Prediction Of Ship Motion State Based On Digital Twin

With the in-depth integration and development of new-generation information technologies such as cloud computing,Internet of Things,big data,artificial intelligence and virtual reality,and traditional manufacturing,the concept of digital twin emerges as the times require.As an emerging technology,the digital twin model has self-learning,self-adapting,and self-evolution capabilities.These features enable the use of digital twin technology to maintain a high-fidelity mapping of the physical system between the virtual space and the physical space.Closed-loop connection of data between virtual.At the same time,the vigorous development of the domestic shipbuilding industry and foreign maritime trade in recent years has made the activities of ships on the waterway more intensive.Therefore,in order to improve the safe and efficient navigation of ships under the constraints of the above-mentioned dynamic environment and multi-task scenarios,this paper proposes a ship motion state prediction method based on digital twins,and uses the ship motion twin model to realize the motion state prediction of ships.Firstly,the concept of digital twin requires the iterative interaction between the twin model and the physical system.In this paper,the numerical simulation method is used to construct the digital model of the physical ship.This method makes the design and optimization of the ship motion twin model more convenient and fast,and saves the development cost..In this paper,the digital model of the physical ship is constructed first from the point of view of the ship’s motion mechanism,and guided by the modeling idea of the separated model,a mathematical model of the ship’s maneuvering motion with three degrees of freedom is established.For the marine environment disturbance mathematical model,the wind and wind moment coefficients are estimated by Blendermann’s empirical formula method to construct the wind disturbance mathematical model and the wave disturbance mathematical model based on the P-M wave spectrum.Using the constructed ship motion model,the ship 10°right-turn maneuvering simulation test and 10°/-10° Z-type maneuvering simulation test were carried out in calm deep water and wind and wave environment without wind and waves,and it was verified that the established model conforms to the actual motion condition.It lays a foundation for the subsequent use of the ship motion twin model for digital ship motion prediction and digital ship course keeping experiments.Secondly,a three-degree-of-freedom motion twin model of the ship is designed from the perspective of the mechanism model.The viscous force coefficient and the thrust coefficient of the propeller in the model are used as the identification targets.The extended Kalman filter algorithm and the support vector machine identification algorithm are used to identify the optimize.Among them,the extended Kalman filter algorithm is improved based on the sliding window recursive identification,which makes full use of the historical information data of a certain length of time for recursive identification,which can make the parameter identification achieve higher accuracy;Constraint Cb is added to the bias term b in the identification algorithm to make the results of the identification parameters more accurate.Using the navigation data simulated by the digital ship combined with the two parameter identification algorithms and the ship’s three-degree-of-freedom motion twin model,the digital ship’s three-degree-of-freedom motion forecast for the next 10 seconds in the still water environment and the wind and wave environment was carried out,and the twin model was optimized in real time.The prediction results were evaluated by MSE index,and it was concluded that the ship three-degree-of-freedom motion twin model optimized by the extended Kalman filter algorithm was better than the ship three-degree-of-freedom motion twin model optimized by the support vector machine algorithm,and the differences are analyzed.Finally,the ship heading twin model is designed according to the forward differentiation of the second-order response model structure.Using the improved extended Kalman filter algorithm and the support vector machine algorithm combined with the ship heading twin model,the heading prediction was carried out on the simulated data and the real ship data.The optimized twin model of ship heading is more accurate in heading prediction.The paper also uses the predicted heading information to design the self-correction control law of the steering signal,and conducts a heading keeping simulation experiment on a digital ship in a general environment with disturbances,which verifies the effectiveness of the ship heading control based on the ship heading twin model.