Abstract: The tangent of major influence angle tanβ is one of the most important parameters for mining subsidence prediction with the probability integral method，and it determines the influence range of mining subsidence.In order to improve calculating accuracy of tanβ，and based on analysis of tanβ and its influence factors，5 main influence factors on tanβ as inputting layer neuron are selected.Combining PSO algorithm of quick searching the global optimal solution with RBF neural network，a PSO-RBF neural network prediction model is proposed，and the nonlinear mapping relationship between tanβ and mining and geological conditions is obtained.Then，data from 30 typical observation stations are used as learning and training sample to test the fitness and generalization of PSO-RBF neural network model.The predication results of the PSO-RBF neural network and the observation values are analyzed and compared with each other.The results show that：adopting PSO-RBF neural network to calculate tanβ，the rate of convergence is rapid，with high prediction accuracy.The prediction result of maximum relative error is 6.54%，the minimum relative error is 2.56%，and the accuracy of tanβ is improved to some degree.

Key words: PSO-RBF neural network, Prediction model, Tangent of major influence angle