摘要
采用近红外光谱测定在紫外光谱区表现为弱吸收响应的生化废水的化学需氧量(COD)。通过主成分分析-欧氏空间距离类聚法筛选出具有代表性的样本,从系统独立变量数的角度判断了主成分数,结合变量标准化-基线扣除的预处理方法消除光谱信号背景干扰,建立了偏最小二乘校正模型。在系统主成分数为6时,水样的COD实测值与预测值之间的相关系数为0.9348,预测标准差(RM SEC)为61.89m g/L,模型有较好的预测能力。
The COD in biological and chemical wastewater that performed to weak absorption in the UV spectral region was determined by NIR spectrometry.Principal component analysis-Euclidean distance cluster was applied to select the representative sample.From the aspect of the number of independent variables,the number of principal components was determined.Combined with the SNV-Detrending preprocessing,the background signal of spectra was eliminated.The partial least squares model was built and the predictions were satisfied.When main component in the system amounted to 6,the linear correlation coefficient of measured values and predicted values was 0.9348 and the RMSEC was 61.89mg/L,the model had a better predictive ability.
出处
《光谱实验室》
CAS
CSCD
北大核心
2011年第6期3040-3043,共4页
Chinese Journal of Spectroscopy Laboratory
基金
广西自然科学基金资助项目(桂科自0832064)
广西教育厅桂教科研资助项目(200708MS067)