摘要
当气动肌肉与环境发生碰撞时,冲击会引起内腔压力变化。首先分析了轴向和径向冲击下内腔压力波的传播;采用主成分分析法降低压力响应信号的特征维数,并将主成分作为BP神经网络的输入;然后搭建实验平台,测试分析了不同负载、内压下、轴向和径向冲击下压力响应曲线;基于测试样本对3层BP神经网络进行轴向和径向冲击方向感知训练。实测表明径向冲击下压力响应比轴向剧烈,BP神经网络方向感知精度高于95%。
When the pneumatic muscles and the environment collide,the impact will cause the cavity pressure change.Firstly,the pressure wave spread under axial and radial shock is analyzed,the principal component analysis method is used to reduce the pressure response of the characteristic dimension of the signal,and the principal component is taken as BP neural network＇ s input.Then the experimental platform is built to test and analyze the different load,internal pressure,axial and radial impact pressure response curve.Based on test samples,3 layer BP neural network for axial and radial direction of the impact perception is trained.The experimental results show that the radial pressure response is more intense than the axial＇ s,and the accuracy of the BP neural network direction sensing is higher than 95%.
出处
《测控技术》
CSCD
2016年第5期20-24,共5页
Measurement & Control Technology
基金
国家863计划项目(2015AA042302)
浙江省自然科学基金项目(LY14F030021)
关键词
气动肌肉
压力响应
主成分分析
BP神经网络
感知
pneumatic muscle
pressure response
principal component analysis
back-propagation neural network
perception
