Research On Prediction Of Coal Mine Gas Emission Based On KPCA-IAFSA-ELM Algorithm

Gas disaster accidents not only seriously threaten the lives of coal mine personnel,but also bring huge economic losses to enterprises,therefore,it is very important to prevent gas d isasters.Precisely and quickly predict the amount of gas emission,and make gas prevention and control measures in advance,which are of great significance to mine gas control and coal mine safety production.Gas emission prediction is a multi-dimensional nonlinear small sample data predictio n problem,which has many influencing factors and complex relationships between them.In order to study the specific relationship between the influencing factors of gas emission,18 influencing factors were selected to establish an initial predictive index system according to the source-division forecasting method and selection principle.Through the non-linear characteristic analysis of the data collected in the test mine,the influencing factors of gas emission have the characteristics of non-linearity and overlap of information.Spearman's rank correlation coefficient method was used to clarify the correlation strength between each influencing factor and the amount of gas emission.The kernel principal component analysis method was used to reduce dimensionality,the optimal kernel parameters were determined,and the characteristic indexes and model input variables were clarified.In order to realize the accurate and quick prediction of gas emission,in view of the slow convergence speed of the artificial fish swarm algorithm and the low accuracy of the optimal solution in the later stage,the Levi flight step was proposed to improve the parameters,and the artificial fish swarm algorithm based on Levi flight was obtained,the optimization effect was verified through the multimodal function optimization test experiment,and the results show that the optimization accuracy and optimization speed have been improved.Aiming at the problem that the weight and threshold parameters in the extreme learning machine were initially randomly generated,the IAFSA algorithm and the extreme learning machine were coupled to obtain the IAFSA-ELM prediction algorithm,and through the UCI classic data set prediction experiment to verify the feasibility of the prediction,the results show that the prediction algorithm is feasible and credible.the nuclear principal component analysis method and the IAFSA-ELM prediction algorithm were combined to construct a gas emission prediction model based on the KPCA-IAFSA-ELM algorithm,and the graphical user interface design tool in MATLAB software was used to developed a universal coal mine gas emission prediction software.Forecasting software was used to carry out example applications in test mines,a mine in Hunan and a mine in Shenyang,The results show that the prediction relative error of the split-source prediction method is about 10.27%,and the prediction accuracy of the IAFSA-ELM algorithm is about 12.8 times higher than that of the split-source prediction method,and the prediction accuracy of this prediction algorithm is higher than that of the traditional method.The number of iterations of the prediction algorithm has also dropped significantly,the prediction difference basically does not exceed±0.4,and the average prediction relative error does not exceed 4%,this further shows that the IAFSA-ELM algorithm has fast prediction efficiency,high prediction accuracy,and stable prediction effect,and the prediction software has a certain degree of compatibility,It can realize the accurate and quick prediction of gas emission,which has important on-site guiding significance for the prevention and control of gas disasters in coal mines.