Self-optimizing dynamic position control of UUV in the underwater transportation

In the Underwater Transportation processing, the fixed set-point control of a dynamic positioning system enables an unmanned underwater vehicle (UUV) automatically to move to a set position by means of a control system, as well as to maintain its precision when deviating from that set position. In response, this paper aims to improve its generalization ability by indirectly pruning the structure of neural networks proposes a novel neural network with a penalty term (P-KWNN). It then goes on to establish a dynamic positioning (DP) system model and introduces a P-KWNN algorithm so as to tune the parameters of the PID controller. Finally, this paper transforms the controller output into an actual torque by introducing a proportional switching unit, thus achieving the adaptive dynamic positioning of UUV.