Abstract: To accurately predict the basic top cycle pressure law of a fully mechanized mining face, the gray system theory is used to extract eight significant features that affect the cycle pressure of a fully mechanized mining face; the support vector machine (SVR) prediction model is overly dependent on subjectively selected parameters. Established a support vector machine (PSO-SVR) prediction model for particle swarm optimization parameter selection. The test results show that the mean square error of the PSO-SVM model during the period of compressive strength and step distance is reduced to 47.7% and 74.3% respectively, and the coefficient of determination is increased to 45.7% and 44.6% respectively. To highlight the superiority of the PSO-SVM algorithm model, the most widely used BP general neural network is established for comparison experiments. The results show that the particle swarm algorithm has a significant effect on the performance optimization of the standard support vector machine model, and has obvious advantages over the ordinary BP neural network. Therefore, PSO-SVM has high accuracy and strong generalization for the nonlinear coupling prediction affected by multiple factors. This study can effectively predict the periodic pressure law to provide a reference value for safe and efficient mining in a fully mechanized mining face.