论文部分内容阅读
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。
Near infrared spectroscopy (1300 ~ 2100 nm) of apple was preprocessed by OSC and NAP respectively, and the prediction model of sugar content was established by partial least squares (PLS) . Application results show that the partial least squares sugar model (OSC / PLS model and NAP / PLS model) is the best one to adopt as the quadrature signal correction factor or the net analyte pretreatment factor is gradually increased during the pretreatment The number of factors will also decrease, even to 1. The OSC / PLS brix model achieved the best performance when pretreatment of apple spectra with 10 quadrature signal correction factors, and the best model adopted a factor of 2. When 11 net pretreatment factors were used to preprocess the spectra, NAP / PLS brix model to achieve the best performance, the best model to adopt a factor of 1. Overall, both the best OSC / PLS brix model and the best NAP / PLS brix model performed significantly better than the best partial least squares model of the original spectrum. These results show that both the quadrature signal correction method and the neat analyte pretreatment method can effectively simplify the apple brix prediction model while ensuring the accuracy.