Abstract: In order to correctly predict the damage degree of coal seam floor, the genetic algorithm (GA) and particle swarm optimization (PSO) algorithm have the problems that optimization support vector machine (SVM) is easy to fall into the local optimal solution and the classification accuracy is relatively low. GAPSO-SVM is proposed. The parameters of SVM are optimized by considering the advantages of GA and PSO algorithms. The optimized algorithm can better adjust the balance between the global and local search capabilities of the algorithm. The prediction of the damage degree of the bottom plate of Caozhuang Coal Mine shows that the method can not only achieve good classification effect, but also has higher classification accuracy than GA-SVM and PSO-SVM, and has better robustness.