中文摘要：

利用衰减全反射傅立叶变换红外光谱（FTIR／ATR）以及多元线性回归模型，分别采用经典最小二乘回归算法和逆最小二乘回归算法，建立了一种多组分同时在线的定量分析方法．这种方法精度高，结果准确，平均相对误差不到3％．两种回归算法中，以逆最小二乘算法获得的结果具有更高的精度和准确性，更适合含有少量杂质的体系．

英文摘要：

A quantitative multi-component analysis by FTIR/ATR with multiple regression is performed. Through this method, many components can be quantitative analyzed on-line at the same time. In Multiple Regression model, Classic Least Squares or Inverse Least Squares were adopted. The calculated value from this method has higher precision and accuracy compared with traditional methods with its average relative error less than 3%. The calculations from two regressions, value of ILS have higher precision and accuracy and it is more adaptable to the system which has little impurity.