【摘要】 利用模糊神经网络的模糊推理能力以及前馈神经网络的逼近能力 ,将其与自适应控制方案结合 ,并取带有控制增量约束的广义目标函数作为优化指标 ,从而推导出一种能对非线性非最小相位系统进行有效控制的模糊神经网络间接自适应控制器。在网络学习算法上采用带有动量项的BP算法。仿真结果表明了该方法的有效性。更多还原

【Abstract】 In this paper, an adaptive control scheme is combined with neuro fuzzy networks which can human like reason and with feed forward neural networks which can step by step approach and a general object function with constraints of control increments is employed as optimization index, leading to a PI adaptive controller by which systems with non linearity and of non minimum phase can be controlled effectively. BP algorithm with momentum term is used for training the neuro fuzzy networks serving as controller and the multi layer feed forward networks as identifier. Simulation results have demonstrated the effectiveness of this controller.更多还原