Ship motion online forecasting method oriented to different control working conditions

The invention discloses a ship motion on-line forecasting method oriented to different control working conditions, which comprises the following steps: constructing a three-degree-of-freedom ship motion model, collecting ship motion data sets under different control working conditions through a 360-degree full-task control simulator, including different rudder angles and different engine telegraph orders, and preprocessing the ship motion data sets; constructing a relevance vector machine forecasting model, and optimizing hyper-parameters in the relevance vector machine forecasting model by using a fast marginal likelihood maximization algorithm to obtain an optimized model; acquiring a motion training set from the ship motion data set, updating the motion training set based on a sliding time window method, and training the motion training set by using the optimization model to obtain a three-degree-of-freedom ship motion prediction model; and according to the three-degree-of-freedom ship motion prediction mo