Condition Monitoring of Coal Mine Using Ensemble Boosted Tree Regression Model

In recent years, Fires and explosion in coal mines imposes number of life threats for mine workers along with a rapid increase in environmental air pollution. By using various risk assessment methodologies, coal miners can easily predict the potential risks of forthcoming hazards in advance. In this work, a novel approach is proposed for monitoring the fire-resistant hydraulic fluids (HFA) contamination level. Fire resistance property of HFA fluids varies with the viscosity. Water content. By monitoring the water content in HFA fluids, fire resistance can be easily predicted. Fire resistance hydraulic fluid properties are trained in Ensemble Boosted Regression Tree (EBRT) to predict the potential risk in coal mines. EBRT is the supervised training algorithm which is proposed for leveraging an efficacious coal mine monitoring into existence. EBRT model estimates stronger prediction by linearly integrating the weaker estimations. Threshold rule-based decision making is adopted for the effective mitigation of risks. EBRT is optimized to minimize the cross-validation loss. Furthermore, Bayesian optimizer is used to minimize the objective function to 7.81 with regularized parameter lambda is chosen as 0.34 to minimize the ensemble trees. The root Mean square error is optimized to 31.68.