Quantitative analysis of the 2-mercaptobenzothiazole based on terahertz time-domain spectroscopy and an improved support vector regression

Abstract

2-Mercaptobenzothiazole is an irreplaceable accelerator for rubber vulcanization. It is important to study the content of 2-mercaptobenzothiazole which has an important influence on the vulcanization process of rubber. In order to improve the detection efficiency of 2-mercaptobenzothiazole, a new quantitative analysis method was proposed based on terahertz time-domain spectroscopy and an improved support vector regression. Experimental samples are mixtures of 2-mercaptobenzothiazole and polyethylene with different mass fractions. Their time-domain spectroscopy was obtained by terahertz time-domain spectroscopy system. It is found that 2-mercaptobenzothiazole has obvious absorption characteristics at 0.88 THz. Use Least squares support vector regression optimized by bat algorithm, Correlation coefficient and root mean-square error of the prediction set were 0.9987 and 0.5564%. Experimental results demonstrate that the predictive result of least squares support vector regression optimized by bat algorithm model is better than that of partial least squares and support vector regression model. It provides a potential approach for quantitative analysis of other rubber additives.