EUVIMEDObjevte EUVIMED
Open access · CC BY 4.0

INTELLIGENT EARLY-WARNING AND DECISION-MAKING FOR COALFIELD FIRE RISK BASED ON RANDOM FOREST: A CASE STUDY OF ABANDONED GOAFS IN NORTHERN CHINA

Rui Sun, ZiXue Wang

Frontiers in Environmental Research · 2026-05-09 · 4 (1) · 64-72

DOI: 10.61784/fer3050 ↗

At a glance

Journal article in Frontiers in Environmental Research, published online on 2026-05-09. Written by Rui Sun and 1 co-authors. The full text is freely available (open access, CC BY 4.0).

AbstraktBibliografické údajeCiteRelated studies

Abstract (original language)

Coalfield fire risk identification is challenged by the strong concealment of goafs, the development of surface fissures, the unclear distribution of residual coal in abandoned workings, and uncertainties associated with multi-source detection data. Conventional approaches based on a single temperature threshold or expert empirical judgement are therefore insufficient for accurate early warning. To address field problems observed in an abandoned goaf of a coal mine in northern China, including temperature anomalies, abnormal CO concentrations at small-mine openings, smoke emissions from surface fissures, and the potential hydraulic or fracture connection between upper and lower goafs, this study develops an intelligent coalfield fire risk early-warning model based on the random forest algorithm. Thermal infrared imaging, isotopic radon measurements, and geological-goaf spatial information were used as the primary data sources. Surface temperature anomalies were extracted from UAV-based thermal infrared images, gas-migration signals along underground fracture pathways were obtained using alpha-cup radon measurements, and spatial-distance features to Jurassic abandoned-goaf boundaries and small-mine openings were quantified. Using expert interpretation results and field-observed anomaly points as labels, a random forest classification model was established to quantitatively classify coalfield fire risk levels in the study area. The results show that thermal anomaly features directly represent surface temperature responses, radon anomaly features effectively reveal concealed fractures and gas-migration pathways, and geological and goaf-related spatial features characterize the controlling conditions for fire development. Fusion of the three types of features substantially improves the stability and interpretability of coalfield fire risk identification. The proposed model provides a scientific basis for delineating temperature anomalies in abandoned goafs, optimizing verification borehole layout, zoning fire prevention and control measures, and ensuring safe longwall panel retreat.

Cite

APA

Sun, R., & Wang, Z. (2026). INTELLIGENT EARLY-WARNING AND DECISION-MAKING FOR COALFIELD FIRE RISK BASED ON RANDOM FOREST: A CASE STUDY OF ABANDONED GOAFS IN NORTHERN CHINA. Frontiers in Environmental Research, 4(1), 64-72. https://doi.org/10.61784/fer3050

Vancouver

Sun R, Wang Z. INTELLIGENT EARLY-WARNING AND DECISION-MAKING FOR COALFIELD FIRE RISK BASED ON RANDOM FOREST: A CASE STUDY OF ABANDONED GOAFS IN NORTHERN CHINA. Frontiers in Environmental Research. 2026;4(1):64-72. doi:10.61784/fer3050

BibTeX
@article{Sun2026,
  author = {Sun, Rui and Wang, ZiXue},
  title = {INTELLIGENT EARLY-WARNING AND DECISION-MAKING FOR COALFIELD FIRE RISK BASED ON RANDOM FOREST: A CASE STUDY OF ABANDONED GOAFS IN NORTHERN CHINA},
  journal = {Frontiers in Environmental Research},
  year = {2026},
  volume = {4},
  number = {1},
  pages = {64--72},
  doi = {10.61784/fer3050}
}
Start for free →