最小一乘稳健多元分析校正

Robust Multivariate Calibration Based on Least Absolute Deviation

本文论述最小一乘求解的多元分析校正算法,探讨了最小一乘较常规最小二乘法及其他稳健算法的优点。用计算机数值模拟及实际多组分光谱体系对方法进行了检验,展示了最小一乘法在分析化学计量学中实际应用的可行性。

The robust multivariate calibration algorithm based on least absolute deviation (LA) is introduced. The advantage of LA algorithm over the least square method has been verified by using numerical simulation and the treatment of real spectrophotometric data. The feasibility of the application of robust LA calibration in chemometrics has been con-firmed.