Fast Detection of Centrifugal Pumps Condition by Structural Analysis of MEMS Sensor Signals

• Published in: Journal of control, automation & electrical systems Vol. 33; no. 1; pp. 293 - 303
• Main Authors: de Araújo, Rodrigo D. B., Rocha, João M. M., Santos, Matheus A. dos, Ramalho, Geraldo L. B.
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.02.2022; Springer Nature B.V
• Subjects: Accelerometers; Axial flow pumps; Centrifugal pumps; Control; Control and Systems Theory; Electrical Engineering; Embedded systems; Engineering; Feature extraction; Gyroscopes; Mechatronics; Microelectromechanical systems; Predictive maintenance; Robotics; Robotics and Automation; Signal classification; Structural analysis; Vibration analysis; Condition monitoring; Centrifugal pump; MEMS sensors; Predictive maintenance; Structural analysis
• ISSN: 2195-3880; 2195-3899
• DOI: 10.1007/s40313-021-00806-w

Predictive maintenance techniques allow monitoring machines without the need to interrupt its operation, reducing the frequency of corrective maintenance interventions. This work proposes a noninvasive vibration detection system to analyze the state of a three-phase electric powered centrifugal pumps using a micro-electro-mechanical system sensor composed of three-axial accelerometer and gyroscope. The structural analysis is used to extracting time domain signal characteristics. Classification experiments were performed in order to analyze the signal discrimination power for different machine conditions. The results show that using the signals from accelerometers and gyroscopes improves the performance of the classifiers, achieving more than 98% correct condition classification and 100% pump classification. In addition, using the SCM structural analysis method as feature extraction is fast and produces a relative small feature dimension enabling it to be used in embedded systems.