Multi-beam fishery sonar is an indispensable equipment in marine fisheries and plays an important role in detecting fish schools. However, the directionality of such sonar is easily affected by wind, waves, and ship attitude. To address this issue, this study established a simplified attitude change model, identified the factors that affect beam stability error, and proposed a beam stability algorithm based on Kalman filtering and digital beamforming technology: first, filter and interpolate the attitude values to obtain smoother and higher sampling rate attitude data; By using coordinate rotation to compensate for the target pointing of the beam, stable and accurate pointing of the sonar beam can be achieved. This study implemented a beam stabilization module based on FPGA (Field Programmable Gate Arrays). The shaking table test shows that under the fourth level sea state, the application of this module can ensure that the maximum deviation of the pointing angle of each beam does not exceed ±1.8°, which is basically consistent with the simulation results. This study has been applied to multi-beam fishery sonar, greatly improving the fishing efficiency of fishing vessels.