摘要
针对压力传感器对温度存在交叉灵敏度这一具体问题,采用LM算法对其进行数据融合,消除温度对压力传感器的影响,大大提高传感器的稳定性及其准确度;针对传感器存在测量误差问题,提出了增加一个噪声来训练网络,增强了网络的容错性。对所提方法进行了仿真研究和简单分析,并与非线性插值(三次多项式插值)作了比较,粗浅地指出:神经网络在数据融合过程中的实现机理。
LM algorithm is adopted to solve the problem of intercross sensitivity of pressure sensor to temperature and to perform data fusion disposal.The method eliminates the influence of the temperature on pressure sensor and improves the stability and accuracy of the sensor.The effect is proved to be fairly well.Considering the measurement error of the sensor,a noise can be added to train the network to improve its fault tolerance.Emulation research and simple analysis are taken in this method,and through comparing whit non linear interpolation algorithm (3 order polynomial interpolation),some mechanisms when implementing neural network in data fusion are pointed out.
出处
《传感器技术》
CSCD
北大核心
2005年第5期77-79,85,共4页
Journal of Transducer Technology
关键词
BP神经网络
传感器
插值
容错性
LM算法
BP neural network
sensor
interpolation
fault tolerance
LM algorithm
