A novel ship short-term speed prediction method under the influence of currents

The short-term speed prediction can effectively improve the accuracy and stability of ship motion control, especially for model-based controllers. To acquire accurate speed prediction with specific inputs in future moments during ship sailing, a novel short-term speed prediction method under the influence of currents is proposed. A ship speed model is first built to describe the relationship between the propeller rotation speed and ship navigation speed, which is identified by the least squares method with the moving average filter strategy, i.e., least squares moving average filter (LSMAF). To handle the uncertain currents, a current observer based on PI compensation is designed and embedded in the speed model. Furthermore, an online identification and compensation strategy is designed to reduce the impact of changes in ship load and mechanical conditions, and the step size extended mechanism is proposed to improve the response speed and prediction accuracy. The actual offshore test results show that, compared to the dynamic ship model prediction and inertia prediction method, the average error is reduced by 80.5% and 60.4%, and the mean square error (MSE) is reduced by 93.2% and 84.8%. Moreover, the average error and the MSE are reduced by 36.7% and 77.9% by using the step size extended mechanism.