Rapid Multi-Factor Evaluation System for Full-Process Risk Assessment of Coal Spontaneous Combustion in Engineering Applications

Existing coal spontaneous combustion liability assessments suffer from incomplete temperature range coverage, poor cross-rank comparability, and weak correlations between microscopic essence and macroscopic criteria—issues that undermine reliability and risk coal mine safety. This study aims to establish a structure-driven intrinsic identification system to address these gaps. Using 10 cross-rank coal samples (lignite, bituminous coal, and anthracite), we conducted systematic research via experiments, model building, and theoretical verification. We integrated three stage-specific parameters (each matching a combustion phase): saturated oxygen uptake (VO2, 30 °C chromatographic adsorption), average heating rate R70 (40–70 °C adiabatic oxidation), and Fuel Combustion Characteristic index (FCC, 110–230 °C crossing point method). With Information Entropy weighting (VO2: 0.296; R70: 0.292; and FCC: 0.412), we constructed the Multi-Factor Comprehensive Spontaneous Combustion Index (MF-CSCI). We also screened functional groups via FTIR, built a microstructure-driven model (MD-CSEI, linking groups to MF-CSCI), and verified mechanisms via DFT. Results show MF-CSCI covers the full “adsorption-heat accumulation-self-heating” process: HG lignite (MF-CSCI = 1.0) had high liability and YCW anthracite (MF-CSCI = 7.98) had low liability, solving cross-rank issues. Pearson analysis found –OH positively correlated with MF-CSCI (r ≈ −0.997), C=C negatively (r ≈ −0.951); MD-CSEI achieved R2 = 0.863 (p = 0.042). This study improves cross-rank assessment accuracy, enables rapid micro-to-macro risk prediction, and provides a theoretical basis for on-site coal safety management.