Allan variance segmented circular fitting method for laser gyroscopes random error analysis

Abstract

The curve fitting process is the key process of Allan variance algorithm which is an effective method for laser gyroscope random error analysis. The traditional piecewise regression method leads to the fitting curve shift up because the cross-impact among the five kinds of random errors in Allan variance hasn't been taken into consideration especially in the case of long correlation time, which will leads to the random error characteristic of Laser Gyroscope can't been evaluated accurately. This paper proposes an Allan variance segmented circular fitting method, which improves the accuracy of curve fitting and especially reduces the fitting error in long correlation time. The proposed method is applied to 7-hour experimental static data analysis of laser gyroscopes, and the random error coefficients estimated with modified Allan variance algorithm can converge with better stability.