对神经网络和模糊理论在水下机器人运动控制中的应用进行了探讨 ,提出了神经网络多步预测模型的非线性广义模糊预测控制算法 ,用神经网络方法实现了对水下机器人这一非线性系统的在线计算、滚动优化和在线控制 ,采用强化学习方法来构筑模糊控制系统中的神经网络 ,给出了神经网络多步预测模型及相应的控制算法和操作过程 ,计算机仿真结果验证了本文所提方法的有效性 The application of neural network and fuzzy theory in the motion control of underwater robots is discussed. A nonlinear generalized fuzzy predictive control algorithm for multi-step neural network prediction model is proposed. The neural network method is used to realize the nonlinearity of underwater robots The system online calculation, the rolling optimization and the on-line control, adopts the reinforcement learning method to construct the neural network in the fuzzy control system, and gives the neural network multi-step prediction model and the corresponding control algorithm and operation process. The computer simulation results verify that this article The effectiveness of the method