Research on a Coal Seam Gas Content Prediction Method Based on an Improved Extreme Learning Machine

With the rapid advancement of artificial neural network (ANN) algorithms, many researchers have applied these methods to mine gas prediction and achieved numerous research achievements. It is great significance study that can accurately predict content for prevention disasters in mining areas. In order enhance accuracy, stability, generalization capability model, GASA-KELM model was established using GASA algorithm improve KELM initial parameter assignment method, based on BPNN SVM under same conditions. The experimental results show GASA-BPNN failed achieve desired outcome within 800 iterations. On other hand, GASA-SVM models accomplished goal significantly fewer iterations, taking only 673 487 respectively. Moreover, overall average relative errors cross-validated predictions were 15.74%, 13.85%, 9.87% three models, Furthermore, total variance test set 3.99, 2.76, 2.05 GASA-BPNN, GASA-SVM, As a result, compared with ANN demonstrates higher stronger ability practical application. This novel provides basis predicting proposing effective regional management measures.