论文部分内容阅读
模糊神经网络控制已经成功应用于水下机器人运动控制中,但其运算过程和训练算法比较复杂,对嵌入式硬件要求也较高。根据带翼水下机器人的运动特性提出了S型模糊神经网络控制方法,并推导了网络权值学习算法,最后以XX水下机器人为研究对象进行了仿真实验。试验结果表明,与基于高斯型隶属函数的模糊神经网络控制器相比,在没有过多损失整体控制品质的情况下,其网络算法得到极大简化,运算速度得到了提高,反应能力增强,非常适用于对精确定位能力和运动速度要求不高,但要求高机动性的水下机器人。
Fuzzy neural network control has been successfully applied to the underwater robot motion control, but its operation process and training algorithm is more complex, and requires high embedded hardware. According to the motion characteristics of the winged underwater robot, a S-type fuzzy neural network control method is proposed and the network weight learning algorithm is deduced. Finally, a simulation experiment is carried out using XX underwater robot as the research object. The experimental results show that compared with the fuzzy neural network controller based on Gaussian membership function, the network algorithm can be greatly simplified without any loss of overall control quality, the computing speed is improved and the reaction ability is enhanced. It is suitable for underwater robots that do not require precise positioning ability and movement speed, but require high maneuverability.