Affiliations: [a] College of Safety Science and Engineering, Xi’anUniversity of Science and Technology, Xi’an, China
| [b] Institute of Safety and Emergency Management, Xi’an University of Science and Technology, Xi’an, China
Abstract: Aiming at the problem of large error and long time of early warning response in the traditional system, this paper designs a hazard identification early warning system based on random forest algorithm in underground coal mine. By random classification decision forest created dangerous content in different areas of the downhole information input into the decision tree as a test sample, according to the result of the output of the leaf node determine the risk level of decision trees, and USES the high precision of decision forest classification ability the threat level assessment test sample, radically reducing hazards identification error. Then, based on the evaluation results, combined with the threshold value of warning criteria to identify the gas exceeding limit area, and determine the fire source warning level, so as to realize the hazard source identification and warning. The simulation results show that the average hazard location identification error of the system is only 4.1%, and the warning response time can be controlled within 9 s.
Keywords: Underground hazard sources, identify early warning, random forest algorithm, decision forest, risk assessment, alarm criteria threshold
