The fault prediction of weapon is getting more difficult for its complex structure, unique operating environment and multi-plicate faults. Currently although the main fault prediction methods have achieved certain success in practical application, they all fall short in some aspects. Based on the discrete grey prediction theory and with an analysis of the disadvantages of the discrete grey model, an adaptive prediction model with several characteristic parameters for small samples is proposed. This model modifies the initial value and takes into account the interrelations of the parameters and characteristics of prediction series. The data of a certain aircraft engine are taken as an example for prediction and analysis, and the results showed that the model had high availability and precision.