Investigation of the impact of pyrite content on the terahertz dielectric response of coals and rapid recognition with kernel-SVM

Abstract

Pyrite promotes the coal spontaneous combustion of coal. Owing to the highly uneven distribution of pyrite in coal seams, different areas of the same coal seam may have significantly different spontaneous combustion tendencies. To mitigate the adverse effects of pyrite, a method for detecting pyrite distribution in coal seams is imperative to be developed. This study employed terahertz time-domain spectroscopy to measure the terahertz dielectric response of coals with varying pyrite contents, focusing on discerning the underlying differences and mechanisms. Results emphasize the coal's metamorphism as a key influencer of terahertz signals. Coals that are more highly metamorphosed generally exhibit higher signal amplitude and lower complex permittivity. Signal peak amplitudes demonstrate a linear attenuation trend with increasing pyrite content, particularly pronounced in highly metamorphic coals. Regarding the complex permittivity, the real part displayed irregular changes with variations in pyrite content, whereas the imaginary part exhibited linear growth. Terahertz waves prove sensitive to pyrite-induced relaxation-type polarization. Moreover, support vector machine classification, employing terahertz imaginary spectrum, achieves an approximately 90 % multiclass accuracy in pyrite estimation across diverse coal types. The conclusions offer valuable theoretical insight for terahertz technology applications, including controlling coal spontaneous combustion, mining associated pyrite, and analyzing coal quality.