红外光谱法用于单组分含氧化合物的定量分析研究

Quantitative analysis of single-component oxygen compound by infrared spectroscopy

注空气催化氧化采油技术是目前应用较为广泛的新型采油技术。含氧化合物含量是表征石油的重要性质,根据氧化前后含氧化合物中氧含量的差值,分析氧气的去向,判断采油过程是否安全,在采油工艺技术上起到指导作用。通过采用主成分分析法压缩环己酮模拟油组分的红外光谱全谱数据（4000-400）cm^-1将其主成分作为模型的输入信息,并结合反向传播人工神经网络以及支持向量机回归方法建立测定模拟油中单组分含氧化合物含量的校正模型,对石油产品中含氧化合物的浓度进行有效的预测。3种方法建立的定量校正模型中环己酮浓度的预测值和实验检测值数据拟合程度良好,相关系数R分别为0.9916,0.9924,0.9961,均方根误差分别为0.03790,0.04787,0.03775。研究结果表明,3种方法建立的定量校正模型均具有较强的稳定性和良好的预测能力,但支持向量机回归模型的预测结果稍好于其他2种方法。与传统测定含氧化合物含量的方法相比,红外光谱法测量速度快,测试过程简便,分析成本低且绿色环保,可以满足采油工业中快速测量的需要。

Oxygen compound content is an important factor for characterization of oil.In the process of catalysis and oxidation in air injection in oil, measuring the concentration of oxygen compounds has an important significance for exploring the mechanism of catalytic oxidation and guiding the technology of oil production. Titration and other traditional methods are tedious and time-consuming for the determination of the oxygen compounds. As an ideal method, infrared spectroscopy technology has a fast response, wide measurement and simple using. Build an accurate and efficient model is the key of using infrared spectroscopy to measure the concentration. Component regression, hack propagation neural algorithm and support vector machine for regression are used to determine the concentration of cyclohexanone.The full spectra as the calculation range of spectral matrix to make the model a high extension. The infrared spectroscopy datas were compressed by principal component analysis and were used as input information to develop models. The three quantitative models can respectively predict the oxygen compound content, but the results of SVM models are superior to the results of the other two models. And the excellent statistical parameters reveal that the models are robust and have high predictive capability.this method,comparing with traditional standards, provides with advantages,such as simplicity, high-speed,no damage and green.