基于RBF网络和遗传优化的船舶操纵模糊控制器

Fuzzy Controller for Ship Steering Based on RBF Networks and Genetic Algorithms

设计了一种基于 RBF网络和遗传优化的船舶操纵模糊控制器。首先讨论了传统模糊控制器应用于船舶操纵控制的不足 ,然后根据模糊系统在特定情况下与 RBF网络具有等价关系的特点 ,采用具有加权平均输出的 RBF网络构造了一个船舶操纵模糊控制器 ,有效地消除了小偏差范围的舵角抖动现象。在此基础上 ,根据船舶操纵的特点提出了一种尺度变换因子的自整定方法 ,并采用遗传算法对自整定过程中的可变参数进行优化 ,以使控制器能够适应实时控制过程中的时变性和不确定性 ,保持良好的控制性能。最后针对某大型船舶的非线性模型 ,采用 Matlab6.1的 Simulink工具进行了转艏操纵仿真试验 。

A fuzzy controller for ship steering based on RBF networks and genetic algorithms is designed. Firstly, the drawbacks of conventional fuzzy controllers for ship steering are discussed. Then, based on the equivalence relation between fuzzy system and RBF networks under certain conditions, an alternative method is proposed using RBF networks with weighted average output, which shows the advantage of eliminating the tremble of rudder within small error regions. Further, a new method is also proposed for the self tuning of scaling factors and genetic algorithm is employed to optimize the para meters used in tuning process in order to keep good controlling performance in case of time varying and uncertainties. Finally, with the nonlinear ship model of a mariner level vessel, simulation tests are carried out using Simulink tools of Matlab 6.1 and results are satisfactory.