Terahertz Spectroscopy and AI Enable Coal-Rock Interface Detection

Why It Matters

The technology delivers high‑precision, non‑radioactive sensing that can dramatically improve safety and productivity in intelligent mining, replacing hazardous gamma‑ray systems.

Key Takeaways

• THz‑TDS + Random Forest reaches 96% detection accuracy.
• Optimal frequency band 0.7‑1.3 THz for coal‑rock discrimination.
• Transmission mode works best up to 30% coal concentration.
• Enables real‑time adaptive cutting, lowering tool wear and collapse risk.

Pulse Analysis

Intelligent mining faces a persistent challenge: accurately locating the coal‑rock seam in harsh underground environments. Conventional acoustic, gamma‑ray, and visual systems struggle with dust, humidity, and vibration, often compromising safety and efficiency. Terahertz spectroscopy bridges the gap by combining microwave penetration depth with infrared sensitivity, offering a non‑ionizing, high‑resolution alternative that can operate through dense coal dust and rock matrices.

The recent study published in *Photonics* demonstrates how THz‑TDS data, when distilled through principal component analysis, can feed robust machine‑learning models. Among the four algorithms tested—SVM, LS‑SVM, ANN, and Random Forest—the latter achieved a 96% correct classification rate, especially within the 0.7‑1.3 THz window where C‑H and C‑C bond resonances are most pronounced. While transmission‑mode sensing loses fidelity above 30% coal content due to signal attenuation, the approach remains highly effective for early‑stage mixing and rock‑dominant interfaces, delivering sub‑8 ms response times.

For the mining sector, the implications are immediate. Embedding THz‑TDS sensors into robotic shearer arms enables instant seam detection, reducing unnecessary rock cutting, extending tool life, and mitigating roof‑collapse hazards. The non‑radioactive nature simplifies compliance with safety regulations, and the rich spectral data can feed digital‑twin platforms for real‑time operational optimization. Future work on reflection‑mode THz probes and multimodal sensor fusion promises to extend coverage to high‑coal seams, positioning terahertz technology as a cornerstone of sustainable, automated mineral extraction.