摘要
针对常用方法对环形球栅扭矩传感器信号的解调效果不明显的状况,提出基于小波神经网络的解调方法。利用非抽样小波变换的时频局部化特性和平移不变性,对预处理后的传感器信号进行多分辨率分解,并在神经网络的干预下进行小波重构。构建了小波神经网络模型,并在LabVIEW中编程实现了该算法。在测试平台中分别对同向转动机械轴、来回换向转动机械轴以及对机械轴敲击时所对应的传感器信号进行解调实验,其结果表明解调的效果好,可靠性高。
Aiming at the status that the signal demodulation effect of ring ball grating torque sensor is not obvious with common methods,a demodulation method based on wavelet neural networks(WNN) is put forward.According to the localization property of time and frequency domains and the translation invariant property of undecimated wavelet transform,the preprocessed signal of the sensor is decomposed with multiple resolutions,and the signal is reconstructed with wavelet under the intervention of neural network.A WNN model is constructed and the algorithm program is realized on laboratory virtual instrument engineering workbench(LabVIEW) platform.With the test platform,demodulation experiments are carried out on corresponding sensor signals when the shaft continuously rotates along the same direction,rotates along alternate directions and is knocked.Experiment results indicate that the demodulation effect and reliability are very well.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2011年第10期2216-2221,共6页
Chinese Journal of Scientific Instrument
基金
国家自然基金(50975300)
中央高校基本科研业务费专项资金(XDJK2009C051)
高等学校博士学科点专项基金(20100191110038)资助项目
关键词
环形球栅扭矩传感器
小波神经网络
非抽样小波变换
ring ball grating torque sensor
wavelet neural network
undecimated wavelet transform