花生种子样品化学成分的潜力。GA-PLS 和 Si-GA-PLS 与经典 PLS 模型相关。为预测和 Rc 校准集确定的 Rp 的结果相应地从 0.7473 到 0.9420 和 0.7794 到 0.9623 不同。这些结果表明，便携式 NIR 光谱结合不同的化学计量算法具有用于预测花生种子样品化学成分的潜力。