The Improvement Of The AR Method In Tow Parallel Sailing Ships' Roll Attitude Motion Forecast

Ship sailing replenishment don't only provide logistics guarantee for improving cruising ability,but reduce the supply ship in and out of the busy port that at the same time reduce the ship's directly dependence on the station and raised the navigational safety.In recent years,the ship underway replenishment problem widely attracted attention.The alongside replenishment is one of the commonly used supply modes,the process of which is affected by the weather factors outside such as waves,wind and so on,that causes the swaying problem between the two ships which brings the lateral supply a certain risk.Due to the effect factors of ship is relatively complicated and varied,the integrated ship posture prediction system used to be the multidimensional chaotic model,which based on the single dimension linear prediction model.Theoretically,to improve the existing linear forecast model,and applied to the comprehensive forecast system,will be able to promote the forecasting accuracy and the forecast time in the standard precision of the comprehensive forecast system effectively,strengthen early warning function and improve the safety of ships in the process of supply.When the forecast results reach a higher standard,it can apply comprehensive forecast system to ship heave compensation control system,support automation control attitude correction of the ship.This paper carried out the following research work basing on a large number of literature at home and abroad related information access and the combination of the domestic actual situation:The first part of this paper expounds the present situation of the research on the online prediction and simulation technology about the ship attitude motion and the development and application of the corresponding related technology at home and abroad.Then gives the quantitative analysis to the characteristics of ship motion attitude.For analyzing the statistical characteristics of double ships' rolling motion time series data,introduce the test methods about the stationarity of ship roll motion time series data,such as autocorrelation and partial autocorrelation,ADF and so on,to make the linear features clear.At the same time,introduce some common methods to smooth the non-stationary data in.case.Finally,get the result that double ships' rolling motion time series data is very suitable to be analyzed by linear methods,especial the AR model,and the stability is good.What's more,the AR models are established on the two ships individually through the known data features using prior knowledge without considering the influence of the ship close interactions between pairs of roll motion.In the order fixing process,determine the most appropriate rule for the double ships' roll motion through the comparison and analysis of the several conventional order-fix methods' effect in the forecast model.And the effect of the order to the forecasting model is affected by the selected learning step size,so we get the suitable range of learning step size through the study of the effect's trend diagram in multiple conditions step length and choose to applied AIC criterion to decide order.Finally,the forecasting results of model are tested and analyzed.Finally,study the physical factors that affect the ship motion posture and analyze the special response relationship in the cases of double ship related factor data analysis for use.Gives quantitative correlation calculation on the interaction of the double ships' rolling to verify the size of the impact factor effect,at the same time calculate the delay through the analysis of the wave theory in the hydrodynamic aspect.Considering the impact factor data,build a special form of improved AR linear forecast model with the application of the improved AIC criterion to fix the order.Then use the delay computed with the hydrodynamic method and corrected with mathematical statistics method for forecast respectively,and compare the results of the predictions with the AR method.The compare that the improved AR model is better than the AR model,and the improved AR model using the cha corrected with mathematical statistics method is a little better than the one using the hydrodynamic theory in most cases.At last,the feasibility of using the double ships' roll attitude motion time series data for each other to predict is verified.