A novel algorithm of computational spectral analysis integrating factor analysis with Kalman filter

Abstract

By integrating factor analysis (FA) with Kalman filter (KF), a novel algorithm used for computational spectral analysis is proposed. In the algorithm, the factors extracted from the data set of the multicomponent system with factor analysis technique have been employed to construct the calibration model of KF. In this way, the inherent limitation of KF, in which calibration modeling highly relies on the pure spectra of all the components in the multicomponent system, has been overcome. Computer simulatin indicates that the results obtained by the proposed algorithm are accurate. A typical example of three-lomponent drug including amino-pyrine, caffein and phenacetin has been used to verify the effectiveness of algorithms. Compared with what obtained by KF, the total mean relative error obtained by the proposed algorithm decreases from 2.78% to 1.78%.