为解决模型预测控制器(MPC)在执行器约束下初始状态量偏差大导致的无解问题，设计一种基于级联思想的MPC控制方法：首先，针对欠驱动AUV模型进行级联系统划分，再对划分出的模型分别进行MPC控制器设计；然后，利用微分跟踪器将期望量平滑处理；最后，基于非线性干扰观测器对外界干扰进行补偿。仿真结果验证了该方法的有效性。

In order to solve the problem of model predictive controller (MPC) with large initial state deviation under actuator constraints, a cascade based MPC control method was designed. Firstly, cascade system was divided according to the underactuated autonomous underwater vehicle (AUV) model, and then MPC controllers were designed for the divided models. Secondly, the expected quantity was smoothed by differential tracker. Finally, the nonlinear disturbance observer was used to compensate the external disturbance. Simulation results verify the effectiveness of the proposed method