Machine Learning Based Safety Assurance System And Its Application With Mine Water Inrush

The water inrush will often cause significant casualties,huge property damage,and even environmental pollution that is difficult to recover during mining.In order to prevent mine water damage accidents,various countries and mine enterprises have attached great importance to the safety of water disasters.On the basis of conventional prevention and control,informational and intelligent methods were constantly tried to solve the problem of water inrush.It makes the hydrological data explode during mining in the context of digital mines.Therefore,how to fully mobilize mine hydrological data to better serve mine water inrush safety is the focus of this dissertation.1)Establishing of mine water inrush knowledge base.The mine water inrush knowledge base consisted of three parts:the mine basic hydrological database,the hazards inventory database,and the cases of water inrush database.Firstly,the basic hydrological database was mainly used to store the monitoring data generated by the daily monitoring equipment of the mine,such as the water temperature,water level,and water inflow,which was used to meet the data needs of the late data outlier mining.Secondly,the hazards inventory database refers to the domestic and foreign laws,regulations,policy documents,normative standards,and investigation reports of water inrush accidents that occurred in the past.With reference to the three-dimensional matrix evaluation of accident risk and the analysis of hidden dangers in other industries,the hazards of mine water inrush were analyzed from the three dimensions of exposure rate,controllability,and severity.This was combined with the three-dimensional matrix evaluation and mine production life cycle to form a hazard classification database.Through the preliminary investigation of the mine water inrush accident investigation reports issued by the national safety supervision department,extracting the basic information of the accident unit,the accident occurrence,the emergency rescue situation,the cause analysis of the accident,and the preventive measures,the database of mine inrush case was estabilished.2)Establishing mine hydrological data mining.Using the basic hydrological data in the knowledge base,combined with unsupervised learning based on machine learning,a multi-dimensional hydrological data outlier mining method was formed.Through the data mining of self-organizing neural network(SOM)for different types of one-dimensional hydrological data and the cluster analysis of DBSCAN integrated multi-dimensional hydrological data,the model for judging the abnormal fluctuation of hydrological number was finally formed.On the one hand,a large amount of valuable unmarked hydrological data was fully used.On the other hand,high reliability data outlier mining could be realized with large amounts of hydrological data.3)Establishing mine water inrush case recommendation.Using the mine water inrush case data in the knowledge base,based on machine learning recommendation algorithm and case-based reasoning theory,a case recommendation method was established for mine water inrush disaster control,disaster situation disposal,and post-disaster rescue.By combining the cosine similarity in the recommended system method with the case-based reasoning theory,on the one hand,the comparison mechanism of different mines was established;on the other hand,the method of forming structured knowledge and experience of historical water inrush accident cases were summarized.Quickly match the mines that meet different geological environments and mining conditions,and realize the effective use of knowledge and experience in previous water inrush cases.4)Establishing a cloud platform for mine water inrush safety.Through the analysis of Hadoop and cloud technology,the service content,framework,and operation mechanism of the mine water inrush security cloud platform were proposed.The knowledge base,outlier mining,and case recommendation were integrated to the mine water inrush security cloud platform,to achieve multi-directional,multi-angle,and systematic supply of mine water inrush safety guarantee.