Study On Discrimination Of Mine Water Source Based On Principal Component Analysis Weighted Som Neural Network

Mine water inrush accident is one of the five major disasters in coal mines,which seriously threatens coal mine safety production.The main types of mine water damage include Ordovician confined water,old kiln water,pore fissure water and so on.The severity of water damage depends on the nature of water inrush source.Therefore,fast and accurate identification of mine water source is of great significance for mine water disaster prevention and control.The paper takes the water inrush water source as the research object.Through the theory of mine water inrush related,this paper discusses the feasibility of distinguishing the mine outburst water source according to the hydrochemical characteristics.After the statistics of the hydrochemical characteristics of the mine water source,the relevant indexes to distinguish the mine water source are selected,and then the water sample data is treated with the principal component analysis theory.After weighting the main component factors of mine water samples extracted by the principal component analysis and reducing the dimension,the SOM(Self-Organizing Feature Map)neural network is put into the neural network,and the mine water source discriminant model of the principal component weighted SOM network is established.The recognition of the ion ion water sample verifies the reliability of the method.Based on that,we develop water control strategies and establish relevant databases.Research shows:(1)The SOM neural network can improve the accuracy of the mine water source with more variable characteristics and strong correlation between variables by using the training samples of the neural network to reduce the dimension of the principal component analysis.(2)Compared with the traditional SOM neural network,the principal component weighted SOM neural network has the advantages of convergence speed and error control.(3)Principal component analysis weighted SOM Neural netwo'rk can successfully identify the main components of mine water source with low mixing degree,which can be applied to most cases of actual mine water discrimination.(4)The principal component weighted SOM neural network is applied to the source discrimination of Yuxing coal mine in Inner Mongolia,which can accurately identify the source of water inrush and formu]ate corresponding water control countermeasures.