A Gas Prominence Prediction Model Based on Entropy-Weighted Gray Correlation and MCMC-ISSA-SVM

To improve the accuracy of coal and gas prominence prediction, an improved sparrow search algorithm (ISSA) optimized support vector machine (SVM) based on Markov chain Monte Carlo (MCMC) filling prediction model were proposed. The mean value data after in missing values using MCMC was 2.282, with a standard deviation 0.193. Compared fill method (Mean), random forest (random forest, RF), K-nearest neighbor (K-nearest neighbor, KNN), showed best results. parameter indicators salient ranked by entropy-weighted gray correlation analysis, experiments divided into four groups different numbers according to correlation. results obtained fourth group, maximum relative error (maximum error, REmax) 0.500, average (average MRE) 0.042, root square (root RMSE) 0.144, coefficient determination (coefficient determination, R2) 0.993. predicted parameters initial velocity dispersion (X2), content (X4), K1 desorption (X5), drill chip volume (X6). (sparrow algorithm, SSA), adaptive t-distribution variation operator introduced obtain ISSA, models (MCMC-ISSA-SVM), (MCMC-SSA-SVM), genetic (MCMC-GA-SVM) particle swarm optimization (MCMC- PSO -SVM) established for SSA, (genetic GA) (particle optimization, PSO) respectively. Comparing experimental each model, MCMC-ISSA-SVM is 98.25%, 0.018, convergence speed fastest, number iterations least, fitness are highest among models. All significantly better than other three models, which indicates that improvement effective. ISSA outperformed PSO, GA, able effectively enhance generalization ability.