Development and application of a comprehensive experiment for prediction coal spontaneous combustion on electronic nose technology

Coal spontaneous combustion fires are a major hazard threatening coal mine safety. Accurate and efficient prediction technologies are essential for underground mine fire prevention and control, with odor analysis emerging as an important research area for the early warning of coal spontaneous combustion. Owing to its advantages, including rapid response, short detection times, long monitoring periods, low power consumption, high sensitivity, simple operation, and portability, electronic nose technology has been widely applied in food safety supervision, environmental monitoring, clinical diagnosis, and chemical safety evaluation. To facilitate the reform of interdisciplinary experimental teaching and broaden senior undergraduate students' understanding of cutting-edge scientific research, this study introduces electronic-nose sensing technology into the prediction of coal spontaneous combustion and develops a comprehensive experimental course entitled Prediction Method of Coal Spontaneous Combustion Odor Based on Electronic Nose Technology. The experimental design enables students to independently complete literature reviews, experimental scheme design, and result analysis, thereby stimulating their enthusiasm for scientific exploration and cultivating their innovative awareness and comprehensive scientific literacy.

A self-developed coupled experimental system integrating programmed temperature rise and electronic nose detection was constructed. The system comprised three functional modules: a programmed coal-sample temperature control unit, a gas supply unit, and an electronic nose signal detection unit. A teaching framework integrating instructor guidance with group-based collaborative experimentation was adopted. Students were divided into multiple groups to conduct programmed heating tests that reproduced the complete evolutionary process of coal spontaneous combustion. Characteristic odor signals released during coal oxidation were captured using the electronic nose system. Following data preprocessing and quantitative analysis, radar charts illustrating the relationship between odor response signals and coal temperature were generated. The coupling relationship between odor characteristics and temperature variations was quantified, and linear discriminant analysis was implemented to distinguish various oxidation stages of coal spontaneous combustion. Finally, an odor-driven prediction model for coal spontaneous combustion was established.

The self-built integrated platform for programmed heating and odor signal collection enabled real-time monitoring and feature extraction of volatile odor signals released during the thermal oxidation of coal. The system effectively tracked the dynamic evolution of typical gaseous products, including aldehydes, ketones, benzene homologs, and alkanes, throughout the coal oxidation process, providing a stable and reliable experimental platform for analyzing odor signatures induced by coal spontaneous combustion. Significant correlations were observed among electronic nose sensor responses, coal temperature, and oxidation stages; specifically, the Spearman correlation coefficient between acetaldehyde response intensity and coal temperature reached 0.92. Meanwhile, the oxidation stage classification model constructed using an artificial neural network achieved an overall recognition accuracy of 91%, confirming the reliability of the proposed method for identifying different stages of coal spontaneous combustion.

Two semesters of teaching practice involving seven student groups verified the effectiveness of the proposed experiment. By integrating multidisciplinary theories and techniques related to coal spontaneous combustion monitoring, sensing technology, and artificial intelligence, the experiment demonstrates highly comprehensive, forward-looking, and exploratory characteristics. It effectively improves students' abilities in literature screening and analysis, experimental equipment operation, and data processing. Furthermore, it strengthens their skills in academic writing, research result organization, and oral presentation; stimulates their motivation to address practical engineering problems using emerging technologies; and improves their innovative thinking and teamwork abilities. Satisfactory outcomes have been achieved in undergraduate experimental teaching. Future research could expand the range of coal samples and further optimize the generalizability and engineering applicability of the classification models, thereby continuously updating and enriching the content of experimental teaching.