A New Identification Algorithm Of Ship Motion Mathematical Model Based On Nonlinear Innovation

With the rapid development of large scale and intelligent ships,the conventional models of ship have been found their difficulties in depicting its precise motion.Furthermore,the problem of motion modeling and identification for different types of ships has always been a difficulty in the field of navigation research.Therefore,it is important to investigate the interaction mechanism of ship motion and seek a mathematical model identification algorithm of ship motion with high accuracy for the development of high-quality navigation simulator and the design of high-performance controller,which is in line with the strategic needs of the national development of a strong transportation country.Firstly,from the perspective of identification theory,this thesis analyzes the dominant factors affecting the identification accuracy of different ship types and different mathematical models,and improves and reconstructs the traditional integral Norrbin mathematical model,the separated ship motion mathematical model MMG(Maneuvering Model Group)model and the Nomoto mathematical model.The constraint conditions of power series expansion in the simplified derivation process are analyzed,and the inaccuracy of Nomoto mathematical model simplification derivation process from second order to first order is further corrected.The scope of application of Nomoto mathematical model is clarified,and the model accuracy is improved.According to the navigation practice and theoretical analysis,the force and moments related to the yaw rate and yaw acceleration will increase after the ship is large-scale.At this time,the four hydrodynamic derivatives related to it are amplified.The improved Norrbin mathematical model is more suitable for ships,while retaining the advantages of the Norrbin mathematical model with less requirements on ship parameters.To make it more effective and versatile on ships.In this thesis,the nonlinear innovation(the abbreviation of new information)system identification algorithm as the framework combined with the extended Kalman filter algorithm,carried out the research of ships motion mathematical model parameter identification,to explore the efficient way to reduce the parameter drift.Firstly,a nonlinear innovation identification algorithm is proposed.The full-scale trial is carried out by the research ship“Yukun” of Dalian Maritime University to provide data support,verify the effectiveness of the algorithm and improve the identification accuracy,and establish a 4-DOF(degree of freedom)mathematical model that can accurately reflect the actual ship dynamics.Then combined with extended Kalman filter,forgetting factor and other methods,the ship real-time identification and attitude prediction are carried out.A 4-DOF identification and attitude prediction method for ships under actual sea conditions is established.The results of this research can be widely applied to the identification and modeling problems of intelligent ship design,large ship maneuvering simulator and highly reliable intelligent navigation equipment development.The establishment of parameter identification,attitude prediction and identification model are the prerequisite for the safe,fast and stable navigation of large ships and intelligent unmanned ships.In this thesis,a 4-DOF ship identification model is established by proposing a nonlinear innovation identification algorithm and introducing a tangent function to deal with errors.The problem of inverse of multi-innovation matrix in the traditional multi-innovation least squares algorithm is avoided,and the improved nonlinear innovation identification algorithm can identify the model structure and hydrodynamic derivatives simultaneously.In order to avoid the problems of many erroneous data or duplicate data leading to poor prediction accuracy and inefficient identification,an improved extended Kalman filter(EKF)algorithm is proposed based on the nonlinear innovation identification algorithm combined with the forgetting factor.The forgetting factor was introduced to reduce the cumulative influence of the historical interference data,and the extended Kalman filter was used to optimize the parameters in the motion model to reduce the estimation error of the minimum variance,reduce the fitting time and improve the fitting degree.The large inertia coupling problem is solved and the algorithm is more suitable for ship parameter identification and maneuverability prediction.It is a guideline for practical engineering applications.All experiments in this thesis are implemented using Matlab and Visual Basic,and the experimental results also verify the reliability of the algorithm.The results obtained will be directly applied to the optimization and improvement of the simulator in author’s laboratory and other core key technologies,to give full play to the spirit of independent innovation and make contributions to the national marine development and transportation power strategy.