Time-delay wavelet network predictor based on sensitivity analysis with application to predictive ship course control

摘要

An improved time-delay wavelet neural network (WNN) is proposed to represent the complex nonlinear and time-varying dynamics of ship motion based on sensitivity analysis approach. To improve the generalization performance of WNN, inputs of the wavelet network are selected based on their relative contribution to the overall output. To evaluate the contribution of inputs in the WNN, an index is proposed referred to as relative contribution rate (RCR). The resulted network is utilized as an online ship motion predictor. Based on the predictor, a predictive PID controller is presented and implemented in a ship course-following control. Simulation of online predictive ship course-following control was conducted and results demonstrate the feasibility and efficiency of the WNN predictor and the WNN-based predictive control strategy.