Research on odor prediction methods for coal spontaneous combustion based on E-nose technology

Area of Science:

• Mining Engineering
• Analytical Chemistry
• Sensor Technology

Background:

• Coal Spontaneous Combustion (CSC) poses significant safety risks in mining.
• Early detection of CSC is crucial for preventing catastrophic events.
• Traditional monitoring methods have limitations in detecting early-stage combustion.

Purpose of the Study:

• To develop and validate an electronic nose system for predicting Coal Spontaneous Combustion (CSC).
• To analyze odor characteristics associated with different CSC stages.
• To assess the efficacy of machine learning models in CSC stage prediction.

Main Methods:

• Construction of a Programmable Temperature Coal Electronic Nose Testing Device (PTC E-nose).
• Detection of volatile compounds emitted by lignite during simulated combustion (30-200 °C).
• Application of Principal Component Analysis (PCA) and machine learning models (e.g., PCA-SVM) for odor analysis and stage classification.

Main Results:

• Acetaldehyde identified as a key volatile compound in early CSC stages (feature importance 0.38).
• Acetaldehyde sensor response strongly correlated with coal temperature (R²=0.97).
• PCA effectively differentiated CSC stages, explaining 92.45% of variance; PCA-SVM achieved >95% accuracy.

Conclusions:

• Electronic nose technology provides an effective method for monitoring and early warning of CSC.
• Odor analysis, particularly acetaldehyde detection, is a reliable indicator for CSC prediction.
• The developed system is suitable for high-risk mining environments like goaf areas and fractured seams.