Research On Error Analysis, Modeling And Filtering Of Fog

The Fiber Optic Gyroscope (FOG) is a kind of angle velocity sensor based on Sagnac effect. As a new of type of inertial measurement equipment, it is of great importance in theory and practical application to research and develop FOG. The paper researches on some key problems including the random error analysis, modeling and simulation, and FOG signal filtering, aiming at finding new solutions for these problems by applying the new research methods and by introducing new theoretical analysis techniques.To solve the problem that the estimation error of Allan variance is great at long-termτ-values, according to the frequency properties of FOG random errors, a new method for characterizing the random errors based on Total variance is presented. The Total variance characteristics of the random errors are analyzed, and the estimation performance is compared between Allan variance and Total variance. The experimental results show that, the values of Total variance are consistent with the properties of the actual power-law noise at long-termτ-values. It can efficiently improve the confidence in the case of great mean time index with the Total variance method, and its estimation performance outperforms that of Allan variance.Furtherly, the paper proposes a novel method for analyzing the random error properties of the FOG based on a new special purpose statistic which is called Theoretical variance #1. Theoretical variance #1 has two significant improvements over Allan variance in estimating long-termτ-values: (1) it can evaluate variance at averaging times 50% longer than that of Allan variance, and (2) it can estimate variance with greater confidence than any other estimator. With these merits, it can identify error source type and estimate the error coefficients more accurately by using Theoretical variance #1 method.It is important to build a complete system model and conduct the simulation in the research and application of FOG. The dynamic model and stochastic model of digital closed-loop FOG are developed in the paper. The nonlinear dynamic model is simplified into a linear discrete model with a number of reasonable approximations. The digital control algorithm in the closed-loop control is designed, and the dynamic simulation is conducted. In the stochastic modeling and simulation, a method for synthesis of 1/f noise with Shannon orthogonal wavelet transform and inverse whitening filtering is presented. Various stochastic processes are investigated to simulate various types of random noise exiting in FOG. The bias instability is simulated with 1/f noise. Angular random walk noise is simulated with one-order discrete integration of white noise, and the exponentially correlated noise is simulated with one-order Markov Process or exponentially correlated stochastic sequence. The stochastic simulations are conducted, and the effectiveness of the stochastic modeling and simulation is validated with Theoretical variance #1 analysis. The paper provides a way to simulate the random drift of FOG, and gives a reference for error analysis, performance evaluation and filter design, etc.The output signal of FOG is weakly correlated and nonstationary, and it usually contains the fractional noise. It can't efficiently eliminate the random noise by using the traditional filtering methods. The denoising method based on wavelet analysis is introduced. A wavelet-based equivalent FIR filtering algorithm is deduced on the basis of Mallat pyramid algorithm, and a real-time wavelet-based filtering method for denoising the output signal of FOG is proposed.To eliminate the random noise of FOG and to improve the performance of FOG, an implementation scheme of digital close-loop FOG based on FPGA and DSP is presented. To restrain the impact effect of the square wave on other analog circuits, the time-damain comb filter based on analog switch technique is added. According to the property of wavelet transform, a multirate sampling conversion scheme with real-time wavelet filtering is proposed. The design of multi-stage decimation filters is given and its implementation scheme in FPGA and DSP is presented.