Development of a novel algorithm for rapid alignment of big spectral data based on orthogonal base

Abstract

The alignment of an unknown sample spectrum with a standard spectral library is an important approach for qualitative analysis. With the advent of the data era, the data volume of standard spectral library has significantly increased, resulting in increased pressure on the analysis speed of traditional methods. In this study, a novel algorithm for the rapid alignment of big spectral data based on orthogonal base was developed. The proposed method is based on the basic chemical law that compounds with different structures have typical spectral characteristics. The orthogonal base with definite chemical information was constructed by extracting representative information of the spectral characteristics of all types of compounds, and then comprehensively compared with unknown samples to draw conclusions. Taking the standard Raman spectral library (Aldrich Raman) as an example to verify the method, five types of hydrocarbon compounds were properly classified with a correct classification rate of 93.94%. This method can be further applied to subdivision alignment of other structure (such as compounds with different numbers of substituent groups), and has the potential to be extended to other spectral libraries.