Analysis and identification of the parent coal sources of fulvic acid according to convention, spectroscopy and chemometrics.

Abstract

Fulvic acid (FA) is a kind of organic and complex water-soluble components mainly extracted from low rank coals with small molecular weight, active physical properties (such as cation exchange capacity, pH-buffering alkalinity) and positive biological functions. However, the performance of FA varies greatly, mainly induced by its different sources of raw coals. Thus, classifying the fulvic acid obtained from different coal samples is required. According to their chemical differences, two methods are developed in this paper to distinguish the origin of coal in China in combination with chemometric tools. First, the ash content, elemental composition, ultraviolet-visible (UV-Vis) and fluorescence spectra of sixteen fulvic acid samples from peat, lignite and weathered coal are measured and fifteen parameters are obtained from each sample. In the first Linear Discriminant Analysis (LDA) strategy, Orthogonal Projections to Latent Structures Discriminant Analysis (OPLS-DA) and stepwise LDA are employed to reduce variables. A discriminant function (DF) constructed only by EEt/Bz and FI is obtained, with its accuracy verified by clustering and leave-one-out cross validation (LOOCV) with an accuracy of 87.5%. In another machine learning tactics, Pearson correlation and principal component analysis (PCA) reduce the dimensions of all variables. In the end, all sixteen samples are divided into three groups by support vector machine (SVM), with an accuracy of 100%. In conclusion, based on the differences in the chemical composition of FA from different sources, the method for combining UV-Vis and fluorescence with LDA or SVM can effectively classify the coal sources of FA.