Wavelet neural network adaptive learning algorithm and its ship control application

A wavelet neural network approach is proposed for ship control purpose using nonparametric regression estimation method. The proposed wavelet neural network estimator makes use of residual based selection algorithm for hidden units´ selection and adjusts the connecting parameters with back propagation method. It is capable of handling nonlinear regressions of sparse training data. Ship course control simulation examples illustrate the satisfactory performance of this proposed approach.