On stationary and nonstationary models of long term random errors of gyroscopes and accelerometers in an inertial navigation system

The drifts caused by gyroscopes and accelerometers are major sources of error in an inertial navigation system which is designed to operate over long periods of time. Usually, part of these drifts are deterministic and predictable and hence are calibrated and compensated for in a system. However, the random noise drifts that can not be compensated for are to be taken into account while performing an error analysis. In addition, a Kalman filter can be designed to compensate for these random errors if proper models are obtained from the actual test data. This paper presents the model development and evaluation techniques using time series analysis. The usefulness and validity of the models developed is demonstrated by simulation and covariance analysis results.