Artificial Neural Network And PD Hybrid-control Of Dynamic Positioning System

As human beings continue to explore areas of the ocean, the demands on the position and attitude of the ship are highly increase. Dynamic positioning systems (DPS) have become a key technology maintain the position of vessels in the field of deep-sea. Measurement system, control system and propulsion system are included in the Dynamic positioning systems. The control system is the core part of the DPS.Offshore platform support vessel is taken as the research object in this paper. The dissertation mainly study the hybrid neural network and PD control algorithm to improve immunity and robustness of dynamic positioning control system. Focuses on the three degrees of freedom surface vessels dynamic positioning system model, including ship models, storm flow model ambient interference and the Kalman filter. And then verify the accuracy of the model in MATLAB/Simulink simulation environment.On the basis of the depth study of the feed forward neural network learning algorithms and network structure the A hybrid control algorithm is proposed based on PD and RBF neural network. Dynamic positioning system simulation program is compiled by MATLAB programming language. Nonlinear PID controller and RBF-PD hybrid controller are carried out on board ship motion control simulation of offshore platform support vessel. The results prove the effectiveness of the algorithm.The study of this dissertation proves the hybrid control algorithm based on PD and RBF neural network is effective to improve immunity and robustness of dynamic positioning control system.