Prediction of Quartz Differential Resonant Accelerometer Zero-bias Based on Long Short Term Memory Neural Networks

摘要

In this paper, a predictive method of zero-bias is proposed for quartz differential resonant accelerometer (QDRA) based on long short-term memory neural networks (LSTM). The stationary and non-linear trend sequence was extracted from nonstationary zero-bias series by stationary processing of time series, then LSTM model was selected for modeling and predicting. Experimental results indicated that the root mean square error (RMSE) was 9.2252e-04, and the short term stability of 0g was improved from 301.3$mu$g to 53.6$mu$g.