MEMS Gyroscope Multi-Feature Calibration Using Machine Learning Technique

Yaoyao Long,Zhenming Liu,Cong Hao,Farrokh Ayazi

2024-10-10

Abstract:Gyroscopes are crucial for accurate angular velocity measurements in navigation, stabilization, and control systems. MEMS gyroscopes offer advantages like compact size and low cost but suffer from errors and inaccuracies that are complex and time varying. This study leverages machine learning (ML) and uses multiple signals of the MEMS resonator gyroscope to improve its calibration. XGBoost, known for its high predictive accuracy and ability to handle complex, non-linear relationships, and MLP, recognized for its capability to model intricate patterns through multiple layers and hidden dimensions, are employed to enhance the calibration process. Our findings show that both XGBoost and MLP models significantly reduce noise and enhance accuracy and stability, outperforming the traditional calibration techniques. Despite higher computational costs, DL models are ideal for high-stakes applications, while ML models are efficient for consumer electronics and environmental monitoring. Both ML and DL models demonstrate the potential of advanced calibration techniques in enhancing MEMS gyroscope performance and calibration efficiency.

Machine Learning,Signal Processing

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the accuracy problem of MEMS gyroscopes when applied in navigation, stabilization and control systems. Specifically, although MEMS gyroscopes have the advantages of small size and low cost, they are affected by errors and inaccuracies in actual use, and these errors and inaccuracies are complex and change over time. Traditional linear calibration methods cannot capture the complex nonlinear relationship between the output signal and the rotation rate, resulting in a decrease in accuracy. In addition, traditional calibration methods require a separate control - rate test for each device to extract the scale factor, and this process is both time - consuming and expensive.

To solve these problems, the paper proposes to use machine learning (ML) and deep learning (DL) technologies to improve the calibration process of MEMS resonant gyroscopes by analyzing multiple signals. Specific methods include using XGBoost and multi - layer perceptron (MLP) models, which are selected respectively for their high prediction accuracy and the ability to handle complex nonlinear relationships. The study found that both models can significantly reduce noise, improve accuracy and stability, and are superior to traditional calibration techniques. Although deep - learning models have a high computational cost, they are suitable for high - risk applications, while machine - learning models are suitable for fields such as consumer electronics and environmental monitoring. Overall, these advanced calibration techniques show the potential to improve the performance and calibration efficiency of MEMS gyroscopes.