Estimation and Statistical Averaging Applied to Redundant Strapped Down Inertial Sensors for Navigation and Flight Control

摘要

The concept of providing inertial data to satisfy on-board avionic functions from an integrated, strapped down, redundant inertial sensor reference system is presently receiving attention for military and commercial aircraft applications. Strapped down inertial sensors are well suited for flight control. However, in the highly dynamic environment of high-performance aircraft, the present accuracy provided by strapped down inertial reference systems (IRS) is insufficient to meet navigation and weapon delivery requirements. Performance improvement of an IRS employing redundant strapped down inertial sensors of specified ensemble is the subject of this study. Two techniques to improve the performance are considered. One technique is to combine all redundant data from non-failed sensors, through a statistical average, into a orthogonal-triad inertial reference frame. This is accomplished by weighted-least-squares averaging. A second technique is the use of an improved gyro and accelerometer output and output-rate estimation scheme. Estimation algorithms are developed using Kalman Filter theory, and evaluated in a highly dynamic environment. Least-squares averaging of the redundant sensor data can significantly improve the navigation system performance. However, sensor misalignment errors must be minimized such that they are not the dominant sensor error source.