论文部分内容阅读
以土壤pH、野外实测光谱以及多元散射校正(MSC)预处理后的光谱数据为基础,利用数学方法(主成分回归PCA、偏最小二成回归PLSR、BP神经网络模型)分别建立了土壤pH的预测模型。结果表明:土壤实测光谱和经过MSC方法预处理的光谱数据均与pH存在良好的相关性,并呈极显著水平,后者的相关性更高。PCA和PLSR两种土壤pH估测模型均具有良好的预测能力。BP神经网络模型则因输入变量多,预测精度较低。但利用PCA和PLSR模型所获得主成分,作为BP神经网络的输入变量所建立的复合模型,可明显提高模型稳定性和预测能力。
Based on the spectral data of soil pH, field measured spectra and MSC pretreatment, the soil pH was established by using principal component analysis (PCA, PLSR, BP neural network model) Predictive model. The results showed that there was a good correlation between the measured spectra of the soil and the spectral data pre-processed by the MSC method and the pH, and the correlation coefficient was very significant. Both PCA and PLSR soil pH estimation models have good predictive ability. BP neural network model because of the input variables, the prediction accuracy is low. However, by using the principal components obtained from PCA and PLSR models, the compound model established as the input variable of BP neural network can obviously improve the stability and prediction ability of the model.