Prediction Of Ship Motion Based On The Cerebellar Model Articulation Controller

Ship motion is the main reason of a variety of sea accidents. In order to increase the carrier-based aircraft landing safety factor, improve the combat effectiveness of ships and reduce the occurrence of maritime accidents, the short-term prediction of ship motion is essential. The ship motion in irregular waves is a non-linear random process, traditional time-series analysis and prediction theories are on the basis of the linear model, and it is an effective method for linear systems, but does not apply to the time-series modeling and prediction of non-linear system. So the study of non-linear system modeling prediction method is of important significance.Cerebellar Model Articulation Controller (CMAC) is applied in the short-term prediction for ship motion posture for the first time in this paper. A real-time CMAC prediction model is established, at the same time, it has been done the numerical simulation and analysis of errors. As a conclusion, the content of this paper is as follow:1. CMAC's composition, operation mechanism and its main features are elaborated. It has the similar composition with perceptron, and it tends to simulate the cerebellum functions of vertebrate animals.2. Through researching the features of ship motion and the cerebellum functions, it has been found that they have the same quick response as conditioned reflix, and the same ability of local generalization. Based on these characteristics, the paper establishes a real-time CMAC prediction model of ship vertical motion, shows the whole modeling process and gives each parameter settings.3. Through the multiple data simulations, the results show that the prediction model can basically predict the development trend of real data, and the predicted data also generally tally with the real data. we have a good result that on the error of 0.2 degree, the model can predict 4-13 seconds for ship motion, and the relative MSE is less than 15 percents, which shows that the prediction model presented in the paper is reasonable and feasible.