Abstract: The bumping pressure in coal layers is one of the major disasters in coal mine.The accident is caused by many factors.It is ambiguous and dynamic,and expresses as a complex nonlinear dynamic process.This makes data processing system for bumping pressure forecasting can not be processed with the conventional linear system.This paper presented a fuzzy neural network algorithm of multi-source information fusion,which based on the clustering idea of integrating potential field topology,hierarchical clustering algorithm and FCM algorithm and introduced the fuzzy set theory into neural network to form a multi-criteria fuzzy neural network model based on information fusion,and optimized the network.Simulation showed that the model was valid.