Adaptive Prediction of the Motion of Marine Vehicles

A predictor of the future motion of a vessel subject to random wave and wind forces, would have a variety of applications in ocean engineering. Most previous work has assumed that the wave spectrum is known and that the vessel is modeled accurately; both factors affect the predictor performance strongly. In practice, the relevant data is difficult to measure on a manoeuvering vessel and can change significantly with operating conditions. Here were describe the application of an adaptive algorithm which predicts the future of a signal from its history. The predictor adapts to the signal and varies its parameters to optimise the prediction. Operating on a signal with a stationary spectrum, the predictor tends to a steady performance; if the spectrum changes, the predictor quickly adjusts to the new situation. We illustrate the performance of the system with examples.