Wavelet Transform-based Residual Modified GM(1,1) for Hemispherical Resonator Gyroscope Lifetime Prediction

摘要

The hemispherical resonator gyroscope (HRG) is a new vibration gyro. It has outstanding characteristics of high accuracy, great reliability, long lifespan, and no wear-out, which make it impracticable to get the lifetime with whole life test. To shorten test duration and predict HRG's lifetime, a prediction method combines grey model and grey correlation analysis is proposed in this paper. In the method, wavelet transform is used to reduce noises in original data and an improved GM(1,1) model with residuals is put forward to predict long-term drift data. At last, grey correlation analysis is utilized to evaluate HRG's failure stage and get HRG's lifetime. Applying the method to drift data of 4 different HRGs and the experimental results show the residual modified GM(1,1) model has higher performances not only on data fitting but also on forecasting than the conventional GM(1,1). The predictive lifetimes of 4 HRGs are: 8622 days for 1#, 5748 days for 2#, 7664 days for 3#, and 6707 days for 4#. Based on the 10 global oldest spacecraft (all more than 8760 days), our predictive results are reliable and our method is valid for long-term prediction.