The In-Operation Drift Compensation of MEMS Gyroscope Based on Bagging-ELM and Improved CEEMDAN
Haoyu Gu, Xianxue Liu, Baolin Zhao, Hao Zhou · IEEE Sensors Journal · 2019
Limited by the micro-electromechanical system (MEMS) fabrication technology, the in-operation drift of MEMS gyroscope which degrades measurement repeatability, accuracy, and stability has non-stationary wide-band components and a large difference between each power cycle. The drift limits the usability of MEMS gyroscope in a variety of field applications where autonomous and repeatable operation is required over a long time in harsh environmental conditions. Therefore, a novel method is proposed to compensate the drift. At first, an improved complete ensemble empirical mode decomposition is used to decompose the original signals into a series of intrinsic mode functions (IMFs), and the threshold de-noising method is adopted to filter the IMFs; then, the de-noised sub-series are reconstructed into training and testing dataset, respectively, and the bagging extreme learning machine-based model has been trained; finally, the compensation signal is predicted by the model with testing dataset, and the desired results can be obtained after compensation. The proposed method has been validated by a 9000-s in-operation experiment of CRG20 by comparing it with a typical method. The experiment demonstrated that the proposed method can enhance generalization performance and can boost compensation accuracy of the model, and the bias instability reduced from 0.0785°/s to 0.0046°/s.