Design of neural network observer for ship dynamic positioning system

A neural network observer is developed for ship dynamic positioning system with uncertain dynamics and disturbances in this paper. For ship dynamic positioning system, in most cases the position measurements are available but the velocity measurements are not. In addition, the measured position and heading signals are often corrupted by noises, which may reduce the performance of the dynamic positioning system. The designed observer is presented to estimate all the state of the dynamic positioning system, including position and velocity signals from the corrupted position signals. Based on Lyapunov theory, it is proved that the observer estimation errors are uniformly ultimately bounded. It is also shown that the designed neural network observer is capable of producing noise-free estimates of velocity and position from corrupted position measurement. An example is given to show the performance and effectiveness of the proposed neural network observer.