No abstract

<p>This paper presents two approaches to power circuit condition assessment. The first one covers a wide variety of machine learning classification algorithms where the input for the classification is a manually selected feature set. The second one utilizes deep learning classification based on the convolutional neural network. Both approaches revolve around the idea behind spectral kurtosis, one of which exploits its visual representation in the form of kurtogram. The first approach uses a spectral kurtosis curve as the base for feature extraction while the second approach uses a spectral kurtosis kurtogram as a single input into the convolutional neural network. The validation is performed on a large set of vibration signatures and compared to competing state-of-the-art algorithms. The results indicate promising features of the proposed approach. <br>  </p>

<p>This paper presents two approaches to power circuit condition assessment. The first one covers a wide variety of machine learning classification algorithms where the input for the classification is a manually selected feature set. The second one utilizes deep learning classification based on the convolutional neural network. Both approaches revolve around the idea behind spectral kurtosis, one of which exploits its visual representation in the form of kurtogram. The first approach uses a spectral kurtosis curve as the base for feature extraction while the second approach uses a spectral kurtosis kurtogram as a single input into the convolutional neural network. The validation is performed on a large set of vibration signatures and compared to competing state-of-the-art algorithms. The results indicate promising features of the proposed approach. <br>  </p>

This paper presents two approaches to power circuit condition assessment. The first one covers a wide variety of machine learning classification algorithms where the input for the classification is a manually selected feature set. The second one utilizes deep learning classification based on the convolutional neural network. Both approaches revolve around the idea behind spectral kurtosis, one of which exploits its visual representation in the form of kurtogram. The first approach uses a spectral kurtosis curve as the base for feature extraction while the second approach uses a spectral kurtosis kurtogram as a single input into the convolutional neural network. The validation is performed on a large set of vibration signatures and compared to competing state-of-the-art algorithms. The results indicate promising features of the proposed approach.