【摘要】 采用正交信号校正法(OSC)和净分析物预处理法(NAP)分别对苹果的近红外光谱(1300~2100 nm)进行预处理,并结合偏最小二乘法(PLS)建立了糖度预测模型。应用结果显示,随着预处理过程中所用的正交信号校正因子或净分析物预处理因子的逐渐增加,偏最小二乘糖度模型(OSC/PLS模型和NAP/PLS模型)所采纳的最佳因子数也会随之减少,甚至可减至1。当采用10个正交信号校正因子预处理苹果光谱时,OSC/PLS糖度模型达到最佳性能,最佳模型采纳的因子数为2;采用11个净分析物预处理因子预处理光谱时,NAP/PLS糖度模型达到最佳性能,最佳模型采纳的因子数为1。从总体上评价,最佳OSC/PLS糖度模型和最佳NAP/PLS糖度模型的性能都明显优于原始光谱的最佳偏最小二乘模型。这些结果表明,正交信号校正法和净分析物预处理法都能在保证精度的同时有效地简化苹果糖度预测模型。更多还原

【Abstract】 Orthogonal signal correction(OSC) and net analyte preprocessing(NAP) were respectively used to pretreat the near infrared(NIR) spectra of apples ranging from 1300 nm to 2100 nm,and the models of sugar content were developed from these two pretreated spectra by the partial least square(OSC/PLS models and NAP/PLS models).Results showed that the optimal number of factors used in OSC/PLS models and NAP/PLS models would reduce as the number of OSC factors and NAP factors increased on by one,even to 1 finally.The best OSC/PLS model with 2 PLS factors was obtained when 10 OSC factors were used in pretreatment;and the best NAP/PLS model with 1 PLS factor was obtained when 11 NAP factors were used in pretreatment.Although the best OSC/PLS model and the best NAP/PLS model(didn’t) improve precision to a great extent,they needed fewer factors and became more parsimonious,compared with the model before NAP pretreatment.In a conclusion,OSC and NAP pretreatment could simplify the prediction model of sugar content.更多还原