Abstract: In order to solve the problem that the error of the target positioning result from the vector sensor measurement data increases with the distance, an adaptive unscented Kalman filter is proposed to optimize the positioning accuracy of non-radial moving target in the horizontal direction. Firstly, based on the vector characteristics of the received signal, the azimuth estimation of the target is obtained in frequency domain by using the complex acoustic intensity method. Then, the horizontal distance of the target is matched by using the time delay information of the direct wave and the primary sea surface reflection wave. On this basis, the horizontal distance and azimuth are constructed as the measurement matrix, and the adaptive unscented Kalman filter is established to track and locate the non-radial moving target in the horizontal direction. Simulation results indicate that, under a certain sound source level condition, the vector sensor can provide relatively accurate measurements of target bearing and horizontal range at close distances, and the adaptive unscented Kalman filter method, under favorable initial conditions, can further improve the tracking and localization performance based on the measurements. The experimental data demonstrate that the positioning error of the target at each sampling point is greatly reduced within the horizontal distance of about 15 km, and the horizontal positioning error at the same position can be reduced from 28.65% to 3.86%, and the positioning accuracy is obviously improved, which verifies the effectiveness of the algorithm in this paper.