A GARCH Modeling Approach for Constant Speed Drive Residual Life Predicting of Aircraft Generator

The estimation of devices states and the prediction of condition developments are always utilized in projects by monitoring time-series data. For solving the problems of non-stationary and time-varying variance in practical data, generalized autoregressive conditional heteroscedasticity (GARCH) model is discussed, compared with traditional time-series model, in this paper in order to gain a better estimation and a greater predicting effect. In the paper GRACH model is applied to state estimation of constant speed generator drive (CSD) of aircraft for the residual life prediction. Finally the suitability of the proposed method is examined by simulating experiment data. And it is necessary in practical projects to analyze the relations among sample data, prediction period and precision. The simulation results show that the residual life prediction error of CSD lies within 10% as expected.