Based on the conception of optimum estimation and AIC criterion, the time-varying coefficient autoregressive (AR) model is established for the random drift rate of the electrostatic suspended gyro (ESG) by recursive algorithm. The time-varying coefficient with its canonical standard error and mode's residual variance are then discussed. According to the variant of the coefficient and its canonical standard error, we distinguish whether or not there is singular signal (or failure).
展开▼