Microgyroscope Temperature Effects and Compensation-Control Methods

Dunzhu Xia, Shuling Chen, Shourong Wang, Hongsheng Li2009 · DOI

TL;DR

This paper tackles the significant temperature-induced performance degradation in microgyroscopes by proposing a combined temperature compensation and control strategy. They demonstrate that a piecewise polynomial fitting approach effectively compensates for temperature-induced zero bias drift, and an integral-separated PID controller maintains the gyroscope's internal temperature at an optimal operating point, leading to improved precision and stability.

What Problem Does This Paper Address?

Microgyroscopes, despite their advantages like small size and low cost, suffer from performance degradation due to ambient temperature variations. These fluctuations affect critical parameters like resonant frequency, quality factor, and most importantly, the zero bias, leading to reduced precision and accuracy in angular velocity measurements. Existing methods often struggle to provide comprehensive solutions, necessitating research into effective temperature compensation and control strategies to ensure reliable operation across a wide temperature range.

Methods

The authors propose a two-pronged approach:

  • Temperature Compensation:
    • BP Neural Network Modeling: Initially explored for building a temperature model, this method leverages its nonlinear fitting capabilities.
  • Piecewise Polynomial Fitting: This method was ultimately chosen for the actual system due to its simplicity and effectiveness. It involves fitting polynomial functions to the gyroscope's zero bias data across different temperature ranges. The derived polynomials are then used to calculate and subtract temperature-induced bias from the gyroscope's output in real-time.
  • Temperature Control:
    • Integral-Separated PID Control: This algorithm is employed to maintain the internal temperature of the microgyroscope at an optimal operating point (around 55°C). It uses a threshold to switch between PD control (for faster response and to avoid overshoot) and PID control (for higher precision) based on the temperature error. A thermoelectric cooler (TEC) is used to actively adjust the temperature.

Datasets

The experiments utilized a miniaturized microgyroscope prototype. The "datasets" were generated through experimental measurements of the gyroscope's zero bias and resonant frequency/quality factor across a range of controlled temperatures (-40°C to 80°C). These experimental results were used to build and validate both the compensation models and the control system. No pre-existing public datasets were used.

Key Insights and Ideas

  • Piecewise polynomial fitting is a practical and effective solution: While BP networks offer theoretical advantages, the simplicity and real-time implementability of piecewise polynomial fitting make it a superior choice for practical microgyroscope compensation.
  • Optimal operating temperature exists: The research identified an optimal working temperature of approximately 55°C for the microgyroscope, where its performance is maximized.
  • Combined compensation and control is crucial: Temperature compensation alone cannot fully mitigate all temperature effects, especially during start-up. Combining it with active temperature control is essential for achieving robust performance across the entire operating temperature range.
  • Temperature control can significantly reduce stable-state error: The PID-based temperature control system demonstrated the ability to stabilize the internal temperature within a narrow range (±0.3°C) under varying ambient conditions.

Key Results

  • Zero Bias Compensation: The temperature compensation system, using piecewise polynomial fitting, reduced the maximum absolute zero bias from 12.33°/s to 0.608°/s, a significant improvement. This performance was comparable to the BP neural network approach.
  • Temperature Control Effectiveness: The temperature control system successfully maintained the internal gyroscope temperature around 55°C, even when the ambient temperature varied between -20°C and +35°C, with a maximum stable-state error of less than 0.3°C.
  • Resonant Frequency and Quality Factor: Experimental results confirmed the theoretical analysis that resonant frequency decreases linearly with increasing temperature, while the quality factor is significantly affected by low temperatures, becoming more stable at higher temperatures.

Limitations and Future Work

  • Start-up Time: The paper acknowledges that the start-up procedure for reaching thermal equilibrium and stable zero bias takes time, even with the control system.
  • Model Complexity: While polynomial fitting is simpler than BP networks, building highly accurate models might require higher-order polynomials, increasing computational load.
  • Universality: The paper suggests that manufacturing errors and peripheral circuit variations can lead to different temperature models for different microgyroscopes, implying that a universally applicable model might be challenging.
  • Future Work: The authors advocate for the combined compensation-control approach to achieve optimal performance across the full temperature range, suggesting further refinement of this integrated system. They also imply that further research into more universally applicable methods for reducing temperature effects would be beneficial.

Summary updated

BibTeX
@misc{xia2009microgyroscope,
  author = {Dunzhu Xia and Shuling Chen and Shourong Wang and Hongsheng Li},
  title = {{Microgyroscope Temperature Effects and Compensation-Control Methods}},
  year = {2009},
  doi = {10.3390/s91008349},
}
Figure