Analysis of Inertial Navigation System errors from van testing using an optimal Kalman filter/smoother

The optimal Kalman filter/smoother is an effective tool to systematically assess the error sources of an Inertial Navigation System (INS) including instrument (gyro and accelerometer) errors and initial conditions. In order to best utilize the entire set of the available test data in the estimation of INS error, a fixed-interval smoother was chosen. The fixed-interval smoothed estimate is the optimal combination of two Kalman filters: forward (conventional) filter and backward filter. Together these two filters utilize all available measurements and yield the optimal smoothed estimate. The data from two van tests was collected in the free inertial mode without updates from the Global Positioning System (CPS). The CPS data and INS data are recorded in two separate files and both data streams are time tagged to Greenwich Time (GMT). The CPS latitude and longitude data obtained from CPS was used as reference. The INS position errors are the difference between the INS and the CPS positions. The measurements of INS position errors are then analyzed to obtain estimates of the error sources discussed above using the Kalman filter/smoother. This paper describes the INS van-test data, the optimal Kalman filter/smoother utilized for the error analysis, the INS error model, and the results of the analysis. Analysis results are verified through measurement reconstruction and the agreement between the actual and reconstructed measurements is excellent. The analysis confirms the existence of a very large scale factor error in one accelerometer.