CAJ | 학술논문

选取116份大花生为实验材料，应用近红外反射光谱技术，结合偏最小二乘法，采用交叉检验建立了大花生含水量的近红外模型。优化结果表明，原始光谱不经过预处理，光谱范围为4597．7～11988C1TI-1，维数12，此时建立的模型校正结果最佳。模型决定系数（形）为93．62，根均方差（RMSECV）为1．17。用该模型对20个未参与建模的实验材料进行预测，偏差为-1．781～1．902，相对误差为0．122Voo～2．855％。结果表明含水量模型具有很好的预测准确性，可用于鲜食花生种子水分含量快速检测。
선취116빈대화생위실험재료，응용근홍외반사광보기술，결합편최소이승법，채용교차검험건립료대화생함수량적근홍외모형。우화결과표명，원시광보불경과예처리，광보범위위4597．7～11988C1TI-1，유수12，차시건립적모형교정결과최가。모형결정계수（형）위93．62，근균방차（RMSECV）위1．17。용해모형대20개미삼여건모적실험재료진행예측，편차위-1．781～1．902，상대오차위0．122Voo～2．855％。결과표명함수량모형구유흔호적예측준학성，가용우선식화생충자수분함량쾌속검측。
A near infrared reflectance spectroscopy calibration model for prediction of moisture content in individual single peanut seed was developed using 116 peanut seed samples, PLS algorithms and cross validation. The best pretreatment method was no treatment, the optimal wavenumber was 4597.7-11988 cm-1, and rank was 12. R2 and RMSECV was 93.62 and 1.17, respectively. The model was further tested with 20 external peanut seed samples, and it was found that the error was --1. 781-1. 902, and relative error was 0. 122%-2. 855~/60, demonstrating the robustness of the model. The NIRS model can be used for rapid analysis of moisture content in freshly harvested peanut seeds.