基于神经网络的潜艇水面航向控制研究

Research on the Submarine Course Control Based on Neural Network

针对水面航行的潜艇易受到风浪高频干扰而产生频繁操舵的问题,提出了采用直接模型参考的神经网络自适应控制方法.在潜艇航向的离线辨识中引入了参考模型,通过参考模型的输出和潜艇模型的实际输出的比较来调整RBF神经网络的权值,以达到潜艇水面航向的自适应控制,并且针对风浪干扰设计了切比雪夫II型滤波器.仿真结果表明:结合切比雪夫II型滤波算法和直接模型参考神经网络自适应控制算法,能够很好地解决潜艇航向控制在海浪干扰下的无效操舵问题.

According to the problem that the high frequency disturbance of wind and wave induced operating rudders , the direct model with reference to the control means of self-adapting neural network was adopted .The reference model was instructed by identifying off line of submarine＆#39;s course.The weight value of RBF neural network was set by the output of reference model and the actual output of submarine model .In this way, the self-adapting control of submarine at surface course was achieved .And Chebyshev 2nd filter was designed to against wave disturb .The simulation demonstrated that the integration of Chebyshev 2nd filter algorithm and self-adapting control algorithm of direct model with reference to neural network can solve the problem of void steering of course control under wind and wave disturb .