Search

In this paper, the amount of SF6 decomposition gases due to the partial discharge (PD) was studied in the SF6 gas-insulated transformer. The long-term PD degradation experiment was performed while controlling the discharge magnitude using the surface discharge, and the gas generation amount was measured by using gas chromatography for SO2F2, SOF2, SO2, CO, and CF4. In addition, to investigate the relationship between the partial discharge energy and the decomposed gas generation amount, partial discharge energy was calculated by a data processing program and converted to the unit of joule per mole. With the finite element method (FEM), the electric field distribution and SF6 gas decomposition mechanism were explained for the partial discharge energy effect on the gas generation. This study helps understand the relationship between the partial discharge energy and the decomposed gas generation ratio with the experimental results and can be used for the diagnosis of PD and maintenance process for the gas-insulated transformers.

Supervised contrastive learning (SCL) has recently emerged as an alternative to conventional machine learning and deep neural networks. In this study, we propose an SCL model with data augmentation techniques using phase-resolved partial discharge (PRPD) in gas-insulated switchgear (GIS). To increase the fault data for training, we employ Gaussian noise adding, Gaussian noise scaling, random cropping, and phase shifting for supervised contrastive loss. The performance of the proposed SCL was verified by four types of faults in the GIS and on-site noise using an on-line ultra-high-frequency (UHF) partial discharge (PD) monitoring system. The experimental results show that the proposed SCL achieves a classification accuracy of 97.28% and outperforms the other algorithms, including support vector machines (SVM), multilayer perceptron (MLP), and convolution neural networks (CNNs) in terms of classification accuracy, by 6.8%, 4.28%, 2.04%, respectively.

The development of industrial technology is based on electronic devices that are sensitive to power quality. Thus, the demand for high-quality and reliable power supplies is increasing. Voltage sag results in severe problems in the manufacturing process of power quality-sensitive industrial loads. When a fault occurs in a multi-ground power distribution system, the magnitudes of the fault current and voltage sag in the faulted and nonfaulted feeders become high. Hence, installing a superconducting fault current limiter (SFCL) is an effective method of compensating for fault current limitation and voltage sag. This study evaluates the effects of improving the magnitude, duration, and frequency of the voltage sag according to the type of SFCL used. First, a fault in the power distribution system is analyzed using PSCAD/EMTDC, a power system simulation software, according to the fault current-limiting element (CLE) and the type of SFCL. Second, the expected voltage sag frequency caused by a feeder fault in the power distribution system is assessed. Finally, the voltage sag improvement effect according to the CLE and the type of SFCL are compared. The trigger-type SFCL with a resistor as a CLE has been evaluated and found to be effective in improving voltage sag.

In recent years, deep learning has been successfully used in order to classify partial discharges (PDs) for assessing the condition of insulation systems in different electrical equipment. However, fault diagnosis using deep learning is still challenging, as it requires a large amount of training data, which is difficult and expensive to obtain in the real world. This paper proposes a novel one-shot learning method for fault diagnosis using a small dataset of phase-resolved PDs (PRPDs) in a gas-insulated switchgear (GIS). The proposed method is based on a Siamese network framework, which employs a distance metric function for predicting sample pairs from the same PRPD class or different PRPD classes. Experimental results over the small PRPD dataset that was obtained from an ultra-high-frequency sensor in the GIS show that the proposed method achieves outstanding performance for PRPD fault diagnosis as compared with the previous methods.