Modeling of nonlinear parameters on ship with fuzzy CMAC neural networks

An intelligent model for the ship´s nonlinear parameters was established based on fuzzy cerebellar model arithmetic computer (FCMAC) neural network. Firstly, the system design comprises the structure determination, and then applies the least square estimation with adaptive learning rate to train the mean and variance of the membership functions and the weights of FCMAC. With the learning algorithm, a well-parameterized FCMAC can be achieved for the required performance. Secondly, with the experimental data of HD702 ship, a research based on FCMAC was done on hydrodynamic parameters´ nonlinear function of three dimensional space, resulting in a nonlinear parameter model which can self-adaptive to change with different navigating speed, ocean condition, and course. Finally, simulation results indicate that the modeling method with FCMAC has high speed and high accurate, with the error rate below 10%. And the algorithm is proved to be effective.