Study On Prediction Of Tectonic Deformed Coal Thickness Based On Particle Swarm Optimization And Hybrid Kernel Extreme Learning Machine

Coal and gas outburst is one of the main dynamic disasters in the process of coal production.The study shows that in coal and gas outburst mines,the coal seams all develop a certain degree of structural coal,and the greater the thickness of the tectonic coal,the more dangerous the gas outburst is.Therefore,if the thickness of the coal in the coal seam can be accurately predicted,it will play a vital role in the safety management of the coal mine and the development and utilization of the coal bed gas.In view of the problem that the precision of the prediction method of structural coal thickness is not high,the method of particle swarm optimization(PSO)to optimize the limit learning machine is proposed to predict the thickness of the coal.First,using the principal component analysis technology to reduce the dimension of 3D seismic attributes,reduce the dimension of the seismic attributes and eliminate the correlation between the variables.After that,a hybrid kernel limit learning machine model based on the global polynomial kernel function and the local Gauss radial basis function is constructed,and the particle swarm optimization algorithm is used to optimize the kernel parameters of the hybrid kernel limit learning machine.The particle swarm algorithm can easily jump out of the local minimum point and get better results by adding the thought of simulated annealing,the inertia weight decreasing with the number of iterations and the mutation operation based on the random reverse learning.In addition,in order to enhance the generalization performance of the model,the L2 regular term is added to the kernel function,which effectively avoids the effect of noise and outliers on the generalization performance of the model.Finally,the prediction model is applied to the 15# coal seam in the middle part of the Lulan two mining area in Yangquan mining area.The predicted thickness of the mining area is in good agreement with the actual geological data,and the predicted data of the known drilling points are compared with the actual borehole values.In order to further explore the performance of the model,the improved model proposed in this paper is compared with the BP neural network model and the support vector machine model which uses K cross validation to optimize the parameters.Finally,better results are obtained.Therefore,the prediction model proposed in this paper has higher prediction accuracy and smaller prediction error,which can be extended to predict the thickness of tectonic coal in actual mining area.