Abstract: In this paper, a Fractal Brownian Motion(FBM) model of ship radiated noise was established based on the study of FBM. Under this model, the fractal dimension of ship radianed noise is extracted,and the Fractal Feature Vector (FFV) is constructed in the Fractal Brownian Random Field(FBRF). By using these parameters, a new method for feature analysis is presented in the underwater acoustic target recognition. The comparative experiment of classification shows that the FFV can be used as important parameters to compensate for the shortcomings of other feature extraction methods.