Search

Given the enormous capital value of power transformers and their integral role in the electricity network, increasing attention has been given to diagnostic and monitoring tools as a safety precaution measure to evaluate the internal condition of transformers. This study overcomes the fault diagnosis problem of power transformers using an ultra high frequency drain valve sensor. A convolutional neural network (CNN) is proposed to classify six types of discharge defects in power transformers. The proposed model utilizes the phase-amplitude response from a phase-resolved partial discharge (PRPD) signal to reduce the input size. The performance of the proposed method is verified through PRPD experiments using artificial cells. The experimental results indicate that the classification performance of the proposed method is significantly better than those of conventional algorithms, such as linear and nonlinear support vector machines and feedforward neural networks, at 18.78%, 10.95%, and 8.76%, respectively. In addition, a comparison with the different representations of the data leads to the observation that the proposed CNN using a PA response provides a higher accuracy than that using sequence data at 1.46%.

A time-domain transmission measurement technique for extracting the complex propagation constant of a cross-linked polyethylene cable is proposed. For this extraction, a pulse transmitted through the cable is measured in the time domain and analyzed in the frequency domain. The proposed technique does not use an open termination for reflection, and therefore increases the upper frequency of the bandwidth for extraction; by reducing the pulse width, it decreases the measurement error as compared with an existing time-domain reflectometry method.

The classical multiple signal classification (MUSIC) method has been widely used in induction machine fault detection and diagnosis. This method can extract meaningful frequencies but cannot give accurate amplitude information of fault harmonics. In this paper, we propose a new frequency analysis of stator current to estimate fault-sensitive frequencies and their amplitudes for broken rotor bars (BRBs). The proposed method employs a frequency estimator, an amplitude estimator, and a fault decision module. The frequency estimator is implemented by a zoom technique and a high-resolution analysis technique known as the estimation of signal parameters via rotational invariance techniques, which can extract frequencies accurately. For the amplitude estimator, a least squares estimator is derived to obtain amplitudes of fault harmonics, without frequency leakage. In the fault decision module, the fault diagnosis index from the amplitude estimator is used depending on the load conditions of the induction motors. The fault index and corresponding threshold are optimized by using the false alarm and detection probabilities. Experimental results obtained from induction motors show that the proposed diagnosis algorithm is capable of detecting BRB faults with an accuracy that is superior to the zoom-based MUSIC algorithm.