Fault Detection and Isolation of MEMS IMU Array Based on WOA-MVMD-GLT.

Area of Science:

• Robotics and Control Systems
• Sensor Technology
• Signal Processing

Background:

• Stable output from micro-electro-mechanical system (MEMS) inertial measurement unit (IMU) arrays is crucial for data fusion.
• MEMS IMU arrays are prone to interference and failures due to their large number of components, leading to noise, outliers, and errors.
• Existing methods struggle with low fault detection and isolation accuracy in MEMS IMUs.

Purpose of the Study:

• To propose a novel fault detection and isolation (FDI) method for MEMS IMU arrays.
• To enhance the accuracy and reliability of FDI in MEMS IMU systems.
• To address the challenges of noise, outliers, and interference affecting MEMS IMU output.

Main Methods:

• A new method, WOA-MVMD-GLT, combining multivariate variational mode decomposition (MVMD), whale optimization algorithm (WOA), and generalized likelihood test (GLT) is developed.
• A multi-index fitness function WOA is employed to optimize MVMD parameters.
• MVMD extracts signal features, and GLT constructs fault detection and isolation functions for gyroscopes and accelerometers.

Main Results:

• The proposed WOA-MVMD-GLT method effectively extracts features from MEMS IMU signals.
• The GLT successfully detects and isolates faults in gyroscopes and accelerometers.
• Experimental results demonstrate a significant reduction in false alarm and false isolation rates.

Conclusions:

• The WOA-MVMD-GLT method offers a robust solution for fault detection and isolation in MEMS IMU arrays.
• This approach improves the overall accuracy and reliability of MEMS IMU systems.
• The study highlights the potential of integrating optimization algorithms with advanced signal decomposition and statistical testing for sensor fault diagnosis.