| Coal resources have always been the main and important basic energy in China,and are expected to remain the most critical industrial raw materials in China for a long time in the future.However,in the current coal mine safety production work in China,many coal mines still rely on the outdated geological information before mining,ignoring that the underground geological structure will change greatly with time.If the types of ground and underground structures are not clearly identified and positioned in the process of coal mining,it will lead to major coal mine production accidents,resulting in huge casualties and property losses.It is a new development trend of interpretation and prediction of geological structure and underground water bearing area to comprehensively analyze the multi-attribute data obtained by three-dimensional seismic exploration and the attribute data obtained by electrical exploration and carry out the fusion research based on multi-attribute and multi-source data.In recent years,the cross integration of artificial intelligence,machine learning and other technologies with the field of geological exploration has promoted the rapid development of related technologies and instruments.Multi-source information fusion has become an important research direction in the field of geological exploration in recent years.At the same time,the progress of technology makes the geological information that can be mined more and more abundant,and also provides sufficient data support for the current geological structure research.Based on the national key R &D plan support project,this study mainly uses the seismic attribute data and resistivity attribute of the exposed area obtained from the field exploration of Xinyuan mining area in Yangquan,Shanxi Province to carry out the research on geological structure identification and water yield identification.Through field exploration in the early stage,it is known that there are mainly geological anomalies such as faults and collapse columns in this area,which has a great impact on underground mining.The seismic attribute information obtained from exploration after processing can provide a favorable reference for the identification of geological abnormal bodies in the process of geological exploration.However,previous work experience and research show that there are problems of multiple solutions and uncertainty in the prediction of ground and underground abnormal structures by a single seismic attribute.Therefore,this study selects the seismic multi-attribute fusion technology for the identification and prediction of geological structures.Firstly,after comparing the prediction results of four algorithms: random forest,decision tree,logistic regression and gradient boosting decision tree(gbdt),the geological structure recognition model constructed by random forest algorithm is selected.Secondly,based on the random forest algorithm of classical grid search,the random forest algorithm is optimized by optimizing the parameters of random forest parameters.Based on the optimized random forest algorithm,a variety of seismic attributes are fused to build a geological structure recognition model.The model is used to predict the faults and collapse columns in the target mining area.The prediction results show that the optimized algorithm model has higher accuracy.At the same time,the visualization results of prediction data show that the model has better recognition accuracy for faults and collapse columns.Finally,based on the optimized random forest algorithm model,multiple seismic attributes and resistivity attributes are fused.Finally,three seismic attributes and resistivity attributes are retained for multi-source information fusion through correlation analysis and attribute importance analysis,and the water abundance of geological structures is identified.The experimental results show that the multi-source information fusion model can better reflect the water abundance of geological structure and the boundary of water bearing area than a single resistivity attribute data recognition,and provide a new idea for the prediction of water abundance of geological structure in coal mining. |
