Abstract: Aiming at the complexity of airflow temperature prediction in the mine and the nonlinear relationship of the various factors, traditional forecasting method is difficult to construct the forecast model, which results in low prediction accuracy. A prediction method of mine airflow based on RBF neural network is proposed, and the parameters of RBF neural network by using particle swarm optimization algorithm are optimized. Simulation study on the prediction model by coal mine historical data is carried out. The results show that the proposed method in this paper based on improved particle swarm optimization algorithm of RBF neural network model (MPSO-RBF) has the characteristics of fast convergence rate, high precision prediction, which provides theoretical support for mine airflow temperature forecast fields.