The unscented Kalman filter (UKF) is a common weighted statistical linear regression AUV tracking algorithm. Its computational redundancy is lower than those of numerical optimization algorithms such as EKF,PF and PSO, and the tracking efficiency is high. However, the systematic sampling interval in the UKF control method is usually set to a constant value, which may result in the accumulation of errors in the predicted values and influence the accuracy of the navigation prediction results. Therefore, an adaptive unscented Kalman filter algorithm (AUKF) based on AUV tracking is proposed in this paper to reduce the cumulative error of the predicted algorithm. Based on the standard UKF algorithm principle, this prediction method adjusts the sampling interval t of each step in the system state function through the construction of corresponding constraint,judgment and feedback mechanism, so as to improve the tracking accuracy of the algorithm and reduce the influence of the process noise and sensor noise on the prediction process. By simulation experiment and the result comparison, the idea put forward earlier is verified.