Yazar "Gurusamy, V." seçeneğine göre listele

Listeleniyor 1 - 2 / 2
  • Küçük Resim Yok
    Öğe
    A Comparative Study of Current, Vibration and Stray Magnetic Flux Based Detection for Parallel Misalignment Fault in Induction Motors
    (Ieee, 2021) Goktas, Taner; Arkan, Muslum; Gurusamy, V.
    The misalignment fault is commonly caused by incorrect shaft positions between motor and load in electrical machines. It affects the mechanical symmetry of machine and thus causes mechanical oscillation on the shaft. In this paper, the parallel misalignment fault is analyzed based on stator current, vibration and stray flux in induction motors (IMs). The three-axis vibration sensor and an integrated flux sensor are used in order to stream vibration and stray flux for diagnostics process, respectively. The comparative results between stator current, vibration and stray flux are presented. Experimental results show that stator current and vibration-based analyses provide highly reliable results than stray flux for parallel misalignment fault. It is also shown that the proposed signatures in current and vibration vary very little with respect to load and motor drive type. Moreover, Multilayer Perceptron (MLP) based machine learning algorithm using vibration and stator current is carried out and it has excellent performance in the automatic detection of parallel misalignment fault.
  • Küçük Resim Yok
    Öğe
    The Performance Evaluation of Machine Learning based Techniques via Stator Current and Stray Flux for Broken Bar Fault in Induction Motors
    (Ieee, 2021) Younas, M. B.; Ullah, N.; Goktas, Taner; Arkan, Muslum; Gurusamy, V.
    In this paper, the machine learning based techniques are evaluated using stator current and stray flux for broken bar fault in induction motors (IMs). The feature extraction is achieved from Discrete Wavelet Transform (DWT) for both healthy and faulty operations. In order to analyze the performance of different classifier, six fundamental classifications with 23 sub-classifiers are used via a toolbox. It has been observed that 18 out of 23 classifiers have shown great performance (100% accuracy) and two more classifier results at accuracy of greater than 90% for stray flux. Both simulation and experimental results show that stray flux provides better diagnostics results than stator current using different machine learning based classification algorithms in IMs.