Abstract: As a kind of signal with large time-bandwidth product, linear frequency modulation (LFM) signal is widely used in active sonar for underwater direction of arrival (DOA) estimation. Fractional Fourier Transform (FRFT) has unique advantages for LFM signal analysis. In this paper, the feasibility of its application to the field of underwater multi-targets DOA estimation is explored based on the theory of FRFT. For the possible increased errors of optimal order searching in the FRFT, norm constrained Capon beamforming (NCCB) is studied in this paper, and it is used to improve the robustness of the DOA based on FRFT preprocessing. Finally, fractional order preprocessing and robust beamforming algorithm are combined in the paper. Computer simulation proves that the NCCB based on FRFT preprocessing can still achieve robust underwater multi-targets DOA estimation in low SNR.