Photoacoustic Time–Frequency Analysis for Quantification of Coal Thermal Properties and Its Classification to Spontaneous Combustion Susceptibility

Abstract

Accurate quantitative assessment of coal thermal properties is critical for predicting spontaneous combustion susceptibility (SCS), particularly for on site evaluation in coal stockyards where early decision making can mitigate unwanted coal burning and greenhouse gas emissions. This work presents AI-supported photoacoustic (PA) evaluator of coal thermodynamics, a compact, pulsed laser diode-based PA sensing system for on-field assessment of coal ignition temperature. The novelty of the proposed method lies in the exploitation of a time-frequency-based PA feature set, derived through a dedicated signal processing framework and enhanced interpretability of the PA data through Wigner–Ville distribution-based spectrograms, enabling ignition temperature quantification with an accuracy of approximately 89%. In addition, the propensity of coal toward spontaneous combustion is classified into high, moderate, and low SCS categories using convolutional neural network classification, achieving an average accuracy of 93%. The developed system is also applied to raw coal samples toward their thermal property assessment and predicting its SCS. The system is integrated with user interface and can evaluate a coal sample in less than 1 min.