On The Identification Of Coupled Pitch And Heave Motions Using Support Vector Machine

With the competition of maritime rights and energy in the increasingly global scale, the ship has become the key for our country to develop the sea strategy and resolve the energy problems. But the complex environments of the sea cause many harmful effects, such as the green water on deck, deceleration, and structural damage of ships and so on. These effects ask for that the ship must have good seakeeping. The coupled heave and pitch motions model of the ship is one important topic of seekeeping. The high precision model of ship coupled heave and pitch motion is the basis of the reality of behavior of the ship handing simulator. In this study, support vector machine (SVM) has been researched for the system identification of ship coupled heave and pitch motions model with real oceanic conditions, and the main innovative work as follows:Firstly, the random wave interference model based on random wave theory is developed with considering the influences of the wind speed and fetch length. Then, the ship coupled heave and pitch motions model at different sea states is developed for the simulation experiments of the the proposed identification algorithms. The characteristics of heave and pitch motions responses are analyzed by the simulation experiments. Fast Fourier transform illustrates that the higher of sea states, the wave interference grow stronger.Secondly, the Random Decrement technique is employed for the identification of the ship coupled heave and pitch motions model without the complete knowledge of wave behavior. The mathematical model of coupled pitch-heave motion identified by using SVM is used for predicting the heave and pitch motion responses at different sea conditions. By contrasting with the prediction of neural network, the predicting results present that the proposed identification algorithm based on SVM can overcome the poor generalization ability and the overfitting problem of neural network.The precision of the identification algorithm based on SVM has not affected by damping parameters.Thirdly, the random wave interference force and moment are determined by the random wave interference model based the wave spectrum. Then, a parameteric identification algorithm based on SVM for the ship coupled heave and pitch motions model is presented. By analyzing the identification precision of the equation parameters of coupled heave-pitch motions with different numbers of data and sea states, this paper presents the proper number of data and sea state for identifying. The mathematic model, which is identified with 1000 data and 5 sea state, is used to predict the motion responses at 3,4,5,6 sea state. The prediction results illustrate that the effectiveness and generalization of the identification method based on SVM are very good.