一种新型人工神经元振荡器的设计与应用  

Design and Application of a Novel Artificial Neural Oscillator

作　　者：张代兵[1] 沈林成[1] 胡德文[1] 

机构地区：[1]国防科学技术大学机电工程与自动化学院,湖南长沙410073

出　　处：《机器人》2007年第6期581-585,共5页Robot

摘　　要：提出一种包含两个神经元和简单连接关系的新型人工神经元振荡器.分析了振荡器的动力学特性,指出振荡的收敛速度、幅度和频率分别由动力学方程中的三个参数独立控制.采用该振荡器设计了鱼形机器人的仿生神经网络控制系统.实验表明该系统能够对起动、停止、直线巡游等动作进行良好控制,实现了仿生机器人对动物从"形似"到"神似"的进步.This paper presents and implements a novel artificial neural oscillator which consists of two neurons and simple connections. The dynamic analysis on the oscillator indicates that the oscillatory convergence speed, amplitude and frequency are individually controlled by three parameters of the dynamic formula. A bionic neural network control system based on the oscillator is designed to control a fish-robot. Experiments are carried out to validate that the neural system can effectively control several motions such as start, stop and cruise, and it promotes the robotic imitation of the living being to progress from ＂similar in shape＂ to ＂alike in spirit＂.

关 键 词：神经元振荡器 仿生神经网络 鱼形机器人 运动控制 

分 类 号：TP271[自动化与计算机技术—检测技术与自动化装置]