铜胁迫下玉米叶片的HHT包络谱变化与污染预测模型

根据不同Cu~(2+)胁迫梯度下玉米叶片光谱及叶片中Cu~(2+)含量数据,利用经验模态分解(EMD)获取不同胁迫梯度下叶片光谱的各阶本征模态函数(IMF),选IMF4进行HHT包络谱分析,研究了玉米叶片在不同Cu~(2+)胁迫梯度下光谱的Hilbert包络谱变化特征与污染程度预测方法。通过构建包络谱峰值指标(E_1)和包络谱峭度系数(E_2)特征参量,分析叶片在不同Cu~(2+)胁迫梯度下的包络谱变化;同时基于包络谱特征参量值与叶片中Cu~(2+)含量的相关性分析和逐步回归统计,建立了玉米叶片重金属污染的Cu~(2+)含量的单、双因素变量预测模型。实验结果表明:不同Cu~(2+)胁迫梯度下,玉米叶片光谱的包络谱为分布在100 Hz频率以内的连续谱;E_1和E_2值均表现出与叶片中Cu~(2+)含量呈正相关的变化趋势,两个包络谱特征参量值都在Cu(150)位置达到最大值;由于E_1和E_2值与叶片中Cu~(2+)含量都具有良好的相关性,可把E_1和E_2作为监测玉米重金属污染预测指标。根据E_1、E_2单因素及双因素变量构建的Cu~(2+)含量预测模型应用结果比较,表明E_1、E_2双因素变量构建的模型具有最优的预测能力。

Changes of HHT Envelope Spectra and Pollution Prediction Models on Corn Leaves Polluted by Copper Stress

Based on the data of leaf spectrum and Cu2+ content in corn leaves under different Cu2+ stress gradients, the intrinsic mode function (IMF) of leaf spectrum under different stress gradients was obtained by empirical mode decomposition (EMD), and IMF4 was selected for HHT envelope analysis. The variation trend and pollution degree of Hilbert envelope spectrum of maize leaves under different Cu2+ stress gradients were studied. The peak index of envelope spectrum (E1) and envelope spectral kurtosis coefficient (E2) were constructed to analyze the envelope spectrum changes of leaves under different Cu2+ pollution levels. The correlation analysis and stepwise regression statistics between the characteristic parameters of envelope spectrum and the content of Cu2+ in leaves were also carried out to establish a single and double variable prediction model of Cu2+ content in maize leaves. The experimental results showed that the spectral envelope of corn leaves under different concentrations of Cu2+ stress was a continuum within 100Hz frequency;values of E1 and E2 showed a positive correlation with the Cu2+ content in leaves. The characteristic parameter values of two envelope spectra all reached the maximum at the Cu(150) position, since both E1 and E2 values had a good correlation with the content of Cu2+ in leaves, E1 and E2 can be used as predictors for monitoring heavy metal pollution in maize plants. According to the application results of Cu2+ content prediction model constructed by E1 and E2 single and double variables, it was proved that the bivariate model constructed by E1 and E2 had the best predictive ability.