采用神经网络数据融合改善传感器的静态特性

Static State of Sensor Using Data Fusion by Neural Network

为解决在全桥差动的电桥测量中,环境温度的变化对因桥臂之间的特性差异所造成的测量误差影响显著,而这种影响却又无法用一般的数学式子进行描述的问题,利用神经网络可以学习的功能,将电桥的两个输出电压信号作为标定数据,采用神经网络对标定数据进行处理,从而既提高了电桥测量的环境温度适应范围,也提高了其测量精度.经过神经网络的数据融合,测量误差可由原来的15%下降到2%.

The objective of this work is to extend the limit of the environment temperature and precision of bridge measurement. Neural network can learn from sampled data by using two output voltages of the bridge. The data fusion is then used for calibration. Precision tuning method is also used to train the network. The same sensor can adapt to different limits for the environment temperature.