Coal Mine Gas Risk Prediction Based on Bn-Elm in Human Reliability Analysis

In the process of coal mine safety production, there are both quantitative risk data and qualitative risk information (such as behavioral risk). Qualitative risk information is difficult to participate in the calculation of early warning, which brings great challenges to improve the accuracy of coal mine risk early warning. Therefore, taking gas as a case, this paper studies the coal mine risk prediction and early warning methods including behavior information. Build BN-ELM (Bayesian network extreme learning machine) model to uniformly quantify behavior risk and gas data. The correction parameters are set to optimize the BN-ELM model. Combined with the control chart, the coal mine safety situation awareness model is constructed, and the risk management and control application software are designed and developed. The results show that the error of gas data prediction is reduced by 0.007, the error of risk value prediction is reduced by 0.01, and the error of safety situation value prediction is reduced by 0.03. This paper considers the behavioral factors of gas risk, and provides research methods and application tools for risk management of coal mining enterprises.