Article: Reciprocating compressor start-up fault monitoring based on sensor and limit learning machine Journal: International Journal of Information and Communication Technology (IJICT) 2025 Vol.26 No.4 pp.72 - 88 Abstract: Condition monitoring of reciprocating compressors (RC) can improve reliability of equipment operation. To address the problem of unsatisfactory monitoring accuracy of the existing RC start-up monitoring methods, the common types of failures were first analysed to obtain the external influencing factors. Second, the overall architecture of RC starter fault monitoring was designed based on the shock pulse sensor to obtain the intrinsic signal data of the RC starter. The internal and external influence data were then pre-processed, the main variables were extracted using principal element analysis, and the variables were decomposed into eigenmode components using an improved empirical modal decomposition method. Finally, the extreme learning machine (ELM) algorithm (OLEM) is optimised by the regularisation term, and the RC start-up fault state is predicted using OLEM. The experimental outcome indicates that the proposed method has a monitoring accuracy of 92.5% and has a strong monitoring capability. Inderscience Publishers - linking academia, business and industry through research

Title: Reciprocating compressor start-up fault monitoring based on sensor and limit learning machine

Authors: Cheng Pan; Lei Guo; Weiwei Zeng; Shihao Hu; Xin Li

Addresses: Petro China Tarim Oilfield Company, Korla 841000, China ' Petro China Tarim Oilfield Company, Korla 841000, China ' Petro China Tarim Oilfield Company, Korla 841000, China ' Petro China Tarim Oilfield Company, Korla 841000, China ' Xinjiang Compressor Branch, CNPC Jichai Power Company Limited, Korla 841000, China

Abstract: Condition monitoring of reciprocating compressors (RC) can improve reliability of equipment operation. To address the problem of unsatisfactory monitoring accuracy of the existing RC start-up monitoring methods, the common types of failures were first analysed to obtain the external influencing factors. Second, the overall architecture of RC starter fault monitoring was designed based on the shock pulse sensor to obtain the intrinsic signal data of the RC starter. The internal and external influence data were then pre-processed, the main variables were extracted using principal element analysis, and the variables were decomposed into eigenmode components using an improved empirical modal decomposition method. Finally, the extreme learning machine (ELM) algorithm (OLEM) is optimised by the regularisation term, and the RC start-up fault state is predicted using OLEM. The experimental outcome indicates that the proposed method has a monitoring accuracy of 92.5% and has a strong monitoring capability.

Keywords: reciprocating compressor; RC; fault monitoring; principal element analysis; empirical modal decomposition; EMD; extreme learning machine; ELM.

DOI: 10.1504/IJICT.2025.144659

International Journal of Information and Communication Technology, 2025 Vol.26 No.4, pp.72 - 88

Received: 08 Dec 2024
Accepted: 18 Dec 2024
Published online: 25 Feb 2025 *

Title: Reciprocating compressor start-up fault monitoring based on sensor and limit learning machine

Keep up-to-date