A fast discriminant model of water inrush source based on improved Fisher discriminant method
In order to effectively and quickly determine the location of water inrush source and prevent the occurrence of mine water dis-aster accidents,according to the actual hydrogeological characteristics of No.5 Coal Mine of Pingdingshan Coal Mine,34 groups of water sample data were collected according to the difference of water quality in sandstone,limestone and cold ash aquifers,and six convention-al hydrochemical ions K+ +Na+,Ca2+,Mg2+,SO42-,Cl- and HCO3- were selected as discriminant factors to analyze the hydrochemical characteristics.An improved Fisher water source discrimination model was established by combining PCA dimensionality reduction sta-tistical algorithm,and the discriminant results of the Fisher model before and after the improvement were compared by using the samples to be tested.At the same time,the training samples were substituted back into the improved model for verification.The results showed that cold grey water and tai grey water cannot be distinguished accurately according to hydrochemical type.The improved Fisher discrim-inant model was used to test 10 groups of samples to be tested,and the discriminant accuracy was 100％.Compared with the basic Fish-er model,the accuracy improved by 20％.The application of the improved Fisher discriminant model can greatly improve the accuracy of water source identification.The back substitution results of the known training samples show that the improved Fisher discriminant re-sults were basically consistent with the actual situation.Through the comparative analysis of the two models,the improved Fisher model for mine water source identification has high accuracy and reliability,which has certain research value and can provide new ideas for mine water source identification.
water inrush sourceFisher discriminant modelwater source identificationprincipal component analysis
NETL
NSTL
万方数据
