Research On Improved RS-SVM Based On Particle Swarm Algorithm And Intelligent Identification Of Goaf Dangerous Degree

With the consumption and use of mineral resources,there are more and more mining production activities.Year-round mining has brought a large number of and a wide range of mined-out areas to the mines.The mined-out areas pose more and more serious threats to the safety of the mines.Phenomena such as rock dislocation and crack deformation have gradually become dangerous,and the danger of goaf has brought certain safety hazards to mine safety,and has become an important factor in evaluating mine safety.In order to overcome the concealment and uncertainty of the danger of the goaf,this paper improves the rough set knowledge of information entropy and support vector machine theory,and combines particle swarm optimization to construct A risk classification model for goafs is developed,and numerical simulation methods are used to explore the temporal and spatial evolution characteristics of goafs with larger hidden dangers.The main research results and conclusions obtained are as follows:(1)On the basis of statistical analysis and actual measurement data,based on the actual situation of the goaf area of an iron mine in East China,the mining method,goaf excavation depth,goaf height,maximum exposure area of the goaf,exposure height,The 9 factors of maximum exposure span,pillar situation,empty area volume,and treatment rate are used as the main influencing factors.Based on the attribute reduction of rough set theory(RS),the attribute reduction method of improved information entropy is used to reduce the attributes of the training samples.Briefly,the final attribute reduction results are obtained: mining method,maximum exposure area,maximum exposure span,pillar situation,maximum exposure height.To a certain extent,the reduction results can provide theoretical guidance for directional control of mined-out areas.(2)Using particle swarm optimization(PSO)for parameter optimization,selecting the optimal support vector machine kernel parameters and penalty parameters,and compiling the support vector machine model with Matlab,using the "one-against-one" method.Construct a binary classifier to realize the multi-class classification algorithm of the goaf,and finally obtain the support vector machine model for evaluating the risk level of the goaf.The results show that the accuracy of the risk recognition of the goaf based on the support vector machine model Reached 94.0%,indicating the feasibility and practicability of the support vector machine for the risk assessment of the goaf.(3)Introduce a 3D laser scanner to scan the goaf area.Based on the results of the 3D laser detection,use 3D mine to model the goaf area with high hidden dangers.Import FLAC3 D to realize the numerical simulation and analysis of the goaf area.Based on the prediction results,it deeply explored the distribution and development of stress,displacement,and plastic zone in the surrounding rock during the mining process of the goaf with greater hidden dangers,and provided a certain reference for mining resources and disaster prevention in actual projects.,Reflects the practicality of the research content of this article.