Research on real-time gear fault detection and classification technology based on EFPI vibration sensor

Abstract

In order to detect the health status of gears working for long periods of time in harsh environments such as high speed and high load, this paper studies the real-time detection and classification technology of gear faults based on fiber-optic extrinsic Fabry-Perot interferometer (EFPI) vibration sensors. A deep learning model of 1D-LSAM-CNN-BiLSTM is proposed to improve the accuracy and speed of gear fault classification. Firstly, three different types of gears are prepared, each including a health gear, a wear gear, and a tooth breakage gear. Then, a feedback based single wavelength intensity demodulation EFPI system is used to collect vibration signals for 27 gear fault categories. The collected vibration signals are transmitted to the computer through the ACM8211 gigabit ethernet module and stored. The 1D-LSAM-CNN-BiLSTM deep learning model automatically reads stored data, preprocesses and trains it. The trained model can perform real-time fault classification on the collected vibration signals. The experimental results show that the real-time gear fault detection and classification system proposed in this paper has high recognition accuracy.