Traditional feature extraction and selection is a labor-intensive process requiring expert knowledge of the relevant features pertinent to the system. This knowledge is sometimes a luxury and could introduce added uncertainty and bias to the results. To address this problem a deep learning enabled featureless methodology is proposed to automatically learn the features of the data. Timefrequency representations of the raw data are used to generate image representations of the raw signal, which are then fed into a deep convolutional neural network (CNN) architecture for classification and fault diagnosis. This methodology was applied to two public data sets of rolling element bearing vibration signals. Three time-frequency analysis methods (short-time Fourier transform, wavelet transform, and Hilbert-Huang transform) were explored for their representation effectiveness. The proposed CNN architecture achieves better results with less learnable parameters than similar architectures used for fault detection, including cases with experimental noise.

No abstract

No abstract

This paper aims at characterizing and improving the metrological performances of Current (CT) and Voltage Instrument Transformers (VT) in harmonic measurements in power system. A theoretical analysis is carried out to demonstrate that, due to the iron core non linearity, CT and VT output signal is distorted even when the input signal is a pure sinusoidal. Starting from this analysis, a new method for CT and VT characterization and compensation is proposed. In a first step, they are characterized in sinusoidal conditions and the harmonic phasors of the distorted output are measured; in the second step, these phasors are used to compensate the harmonic phasors measured in normal operating conditions, which are typically distorted. The proposed characterization and compensation techniques are called SINDICOMP (SINusoidal characterization for DIstortion COMPensation). Several experimental tests, using high accuracy calibration setups, have been performed to verify the proposed methods. The experimental results showed that the SINDICOMP technique assures a significant improvement of CT and VT metrological performances in harmonic measurements.

No abstract