A New Fusion Fault Diagnosis Method for Fiber Optic Gyroscopes.

Area of Science:

• Engineering
• Instrumentation
• Signal Processing

Background:

• Fiber optic gyroscopes (FOGs) are critical for high-precision inertial navigation.
• Vibration interference significantly distorts fault features in operational FOGs, compromising reliability.
• Effective fault diagnosis under vibration is essential for FOG performance assurance.

Purpose of the Study:

• To develop a robust fault diagnosis method for FOGs that minimizes vibration interference.
• To improve the diagnostic accuracy and anti-vibration capability of FOG fault detection systems.

Main Methods:

• Feature extraction using Fast Fourier Transform (FFT) and wavelet packet decomposition (WPD).
• Development of diagnostic classifiers employing sparse autoencoder (SAE) and neural network (NN).
• Establishment of a fusion neural network model integrating outputs from primary classifiers.

Main Results:

• The proposed fusion method demonstrated high diagnostic accuracy for FOGs under random vibration.
• Experimental validation confirmed the method's robustness and superior anti-vibration capability compared to other techniques.
• Successful identification of five distinct FOG fault types under vibration conditions.

Conclusions:

• The fusion fault diagnosis method effectively addresses vibration-induced distortions in FOGs.
• The approach significantly enhances the reliability and accuracy of FOG fault diagnosis in challenging operational environments.
• This method offers a robust solution for maintaining FOG performance under vibration stresses.