Development of an advanced nano-satellite (VELOX-IV) : MEMS-based attitude sensing

• Main Author: Muralidharan, Haritha
• Other Authors: Low Kay Soon
• Format: Final Year Project
• Language: English
• Published: 2016
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Control and instrumentation::Control engineering
• Online Access: http://hdl.handle.net/10356/67686
• Institution: Nanyang Technological University

Microelectromechanical Systems (MEMS) – based gyroscopes are widely used in small satellites for attitude estimation. However, MEMS-based gyroscopes are inherently very noisy, and do not provide an accurate measurement of the true angular velocities of the system. Hence, navigational units often implement filtering techniques to reduce the noise perturbing the system. The focus of this project was to implement a low-complexity Kalman filter that will denoise the raw gyroscope readings of ADIS16405 from Analog Devices. The filter algorithm was designed in the state space domain, and then programmed onto MATLAB for tuning. The performance of the filter was evaluated based on several metrics, including the efficacy of noise removal and the ability of the filter to track the true angular velocity of the system without a time delay. The experimental results indicated a strong trade-off between the above two metrics. As a result, a filter that could maximise noise removal at steady state was not able to track the system under dynamic conditions. To address this shortfall, a novel integrated algorithm that combines the Kalman filter with a moving average filter was designed. The moving average filter acted as an error correction measure that reduced the settling time of the Kalman filter in the presence of discontinuities. Subsequently, the integrated algorithm was further tuned for effectual tracking. The final filter design has excellent denoising capabilities and system tracking properties, even under dynamic conditions. Furthermore, the two mechanisms used in the final algorithm are easy to implement, and are not computationally intensive. This maintains the low-in-complexity nature of the filter design, and makes it suitable for implementation in small satellite systems.