【摘要】 为了更好地辨识和控制非线性动态系统,在FNN基础上对其进行优化和改进,形成了动态模糊神经网络(DFNN)。给出了基于BP梯度算法的参数迭代学习算法,并应用于某非线性动态系统仿真试验中。仿真试验表明,该网络比单纯的FNN具有更强的辨识和控制能力,应用于非线性动态系统的控制中可以有效解决系统的非线性和不确定性,提高系统的跟踪性能,并且控制系统具有很强的鲁棒性。更多还原

【Abstract】 A new dynamic fuzzy-neural network (DFNN) is presented, the proposed DFNN combine the advantage of neural network, fuzzy reasoning and feedback. Not only reduce the time of neural network learning, but also can find the optimum rule of fuzzy reasoning. The experimental results demonstrate that the proposed DFNN can effectively improve tracking performance and have robustness against parameter uncertainty and extern disturbance.更多还原