基于人工神经网络的压力传感器三维数据融合  

3-D data fusion technology of pressure sensor based on neural network

作　　者：钱光耀[1] 杨入超[1] 赵光兴[1] 

机构地区：[1]安徽工业大学电气信息学院,安徽马鞍山243002

出　　处：《传感器与微系统》2007年第2期79-81,84,共4页Transducer and Microsystem Technologies

基　　金：安徽省自然科学基金资助项目(03042309)

摘　　要：针对压力传感器对温度变化和电流波动的交叉灵敏度问题,采用径向基函数(RBF)人工神经网络法对其进行数据融合处理,详细讨论了网络的训练过程和数据融合过程,消除温度和电流对压力传感器的影响。仿真结果表明:当温度变化48.5℃,电流波动3%时,经RBF神经网络数据融合后,压力波动为0.544%,大大降低了交叉干扰,提高传感器的稳定性及其精度,满足在线融合的需要。Aimed at the cross-sensitivity of pressure sensor to temperature flux and the current flux, a data fusion method based on RBF network is proposed for eliminating the influence of two factors on pressure sensor, the training process and data confusion particularly are discussed. The result of simulation shows that for the case of 48.5℃ of temperature flux and 3 % of power flux, the pressure fluctuating is 0.544 %, the cross-sensitivity of system is reduced, the stability and precision of the sensor are increased, it is satisfied of on-line data fusion.

关 键 词：径向基函数神经网络 压力传感器 数据融合 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]