Abstract: The RBF neural network is one of the most widely used neural network. It is used in the pattern recognition of acoustic emission signals generated by the hypervelocity impact of space debris and the spacecraft. The high-speed impact acoustic emission signals are generated by using the numerical simulation software AUTODYN. Some signals are randomly extracted, the magnitude and impact observation points are used as the input parameters, the impact velocity as the output parameter. The RBF neural network is established to inverse the impact velocity of space debris and identify the impact defect types. Calculated results show the effectiveness of inversion to some extent.