Prediction Model of Surrounding Rock Stability of Roadway and Its Maturity Tolerance

For the problems of immaturity and low accuracy of prediction methods of surrounding rock of roadway, we introduced support vector machine（SVM） algorithm and took the evaluation indexes of surrounding rock stability of roadway as the input and its stability levels as the output. A prediction model based on PSO-MKSVM was established by applying particle swarm optimization（PSO） algorithm to adjust the values of penalty factor c, kernel parameter g, and single kernel coefficient λ₁ and λ₂. The precision tolerance P and cost tolerance E under the unequal prices were defined to weigh the prediction results. Then the maturity tolerance M was confirmed as a criterion for assessing model performance. The experimental results showed that multiple sets of λ₁ and λ₂ values make the maturity tolerance M of prediction model of roadway surrounding rock stability reach the maximum value of 1, and the performance is stable.