An Improved Autoregressive Method with Kalman Filtering Theory for Vessel Motion Predication

Abstract

It is significant and valuable to improve the time and the accuracy of ships' motion predication. Autoregressive time series analysis method (AR) with Kalman filtering theory is the mainstream currently and the effectiveness for the prediction of ships' motion attitudes have been fully validated. However, the algorithm fixes the order, i.e. the length of the state vector once only, but forecasts the future data for multi-step, resulting in degradation of the step length and the accuracy, especially when the ship sails in bad sea condition. In order to solve this issue, this paper proposed a new autoregressive-multiple (ARm) method, which can determine the orders and the parameters of model in real-time. The method was applied to forecast a ship's motion attitudes in eight different situations. The simulative results of autoregressive-multiple method show the validity and veracity compared with real value.