改进DOAS技术在混合气体中的定量分析

Improvement of DOAS Technology in the Application of Mixed Gas Quantitative Analysis

针对差分吸收光谱技术(DOAS)中混合气体吸收光谱重叠问题,提出了一种建立在粒子群优化算法和最小二乘支持向量机算法融合上的改进的DOAS方法.采用最小二乘支持向量机技术对SO2、NO分别建立定量分析模型,并运用粒子群优化算法的强大寻优能力对最小二乘支持向量机算法中的参数进行寻优,最后对定量分析模型进行重建.实验结果表明,采用数据融合的DOAS方法,在解决混合气体光谱重叠问题上切实可行,分别将SO2、和NO最大绝对误差率提高到1.200 0%和2.691 8%,具有一定的实际意义.

According to difference absorption spectrum,mixed gas absorption spectrum in technology overlapping problem,a DOAS technology based on the PSO and LS-SVM fusion method was proposed.First,SO2,NO respectively to construct a quantitative analysis model by LS-SVM.Then,using PSO＇s strong optimization ability of least square support vector machine（SVM） the optimization of the parameters of the algorithm,and then,LS-SVM＇s parameter is optimized with PSO of the powerful searching ability.Finally,we rebuilt the quantitative analysis model.Experiments show that using the improved DOAS method is feasible in solving the mixed gas spectrum overlap problem,which has certain practical significance.