Abstract:

An approach to compensating sensor drift based on recurrent neural networks is discussed in the paper. The features of recurrent neural networks will    be used to improve the measurement accuracy of sensors and to reduce sensor calibration times by drift prediction. Lots of “congener data”will be inte egrated into the “real data”to train RNN and predict sensor drift.

Key words: （recurrent neural network, sensor drift, nonlinear compensation, error prediction）