Wuhan Ligong Daxue Xuebao (Jiaotong Kexue Yu Gongcheng Ban)/Journal of Wuhan University of Technology (Transportation Science and Engineering)

About the Journal

[This article belongs to Volume - 46, Issue - 5]

Abstract :

Due to heavy work load of marine diesel engines, the failure in their mechanical components may result in serious accidents. Existing condition monitoring methods for marine diesel engines usually adopt warning after the failure occurred. In order to predict potential faults, this work has put forward a remote intelligent monitoring system for marine diesel engines. The global system for mobile communication mode was employed to construct the basis of data remote transmission, and a new multi-kernel extreme learning machine algorithm was proposed to diagnose the early faults in an intelligent method. Experimental tests were carried out in the marine diesel engine fault diagnosis set-up. The analysis results show that the proposed remote intelligent monitoring system can accurately, timely and reliably detect the potential failures. Meanwhile, the proposed multi-kernel extreme learning machine was compared with the existing methods. The comparison indicates that the multi-kernel extreme learning machine outperforms its rivals in term of fault detection rate by an increase of 3.4%. Therefore, the proposed remote intelligent monitoring system has good prospects for engineering applications.