The Error Analysis Of IFOG Signal Based On Wavelet De-noising

Inertial navigation system(INS) has been widely applied in national defense, aviation,aerospace and other domain. With the development of inertial technology, there is anincreasing trend of INS that traditional mechanical gyros are replaced by new types ofsolid-state gyros and platform inertial navigation system is replaced by strap-down inertialnavigation system. Along with the development of photoelectric technology, optics sensorshave become a new generation of inertial measurement. Depending on the advantages such assmall in size, light in weight, easy to be vindicated and better cost performance, fiber opticgyroscope becomes the preferred IMU in low or medium accuracy INS.In this paper, the self-developed I-FOG (interferometric fiber optic gyroscope) by thelaboratory is taken as the background, and the principle of FOG is the basis. The opticaldevices in the optic circuit of I-FOG are analyzed, the random error coefficient is figured outby Allan variance, and the output signal noise of the FOG is filtered by the noise eliminationprinciple of wavelet analysis.Starting from the Sagnac Effect of the FOG, the principle of FOG is discussed, thereciprocity of optic path and the principle of square wave modulation and demodulation,which is a basic modulation and demodulation method, are described. It adopt step pulsemodulation plus4-status square as offset wave which aiming at the migration of Y waveguide.Each optic device in FOG is a noise source, it brings in various sorts of non-reciprocityeffects, which will cause output migration of FOG and unstable of scale factor. Therefore, thelight source, coupler, Y waveguide and fiber loop are studied, and the main reason that thesedevices cause error is analyzed. Corresponding method is proposed to increase the precisionof the FOG.The characteristics of random noise in output signal of FOG are analyzed, and themathematic statistical characteristics are obtained. The performance of the existing FOG isevaluated by Allan variance method, the influence from all noise sources on the FOG isobtained. So the factors that reduce the performance of the FOG can be figured out.The FOG signal is a sort of stationary random signal with high random noise, theconventional filtering method cannot get a predetermined result. Therefore, the real-time liftwavelet de-noising is applied, and based on characteristic analysis of gyroscope signal wedesign the wavelet base, decomposition scale, threshold and threshold function of filteringalgorithm. Finally the signal is compared to the signal that before filtering, we can verify the effect of wavelet de-noising.