论文部分内容阅读
针对双足机器人单脚支撑期控制问题 ,提出了一种新型的模糊神经网络混杂控制方法 .该种方法结合了模糊神经网络、H∞ 控制及逆系统方法的优点 .应用了一种新的多层模糊CMAC神经网络对系统进行逼近 ,一方面将模糊神经网络的构造误差看作系统的干扰 ,利用H∞ 控制对干扰进行抑制 .另一方面利用模糊神经网络对系统模型进行逼近 ,为逆系统的构建和H∞ 控制率的设计提供了有效的系统信息 .并证明了在采用本文提出的模糊神经网络和自适应算法后可以抑制 L2 增益 .
Aiming at the single-legged support period control problem of biped robot, a new hybrid fuzzy neural network control method is proposed, which combines the advantages of fuzzy neural network, H∞control and inverse system. Layer fuzzy CMAC neural network to approach the system, on the one hand, the construction error of the fuzzy neural network is regarded as the system disturbance, and H∞ control is used to suppress the interference.On the other hand, the fuzzy neural network is used to approach the system model, which is the inverse system And the design of H∞ control rate provides effective system information and proves that the L2 gain can be suppressed after adopting the fuzzy neural network and adaptive algorithm proposed in this paper.