Abstract: The self-radiated noise generated by shell vibration of unmanned underwater vehicle platform greatly degrades the performance of flank array sonar in target detection. To address the issue, a linear interference model is introduced to model the self-radiated noise based on the distribution characteristics. Theoretical noise reduction of different beamformer under the line interference model and point interference model is comparably analyzed. Theoretical and simulation results show that both the noise reduction and the target detection performance of traditional point interference model degenerate badly under line interference conditions. Results conducted by realistic lake trials further validate the effectiveness of the proposed line interference noise model, where the amount of noise reduction of conventional beamformer increases more than 6 dB based on line interference model focusing inverse beamforming preprocessing algorithm when compared with the point noise model in the low frequency band.