Error correction of optical path component manufacture for Fiber Optic Gyroscope using SVM and Monte Carlo

As a kind of complex aerospace electronic instrument, the product quality of Fiber Optic Gyroscope (FOG) is always unstable even we manufacture and assemble them abiding by the process regulations strictly. This paper presents an error correction method for the optical path component of FOG by using the bigdata analysis techniques. First, we analyze the manufacture procedure of optical path component and collect 17 core parameters of it. These data are considered to have some crucial influences on the final quality level of FOG. Second we accumulate more than 200 dataset and use them to train a SVM so that the SVM can be used to predict the quality level of new optical path component. Third, once the output of SVM shows the quality level of new product is low, we will use the Monte Carlo (MC) method to generate the error correction values of these core parameters above stochastically according to their historical distributions. After that the corrected core data will be sent to SVM for quality evaluation again. This process will be repeated until the output of SVM becomes positive. The validity and correctness of proposed technique are presented and the corresponding application hints are also discussed in the end of this paper.