A fault diagnosis method for a practical engineering application based on CEEMD and ELM

Fault diagnosis of rotating machinery is of great significance in preventing catastrophic accidents and beneficially guaranteeing sufficient maintenance, which attracted a large number of researches to propose different diagnostic methods. However, different from the ideal state in laboratory, in practical engineering applications, the quality of collected vibration signals is vulnerable to environment and conditions and tends to be uneven, which lead to difficulties in accurate fault diagnosis. Concentrating on the signals with characteristics of poor quality, high noise and weak separability in the practical engineering applications, this paper proposes a CEEMD and ELM based fault diagnostic method. The proposed method contains four major steps: Row data acquired, feature extraction based on CEEMD and SVD, fault diagnosis based on ELM and calculation of total accuracy. At last, the proposed method is applied to a practical engineering application, fault diagnosis of a centrifugal pump. Results achieve a better fault recognition rate than the peer methods, which verifies the effectiveness and feasibility of the proposed method.