Principal Component Analysis Based Vibration Sensor selection for Fault Diagnosis of an Industrial Gearbox

• Published in: 2024 IEEE International Instrumentation and Measurement Technology Conference (I2MTC) pp. 1 - 6
• Main Authors: Goswami, Priyom, Rai, Rajiv Nandan
• Format: Conference Proceeding
• Language: English
• Published: IEEE 20.05.2024
• Subjects: Broken Gear Teeth Fault; Gear Fault; Gears; PCA; Production; Rendering (computer graphics); Support Vector Machine (SVM); Support vector machines; Throughput; Vectors; Vibration Sensor Selection; Vibrations
• ISSN: 2642-2077
• DOI: 10.1109/I2MTC60896.2024.10560836

The failure of gearboxes that form an integral part of any mechanical power transmission system can severely impact the cycle times of various processes leading to a decline in the throughput of production lines. Literature reveals that failure prediction of gear transmission systems is an uphill task due to its complicated geometry and simultaneous faults interaction, rendering fault isolation in gears a challenging task. Multiple sensors are generally deployed at different locations to isolate and analyse faults in gearbox. In this work data is acquired from multiple vibration sensors from three different locations of gearbox and is analysed using Principal Component Analysis(PCA) based approach for selection of best sensor for fault classification. The experiment is conducted on two different cases; one with a good set of gears and another with seeded broken teeth fault. The proposed PCA based approach for sensor selection enables comprehensive data analysis for gear faults classification.