The Research On The Application Of RS And LSSVM In The Prediction Of Water Inrush From Coal Seam Floor

Water inrush accidents occur frequently in mine of our country,which causes huge losses,and it is a serious threat to the coal mine staff's life,property safety,coal mine economic benefits and the coal industry sustainable sound development,especially affects the coal mine safety production of the mine production base in North China coal field.Among them,the accident of water inrush from coal floor is the most common and harmful water hazard accidents,of which the sudden and destructive is extremely strong.Special hydro-geological conditions of coal mine in our country cause too many influencing factors of water inrush from floor water inrush and the complex relationship among the factors,so many traditional methods for predicting water inrush from coal seam floor have not been able to achieve the ideal effect in practice.Therefore,to achieve the accurate prediction of coal seam floor water inrush risk is the key to effective prevention and control of water disasters,and it has been a major research topic in the field of coal mine safety and the major research issue of the hydro geological workers.In this paper,two algorithms of rough set(RS)and least square support vector machine(LSSVM)are combined to establish the risk prediction model of coal seam floor water inrush based on RS and LSSVM.The continuous attributes discretization method based on rough set theory and fuzzy C-means clustering algorithm is used to discrete the continuous data from the original samples of coal seam floor water inrush.The knowledge acquisition of the information of water inrush from coal seam floor in the decision table after discretization is obtained by using RS theory,which was used to delete the redundant information,then the decision table was implified,the quality of samples is improved under the premise of keeping the classification ability and data integrity unchanged.The influence factors of water inrush from coal seam floor after RS attribute reduction were used as the input vectors,and the LSSVM was used to train the samples after reduction,then choosed appropriate kernel function,and the parameters of LSSVM were optimized by k-fold cross validation.At last,the model was used to predict the water inrush from coal seam floor.At the same time,a comparison of the prediction results with the direct LSSVM,support vector machine and radial basis neural network was made.The experimental results show that the prediction model of water inrush from coal seam floor based on RS and LSSVM has a better training speed and higher prediction accuracy,and it also has good generalization and generalization ability.It has a good application prospect that to apply RS and LSSVM method to predict water inrush from coal seam floor.