The Digital Closed-loop Control Techniques And Random Error Analysis And Restraint Of FOG

According to different achieve method, The FOG, can be divided into open-loop andclosed-loop FOG. In entire dynamic range, the closed-loop FOG has better accuracy、higherscale factor linearity, and better practical effect than the open-loop FOG. The realizationmethod of the closed-loop FOG also can be divided into analog and digital closed-loopimplementation, the analog closed-loop FOG can not achieve very high scale factor stabilityand linearity, but the digital closed-loop implementation can solve the shortcomings of analogclosed-loop implementation. Under the present experimental conditions of the laboratory, thispaper researched the self-study digital closed-loop FOG, then detailedly analysed theprinciples and implementation of closed-loop FOG, and analysed the errors in the randomdrifting data of FOG, meanwhile adopted the filter method to restraint the random errors.Random drift is one of the most important factors affecting the accuracy of FOG. Oncethis drift can not be restrained effectively, it will reduce the measurement accuracy of the FOG.even if random drift error of the measurement be sented to the navigation computer for latercalculation processing, the error will be not converged over time which seriously affect theaccuracy of the entire system. Therefore, the random errors of FOG are needed to berestrained for improving the accuracy of the system.In the light path, optical non-reciprocity error is the main error source of FOGmeasurement. The thermal noise is analysed in the lighr source, the temperature controlcircuit is designed and achieved to control the temperature of the light source. Using Allanvariance analysis method, the various error sources of FOG random drift can be determinedidentificated and characterizated, then the noise figures is calculateed.The digital filtering and Kalman filtering method is introduced to analyse to restraint therandom error of FOG. Before the measurement data send out, random noise in the data can beeffectively restraint by digital filtering, which is not reduced real-time propertity; sample thezero drift data of FOG and establish timing module, then identificate the model coefficients,and at last compensate the random drift of the data by Kalman filtering method. Finally, thetwo filtering methods are compared and the advantages and disadvantages of both approachesdescribed.