Drift Error Compensation for a High-Precision 2-D Angle Sensor Based on EEMD and Multiple Lag Regression

Abstract

Drift issues are commonly encountered in precision instruments operating in standard measurement environments due to temperature fluctuations, material property variations, and environmental disturbances, which represent a significant bottleneck to achieving high measurement accuracy. To address this limitation, a drift error compensation method based on ensemble empirical mode decomposition (EEMD) and multivariate lag regression has been proposed in this article. The structure, principles, and working environment of a high-precision 2-D angle sensor have been analyzed, and the primary influencing factors and mechanisms contributing to drift errors have been systematically investigated. Drift and error source signals have been decomposed and denoised using EEMD, and effective intrinsic mode function (IMF) components have been extracted. The lag characteristics of these components have been analyzed and incorporated into a multivariate regression model for drift error compensation. Partial regression analysis and significance testing have been employed to optimize the model, reducing complexity and enhancing generalization. Experimental results show that drift errors in the yaw and pitch directions have been reduced by more than 60.01% and 67.51%, respectively, after compensation. When compared with classical multivariate regression, LSTM, and support vector machine (SVM) methods, the proposed approach demonstrates certain advantages in error compensation performance, robustness, and complexity. In addition, the method is also suitable for broader applications to other high-precision measurement instruments.