Comparison and Optimization of Monitoring and Warning Methods for Coal Spontaneous Combustion Fire Area

This is the authors' abstract. We don't add key points for this paper.

Coal spontaneous combustion fire is a major hidden danger in coal mine safety production, and accurate monitoring of fire area status is the key to prevention and control. At present, fire zone state recognition based on dynamic changes of gas indicators is the most commonly used monitoring method. However, due to the complexity of the fire area environment, the application of single index, composite index and machine learning model lacks systematic comparison and evaluation, resulting in the lack of scientific basis for the selection of monitoring indicators and early warning models, and it is difficult to accurately grasp the evolution process of fire areas. In this study, by comparing the measured gas data with the temperature, the best gas index suitable for different fire area scenarios was screened out, and the corresponding temperature prediction model was constructed. The results show that the established model achieves accurate prediction of fire area temperature within the average relative error range of 2.367%-5.720%, and the multi-feature temperature prediction model shows the best accuracy and reliability. From index optimization to algorithm adaptation, this study provides a precise and intelligent solution for coal mine fire area management, which helps to reduce fire risk and improve safety production level.