Incipient fault diagnosis for the cam-driven absolute gravimeter

Area of Science:

• Geophysics and Measurement Science
• Signal Processing and Machine Learning

Background:

• Incipient faults in cam-driven absolute gravimeters cause vibration disturbances, reducing measurement accuracy.
• Detecting these faults is challenging due to their small amplitude, poor feature representation, and presence in noise.

Purpose of the Study:

• To develop a novel and effective method for the incipient fault diagnosis of cam-driven absolute gravimeters.
• To enhance the measurement accuracy and reliability of absolute gravimeters by addressing fault detection challenges.

Main Methods:

• Integrated parameter-optimized Variational Mode Decomposition (VMD) with Light Gradient Boosting Machine (LightGBM).
• Utilized Sparrow Search Algorithm for VMD parameter optimization.
• Employed Pearson correlation coefficient for effective intrinsic mode function (IMF) selection.
• Applied adaptive noise reduction and low-frequency IMF reconstruction.
• Extracted multi-scale permutation entropy as fault feature vectors.
• Inputted multi-dimensional feature vectors into LightGBM for classification.

Main Results:

• The proposed method achieved a high identification accuracy of 98.41% for incipient faults.
• Demonstrated effective detection of various incipient failures in cam-driven absolute gravimeters.
• Successfully addressed the issue of low measurement accuracy caused by incipient faults.

Conclusions:

• The integrated VMD-LightGBM method provides an accurate and efficient solution for incipient fault diagnosis in cam-driven absolute gravimeters.
• This approach enables rapid tracing and accurate positioning of gravimeter faults, offering significant engineering application potential.