Abstract: A kind of fitting model and piecewise fitting method based on six kind of noise is proposed to solve the problem that the fitting model will probably be inaccurate by the traditional analysis method for the noise characteristics of fiber optic gyroscope. The reason of leading to the inaccuracy of fitting model is analyzed according to the data of a fiber gyroscope. It is found that a low frequency sinusoidal noise exists in the random noise of fiber-optic gyroscope. Compared with other gyroscope noises, this noise is small so that it was ignored in the traditional fitting model for five noise items, which may lead to the irrationality of fitting model and the inaccuracy of fit coefficients for optical fiber gyroscope,especially high precision optical fiber gyroscope. A fitting model for six noise items, including sine noise, is established by analyzing the time and frequency domains of measured gyroscope signal to determine the frequency of sinusoidal signal. A subsection fitting method is proposed to reduce the fitting error coupled by the approximating correlation time of sinusoidal noise, quantization noise, zero offset instability and rate ramp. The measured data of fiber optic gyroscope is analyzed with Allan variance based on the fitting model for six noise items. The results show that the fit error of the fitting model is small, and the fit error coefficient is accurate and credible.

Key words: ship engineering, Allan variance, fiber optic gyro, fitting, sinusoidal noise