Research And Realization Of Mine Water Source Identification Algorithm

Coal is the main energy source in China,accounting for 95%of China's mineral energy resources.And according to statistics,at present in our country,according to data in the coal production process in a major accident,caused by mine water bursting of casualties and property losses are always located in all kinds of the front of the mine disasters,therefore in the mine water prevention and control work,the recognition of mine water is the basis of the work,in the mine water prevention and control work plays the role of nots allow to ignore,The key to solve the problem is to establish the identification model of mine water source and determine the source of mine water source quickly and accurately through the research on the method of water source discrimination.In order to solve the above problems,this paper chooses the hydrochemical analysis method Piper trigram,the traditional water source discrimination Bayes discriminant method and the emerging artificial intelligence field neural network discriminant method to study.Among them,Piper trigram is to analyze and classify the chemical composition of mine water sources.T raditional bayesian criterion is according to the maximum probability of watery attribution species belong to determine,at the same time for traditional bayesian model of water quality information superposition between ion and heavier limitations such as subjective factor by using principal component analysis and optimization variant weights for fusion,the experiments show that the optimized model accuracy from 92.31%to 96.15%;The generalized regression neural network(GRNN)is proposed to establish the mine water source discrimination model.The model is simple and has fewer adjustable parameters.The optimized fruit fly algorithm is used to adjust the parameters(smoothing factor)in the activation function of the generalized regression neural network(GRNN)to replace the manual setting of the original model parameters.After that,the algorithm model before and after optimization was experimented and the mine water source discrimination model was established.The experiment showed that the optimized GRNN algorithm converges faster,the local optimal value is improved,and the discrimination accuracy is significantly improved.On the basis of the above algorithm research,the software implementation of the algorithm is carried out.Most of the existing mine water source identifi cation algorithms are only in the theoretical stage,and there are few mine water source identification systems that can be delivered to users and can achieve visual mine water source models and integrate multiple algorithms.Therefore,the paper uses C#language to extract the main models through PyChhand 2020 software,and writes relevant call codes through Visual Studio 2013 to realize the asynchronous call of the three algorithms to achieve all the functions described in the paper,and to complete a mine water source type discrimination platform integrating multiple algorithms,which is convenient for users.