Abstract: A new type of robust fault diagnosis strategy based on nonlinear state observer is presented in this paper. In this method, a nonlinear fault detection filter (FDF) is constructed for a class of Lipshitz affine nonlinear system on the basis of traditional FDF method, the algorithm of detection gain is given and the stability of closed-loop error system is proved. Furthermore the equivalence between nonlinear FDF and linear FDF is proofed and the fault detectability is guaranteed. At the stage of residual decision, an adaptive fuzzy threshold method is used to enhance the robustness of fault detection. Finally, this method is applied to the fighter's structure fault detection, simulation results show that the presented method is effective.