The ship motion prediction approach based on BP neural network to identify Volterra series kernels

Ship motion prediction plays a prominent role in the whole ship motion process. This paper presents a new approach for ship motion prediction. In order to obtain more effective prediction result, the paper studied the BP neural network and Volterra series model, and the chaos characteristics of ship motion time series. A novel method of single-output three-layer BP neural network to identify Volterra series kernels is proposed. Multi-step prediction for the roll motion time series of ship at 135° with the method is accomplished. The simulation analysis demonstrate that the ship motion prediction approach based on BP neural network to identify Volterra series kernels has higher precision, longer prediction time, effectiveness and adaptability, and it can predict the ship motion exactly.