【作者】 刘宏达； 马忠丽；

【Author】 Liu Hongda Ma Zhongli (College of Automation, Harbin Engineering University, Harbin 150001 ,China)

【机构】 哈尔滨工程大学自动化学院；

【摘要】 结合隔离小生境技术的特点和免疫遗传算法的全局收敛特性,提出一种基于小生境免疫遗传算法的模糊神经网络控制器优化设计方法,并将其用于船舶自动舵的设计．隔离小生境能有效解决局部和全局收敛的矛盾,可以解决船舶运动控制的FNN控制器难于设计以及传统BP学习算法易于陷入局部收敛的不足．将神经网络结构辨识器作为船舶的辨识模型,船舶运动模型采用二阶非线性方程,模糊神经网络控制器为主控制器,对船舶航向运动控制系统进行仿真研究．结果表明,在船舶无干扰和存在随机干扰的情况下,由此设计的船舶自动舵控制系统均能使船舶转向控制无超调,跟踪快,与BP学习算法相比控制效果更趋于为理想,可以进一步提高航行的性能．更多还原

【Abstract】 Considering the characteristics of isolation niche and global convergence of immune genetic algorithm, a kind of new method to design FNN(fuzzy neural network) controller is discussed, and the method is used to design ship autopilot. Isolation niche can solve efficiently some problems followed: conflicts of convergence between global and local, difficulties of design FNN controller used to control ship movement, and local convergence of classical BP algorithm etc. Regarding artificial neural network identify implement as ship identify model, two orders nonlinear equation as ship movement model, FNN controller as main controller, the results of simulation is given, which can show that there is random disturb or not in ship course, the system designed can control ship course without overflow and quickly, quality of system is improved farther.更多还原