为了建立醇类汽油定性定量分析判别的通用模型,采用WQF-510A傅里叶变换红外光谱仪与OPUS光谱采集软件获得甲醇汽油、乙醇汽油的中红外光谱。利用主成分(PC)分析和偏最小二乘(PLS)判别法对醇类汽油样品进行定性判别；通过无信息变量消除进行波段筛选,并基于无信息变量消除-偏最小二乘（UVE-PLS）和无信息变量消除-最小二乘支持向量机（UVE-LSSVM）两种方法分别建立醇类汽油的通用模型后用数据进行评价检验。结果表明,利用PC和DPLS对醇类汽油样品定性判别准确率达到100%；基于UVE-LSSVM方法建立的通用模型效果最好,决定系数和预测集均方根误差分别为0.945和2.187。该研究表明醇类汽油定性定量分析判别通用模型是可行的,该模型可以作为醇类汽油醇含量检测的技术参考和理论依据。

In order to establish a general model for qualitative and quantitative analysis of alcohol gasoline, mid-infrared spectra of methanol gasoline and ethanol gasoline were obtained by using WQF-510A Fourier transform infrared spectrometer and OPUS spectral acquisition software. Principal component (PC) analysis and partial least square (PLS) discrimination were used to identify the alcohol gasoline samples. Band screening was conducted by uninformation variable elimination(UVE). The general model of alcohol gasoline was established based on UVE-PLS and UVE-LSSVM (least square support vector machine) and the data were evaluated. The results show that, the accuracy rate of qualitative discrimination of alcohol gasoline by PC analysis and PLS discrimination is 100%.The general model based on UVE-LSSVM method has the best effect, with root mean square error prediction of 2.187 and decision coefficient of 0.945 respectively. The general model is feasible to identify alcohol gasoline by qualitative and quantitative analysis, which can be used as technical reference and theoretical basis for alcohol gasoline alcohol content detection.