Rebalance Loop Model Analysis And Noise Elimination

As one of the key factors affecting the accuracy of inertial component system,the rebalance loop, working as the controller and measurement unit in the inertialcomponents (accelerometers/gyroscopes) system, could improve both the systematicdynamic-static property and the testing precision. With the development of navigationtechnologies, there is also an increasing demand for the performance of rebalanceloop in the last decade. Because of substantial electromagnetic noise and thermalnoise existing when working, it is significant for improving both the systematicproperty and navigation accuracy to extract useful signal,to separate and to eliminatethe noise in the loop. Devoted in the rebalance loop, this paper aims to eliminate thenoise existing in the rebalance loop. The research was carried out in the design of therebalance loop, the analysis of the loop’s noise, the calculation of header’sperformance parameter, the identification of model parameters and the adaptive noisecancellation based on Least Mean Square(LMS) algorithm.1. Based on the requirement of controlling and measurement, this paper analyzesthe rebalance loop and designs many circuit modules such as signalconditioning unit, controller unit and signal amplification unit. After that, italso analyzes and quantitatively calculates the effect of circuit noise, andmakes the effect on the loop and systematic measurement.2. Focused on the quartz flexible accelerometer, we calculated manycharacteristic parameters of the header, such as the moment of inertia J, thestiffness Ks, to deduce the theoretical value of the performance index. Afterthat, the model’s identification was conducted on the accelerometer system.The SPHS was designed as the stimulate signal, and the least square method(LSM)was used to estimate the parameters. We set up an ARX model to fit theinput and output data of the inertial system and conducted open-loopidentification to obtain the feature parameters of accelerometer. Comparingthem with calculated value, we analyzed the reason of error to offer prioriknowledge and theoretical foundation for subsequent de-noising processing.3. This paper introduces the LMS de-noising algorithm based on the principle ofadaptive filtering and conducts normalization on it. We also proposed a promoted time-varying step LMS algorithm which is especially applicable forthe rebalance loop technology, and wrote a program for it, adjusted theparameters of filter, simulated it and compared it with other filtering methodsuch as RLS and the Kalman filtering method. Taking the specifieddynamically tuned gyroscopes (DTG) accelerometers as an example, wecollected input and output signals and carried out a de-noising experimentwith the above algorithm, which regarded the electromagnetic noise couplingin the loop as background noise, eliminated it and extracted the useful signal.It is concluded that the LMS adaptive de-noising algorithm could filter out thelow frequency noise in the loop and improve the loop’s performanceeffectively.