Intelligent Solenoid Pump Fault Detection based on MFCC Features, LLE and SVM*

Ugochukwu Ejike Akpudo, Jang-Wook Hur

Abstract: The need for condition monitoring of fluid pumps cannot be overemphasized and this has led to the uproar in research studies on effective and efficient techniques for optimized condition monitoring of VSC63A5 solenoid pumps. From vibrational signals, useful feature extraction can be conducted for accurate and reliable condition monitoring/fault diagnosis of these pumps. By employing the locally linear embedding (LLE) as an effective dimensionality reduction technique, we carried out feature fusion of extracted Mel frequency cepstral coefficients (MFCCs) and employed the Gaussian kernel support vector machine (SVM*) for fault detection. Results show that the proposed diagnostics model is effective and reliable for diagnostics of solenoid pumps.