Research On The Method Of High-precision Prediction Of The Movement Attitude Of The Ship Within A Short Time

A ship will be affected by uncertain sea conditions such as waves,sea breeze and ocean currents,when it sails on the sea.It will inevitably produce a six-degree-of-freedom swaying motion with mutual coupling,which poses a great potential safety hazard for the navigation and operation of the ship at sea.Among them,the effects of roll and pitch are the most serious.Therefore,if we can predict the movement attitude of the ship within a short time(several or ten seconds),it will greatly improve the safety and stability of ship operations at sea.On the basis of modeling and analysis of ship motion,this paper takes the rolling and pitching motion as the research object,and deeply studies the classical algorithm of prediction of the movement attitude of the ship widely used at present.Based on this,the combined prediction model of EEMD-IPSO-SORR is established to realize high-precision prediction of the movement attitude of the ship within a short time.Firstly,this paper analyzes ship motion,and introduces the six-degree-of-freedom swaying motion of the ship in detail.The ship lateral motion and longitudinal motion are deeply studied to get the ship state equation of lateral and longitudinal motion.The perturbation characteristics of the random wave are analyzed in detail,and the wave spectrum is decomposed.The model of wave inclination is established to simulate and analyze the random wave disturbance signal.On this basis,the simulation analysis of the lateral and longitudinal motion of the ship is realized.Secondly,the classical prediction algorithm of the movement attitude of the ship is studied.The basic principles of autoregressive and Kalman filter prediction algorithms are introduced.The autoregressive theory and the parameter estimation method and the ordering criterion of the model are introduced in detail.A method of the recursive least squares parameter estimation is used to establish the recursive prediction model to realize the predictiction and simulation of the movement attitude of the ship.The Kalman filter theory is introduced in detail.And the colored noise of the system must be whitened.And the dimension of the state equation is expanded by the substituted Kalman filter algorithm,which is the extended Kalman filter algorithm.Then the prediction and simulation of the movement attitude of the ship is realized by the extended Kalman filter algorithm.Because of the nonlinear and non-stationary characteristics of the attitude of ship motion,this paper establishes a combined prediction model of EEMD-SVR based on ensemble empirical mode decomposition and support vector regression for predictive and analysis.Thealgorithm of empirical mode decomposition is used to smooth the original time series data,and an improved method of collective empirical mode decomposition(EEMD)is proposed to overcome the deficiency of standard empirical mode decomposition.The decomposed components are classified and predicted by using the support vector regression(SVR)algorithm.Finally,the final predicted value of the movement attitude of the ship is obtained by weighted summation of the results of prediction.Finally,based on the combined prediction model of EEMD-SVR constructed in the previous paper,the improved algorithm of particle swarm optimization(IPSO)is used to optimize the parameters,and an improved algorithm of support vector machine regression is put forward,which is the algorithm of continuous over-relaxation support vector regression(SORR).Then a new method of prediction of the movement attitude of the ship,namely the combined prediction method of EEMD-IPSO-SORR,is obtained.In this paper,the roll and pitch attitude data of a certain type of ship under the three-level and five-level sea conditions are used for simulation experiments.The AR model,SVR model,the combined prediction model of EEMD-SVR,and the combined prediction model of EEMD-IPSO-SVR are used to predict and simulate for different training length and prediction length.At the same time,the prediction accuracy and simulation time of the four methods of prediction under the same training length are compared.The results indicate that the combined prediction model of EEMD-IPSO-SVR has the charcateristics of fast learning speed and high accuracy for the prediction of the movement attitude of the ship within a short time.