Owing to the nonlinearity and uncertainties of the system, big noises of sensors in the harbor, and the precise control of automatic underwater vehicles (AUVs) in some intelligent operations have been puzzling us in the application of AUVs. Process Neural Network is an extension of traditional neural network, in which has increased a for time get together operator, makes network at the same time have two space – time dimensional information processing ability. So imitate the information processor system of biological neuron better. Both inputs and outputs of AUVs motion control system are processing vector which related with time. Based on basic process neuron, integrating S function and pre-planning idea, a process neuron of AUVs control is introduced. At the parameters learning phase, gradually reducing chaotic noise is added to the networks to form a powerful globe optimization algorithm.Simulation results show that new control is a useful control algorithm for underwater vehicles, which features higher precision, simpler design, quicker response, and better robustness.