Grating-Based Fiber-Optic Sensing Using a Single Packaged FBG for Boundary-Dependent Motor Vibration-State...

Area of Science:

• Optical Engineering
• Mechanical Engineering
• Signal Processing

Background:

• Monitoring motor-support systems requires understanding vibration states influenced by boundary conditions.
• Traditional methods often need calibrated speed or acceleration data, limiting application in complex scenarios.

Purpose of the Study:

• To demonstrate single-channel fiber Bragg grating (FBG) sensing for relative vibration-state monitoring.
• To evaluate the system's performance under angle-dependent boundary conditions.
• To assess the efficacy of machine learning models in interpreting optical vibration responses.

Main Methods:

• A fiber Bragg grating (FBG) accelerometer-type sensor was used to record reflected Bragg wavelength fluctuations.
• Adaptive Variational Mode Decomposition (AVMD) was applied for spectral-response tracking.
• LSTM, 1D-CNN, and CNN-LSTM autoencoders were employed to analyze waveform departures from nominal behavior.

Main Results:

• FBG sensing provided distinguishable optical vibration responses under varying boundary states.
• Dynamic angle transitions generated local trajectory changes and potential alarm intervals.
• The CNN-LSTM autoencoder offered a selective response, balancing transition coverage and false alarms.

Conclusions:

• The FBG-based framework enables compact, relative vibration-state transition monitoring.
• This approach is suitable for applications lacking speed references, dense sensor layouts, or labeled fault data.
• The study highlights a novel route for monitoring system dynamics in complex environments.